Proceedings / Actes Volume 29, Division 9, 2008 - Index of
Proceedings / Actes Volume 29, Division 9, 2008 - Index of
Proceedings / Actes Volume 29, Division 9, 2008 - Index of
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Annual Conference <strong>of</strong> the Administrative Sciences Association <strong>of</strong> Canada<br />
Congrès annuel de l’Association des sciences administratives du Canada<br />
ASAC <strong>2008</strong><br />
<strong>Proceedings</strong> / <strong>Actes</strong> <strong>Volume</strong> <strong>29</strong>, <strong>Division</strong> 9, <strong>2008</strong><br />
Human Resources/ Ressources Humaines<br />
Edited by/ Édités par<br />
Zhenzhong Ma, Ph.D.<br />
Odette School <strong>of</strong> Business<br />
University <strong>of</strong> Windsor<br />
Selected Papers / Textes choisis<br />
Halifax, Nova Scotia May 24-27, <strong>2008</strong> / 24 au 27 mai, <strong>2008</strong>
ACKNOWLEDGEMENTS<br />
I would like to thank every author who<br />
submitted papers and symposium proposals to<br />
the HR <strong>Division</strong>.<br />
I would also like to thank the reviewers for their<br />
great work. Their efforts were remarkable and<br />
their valuable comments will no doubt help the<br />
authors in their research.<br />
Thanks to Francine Schlosser, <strong>Division</strong> Chair,<br />
and Sean Lyons, <strong>Division</strong> Program Chair, for<br />
their suggestions and guidance during the<br />
preparations for ASAC <strong>2008</strong>.<br />
Finally, I would like to thank all the colleagues<br />
who volunteered to serve as Discussant and<br />
Chair during the various conference sessions.<br />
Zhenzhong Ma<br />
Odette School <strong>of</strong> Business<br />
University <strong>of</strong> Windsor<br />
REMERCIEMENTS<br />
Je tiens à remercier tous les auteurs qui ont<br />
présenté des documents et des propositions au<br />
colloque de la <strong>Division</strong> des ressources humaines.<br />
Je tiens également à remercier les<br />
commentateurs pour leur excellent travail. Leurs<br />
efforts ont été remarquables et de leurs<br />
précieuses observations vont sans doute aider les<br />
auteurs dans leur recherche.<br />
Grâce à Francine Schlosser, président de la<br />
division, et Sean Lyons, <strong>Division</strong> Programme<br />
président, pour leurs suggestions et conseils au<br />
cours de la préparation de l'ASAC <strong>2008</strong>.<br />
Enfin, je tiens à remercier tous les collègues qui<br />
se sont portés volontaires pour servir en tant que<br />
commentateur et président pendant les<br />
différentes sessions de la conférence.<br />
Zhenzhong Ma<br />
Odette School <strong>of</strong> Business<br />
University <strong>of</strong> Windsor
REVIEWERS / ÉVALUATEURS<br />
Marjorie Armstrong-Stassen University <strong>of</strong> Windsor<br />
Christa L. Austin McMaster University<br />
Nicole Bérubé Concordia University<br />
Akanksha Bedi McMaster University<br />
Travor C. Brown Memorial University<br />
Nita Chhinzer University <strong>of</strong> Guelph<br />
Inta Cinite Carleton University<br />
Julie Cloutier Université du Québec à Montréal<br />
Michel Cossette Université du Québec à Montréal<br />
Rocky Dwyer Department <strong>of</strong> National Defence<br />
Janice Foley University <strong>of</strong> Regina<br />
Ivona Hideg University <strong>of</strong> Toronto<br />
Jack Ito University <strong>of</strong> Regina<br />
Scott Jeffrey University <strong>of</strong> Waterloo<br />
Mark Julien Brock University<br />
Linda Keup Concordia College<br />
Theresa J. Kline University <strong>of</strong> Calgary<br />
Ivy Kyei-Poku University <strong>of</strong> Western Ontario<br />
Richard Lacoursière Université du Québec à Trois-Rivières<br />
Marian Luncasu HEC Montréal<br />
Sean Lyons University <strong>of</strong> Guelph<br />
Pamela Mathews Charles Sturt University<br />
Morris B. Mendelson University <strong>of</strong> New Brunswick<br />
Marc S. Mentzer University <strong>of</strong> Saskatchewan<br />
Denis Morin Université du Québec à Montréal<br />
Peter Mudrack Kansas State University<br />
Floyd Ormsbee Clarkson University<br />
Mark Podolsky McMaster University<br />
Chet Robie Wilfrid Laurier University<br />
Joel Rudin Rowan University<br />
Hermann Schwind Saint Mary’s University<br />
Ge<strong>of</strong>frey Smith University <strong>of</strong> Guelph<br />
Thomas S. Stone Oklahoma State University<br />
Rick Tallman University <strong>of</strong> Northern British Columbia<br />
Andrew Templer University <strong>of</strong> Windsor<br />
Michel Tremblay HEC Montréal<br />
Jean M. Trudel Université de Sherbrooke<br />
Yoshio Yanadori University <strong>of</strong> British Columbia<br />
2
TABLE OF CONTENTS / TABLE DES MATIÈRES<br />
The Relationship <strong>of</strong> Work Status (In)Congruency with Perceived Employer Commitment<br />
and Commitment <strong>of</strong> Community Health Nurses ………………………………………….... 6<br />
Seung Hwan (Mark) Lee (University <strong>of</strong> Western Ontario)<br />
Marjorie Armstrong-Stassen (University <strong>of</strong> Windsor)<br />
Sheila J. Cameron (University <strong>of</strong> Windsor)<br />
Do firm compensation strategies matter to all employees? The scope <strong>of</strong> firm compensation<br />
strategies in high technology firms ………………………………………………………….. 25<br />
Yoshio Yanadori (University <strong>of</strong> British Columbia)<br />
Can the diversity literature help business schools address their relevance crisis? …….… 40<br />
Janice Foley (University <strong>of</strong> Regina)<br />
The Influence <strong>of</strong> Training Transfer Climate and Individual Trainee Characteristics on<br />
Customer Orientation ………………………………………………………………………... 54<br />
Christopher Perryer (University <strong>of</strong> West Australia)<br />
Steven McShane (University <strong>of</strong> West Australia)<br />
Disaggregating Unemployment: Job Leavers, Losers and Lay<strong>of</strong>fs …………………….… 74<br />
Nita Chhinzer (University <strong>of</strong> Guelph)<br />
Khaldoun Ababneh (McMaster University)<br />
Do Employees Pr<strong>of</strong>it from Pr<strong>of</strong>it Sharing? A Longitudinal Analysis <strong>of</strong> Canadian<br />
Establishments ………………………………………………………………………………. 91<br />
Richard Long (University <strong>of</strong> Saskatchewan)<br />
Tony Fang (York University)<br />
HONOURABLE MENTION – MENTION HONORABLE<br />
Explaining Chronological and Subjective Age Differences in Job Satisfaction …………. 111<br />
Nicole Berube (Concordia University)<br />
BEST PAPER – MEILLEURE COMMUNICATION<br />
Are Reactions to Feedback the Mediating Link Between Feedback and Performance?... 125<br />
I. M. Jawahar (Illinois State University)<br />
3
ABSTRACTS / RÉSUMÉS<br />
Leadership and Creativity: The Mediating Effects <strong>of</strong> Intrinsic Motivation, Domain-<br />
Relevant Skills, and Creativity-Relevant Processes …………………………………… 150<br />
Mardi Witzel (Wilfrid Laurier University)<br />
Who receives a performance appraisal and does it matter? An empirical investigation <strong>of</strong><br />
the determinants <strong>of</strong> the receipt <strong>of</strong> a performance appraisal and its effect on job satisfaction<br />
…………………………………………………………………………………………….. 150<br />
Sara L. Mann (University <strong>of</strong> Guelph)<br />
Gary Latham (University <strong>of</strong> Toronto)<br />
Strategic Responsiveness within an IMSA Framework- Institutional and Resource Based<br />
Perspectives ………………………………………………………………………………. 151<br />
Saba Sharih (Wilfrid Laurier University)<br />
Simon Taggar (Wilfrid Laurier University)<br />
Recruiters’ perceptions <strong>of</strong> applicants’ volunteer experience: Are all volunteer assignments<br />
equally relevant? ………………………………………………………………………… 151<br />
Christa L. Austin (McMaster University)<br />
Catherine E. Connelly (McMaster University)<br />
Barriers to Transfer: Participant’s Perspective: Preliminary Results ………………. 152<br />
Travor C. Brown (Memorial University)<br />
Martin McCracken (University <strong>of</strong> Ulster)<br />
Sergio De Leon (Memorial University)<br />
Serious Threats to Canadian University Faculty Members –Survey Outcomes ……. 152<br />
Sunny Marche (Dalhousie University)<br />
John F. Duffy (Dalhousie University)<br />
Geneviève Perron (Dalhousie University)<br />
A Statistical Investigation <strong>of</strong> Antecedents and Outcomes <strong>of</strong> Job-Related Stress for Teachers<br />
…………………………………………………………………………………………….. 153<br />
James Doyle (University <strong>of</strong> Carlton)<br />
Alia El Banna (University <strong>of</strong> Carlton)<br />
Linda Duxbury (University <strong>of</strong> Carlton)<br />
Chris Higgins (University <strong>of</strong> Carlton)<br />
Exploring Faking in the Context Of Multi-Stage Selection Systems ………………… 153<br />
Shawn Komar (University <strong>of</strong> Waterloo)<br />
Douglas J. Brown (University <strong>of</strong> Waterloo)<br />
Chet Robie (Wilfrid Laurier University)<br />
4
Political Skill: A Meta–Analysis <strong>of</strong> Its Predictors and Outcomes ……………….…… 154<br />
Akanksha Bedi (McMaster University)<br />
Mark S. Skowronski (McMaster University)<br />
Aaron C. H. Schat (McMaster University)<br />
Exploring the relationship between organizational culture and Quit behaviour: Evidence<br />
from Canadian call centers ……………………………………………………………... 154<br />
Wendy R. Carroll (Saint Mary’s University)<br />
Terry H. Wagar (Saint Mary’s University)<br />
Kent V. Rondeau (University <strong>of</strong> Alberta)<br />
Developments in Canadian Employment Law and the Supreme Court <strong>of</strong> Canada: Evidence<br />
from an Examination <strong>of</strong> Reasonable Notice in the Wrongful Dismissal Common Law<br />
………………………………………………………………………………………….… 155<br />
Jim D. Grant (Nipissing University)<br />
Terry H. Wagar (Saint Mary’s University)<br />
5
ASAC <strong>2008</strong> Seung Hwan (Mark) Lee (Student)<br />
Halifax, Nova Scotia Ivey School <strong>of</strong> Business<br />
University <strong>of</strong> Western Ontario<br />
Marjorie Armstrong-Stassen<br />
Odette School <strong>of</strong> Business<br />
University <strong>of</strong> Windsor<br />
Sheila J. Cameron<br />
Faculty <strong>of</strong> Nursing<br />
University <strong>of</strong> Windsor<br />
THE RELATIONSHIP OF WORK STATUS (IN)CONGRUENCY WITH PERCEIVED EMPLOYER<br />
COMMITMENT AND COMMITMENT OF COMMUNITY HEALTH NURSES<br />
Using a social exchange framework, we examined the relationship <strong>of</strong> work status<br />
(in)congruency with organization-specific and job-specific aspects <strong>of</strong> perceived<br />
employer commitment and with nurses’ commitment to their organization, to<br />
their job, and to their pr<strong>of</strong>ession. The participants were 1,230 female nurses<br />
employed in community health agencies in Canada. Part-time nurses preferring<br />
full-time work perceived greater job insecurity and lack <strong>of</strong> access to development<br />
opportunities, but reported the highest level <strong>of</strong> commitment to their organization<br />
and pr<strong>of</strong>ession. Full-time nurses preferring part-time work perceived less<br />
organizational support, reported lowest level <strong>of</strong> meaning empowerment, affective,<br />
normative, and pr<strong>of</strong>essional commitment, but reported greatest continuance<br />
commitment.<br />
Work status congruency occurs when a person’s current work status (full-time or part-time)<br />
matches their preference for work status (prefers to work full-time or prefers to work part-time). Work<br />
status incongruency occurs when a person’s work status is inconsistent with the person’s preferred work<br />
status, i.e., the person is working full-time but prefers to work part-time or the person is working parttime<br />
but prefers to work full-time. Work status (in)congruency has been shown to be significantly related<br />
to employee attitudes and well-being including job satisfaction (Armstrong-Stassen, Al-Ma’aitah,<br />
Cameron & Horsburg, 1998; Armstrong-Stassen, Horsburgh & Cameron, 1994; Barker, 1993; Burke &<br />
Greenglass, 2000; Holtom, Lee & Tidd, 2002; Keil, Armstrong-Stassen, Cameron & Horsburgh, 2000;<br />
Krauz, Sagie & Biderman, 2000), organizational commitment (Holtom et al., 2002; Krauz et al., 2000),<br />
turnover intent and voluntary turnover (Armstrong-Stassen et al., 1994, 1998; Burke & Greenglass, 2000;<br />
Holtom et al., 2002), in-role and extra-role performance (Holtom et al., 2002), burnout (Armstrong-<br />
Stassen et al., 1994, 1998; Burke & Greenglass, 2000; Krauz et al., 2000), absenteeism (Burke &<br />
Greenglass, 2000), and psychosomatic and physical health symptoms (Burke & Greenglass, 2000). These<br />
studies have generally supported the prediction that employees with an incongruent work status would<br />
exhibit more negative attitudes toward their organization and their job and report poorer well-being than<br />
employees with a congruent work status.<br />
Much <strong>of</strong> the research to date has investigated the relationship <strong>of</strong> work status (in)congruency<br />
across a wide variety <strong>of</strong> employee attitudes within an individual study. The present study differs from<br />
previous studies by focusing specifically on the relationship between work status (in)congruency and<br />
commitment. The purpose <strong>of</strong> our study was to examine the influence <strong>of</strong> work status (in)congruency on<br />
two aspects <strong>of</strong> commitment: perceived commitment from the organization to its employees and<br />
employees’ commitment to their organization, to their job, and to their pr<strong>of</strong>ession. We examined two<br />
sources <strong>of</strong> perceived commitment from the organization to the employee: organizational practices at the<br />
organizational-level (perceived organizational support, perceived fair treatment, perceived job security,<br />
6
and perceived development opportunities) and organizational practices associated with the employee’s<br />
job (job challenge, job autonomy, and empowerment).<br />
Consistent with many <strong>of</strong> the previous studies investigating work status or work status congruency,<br />
our study was also conducted with nurses. However, unlike previous studies, the nurses in our study were<br />
employed in community health agencies and not in hospitals. Nurses are an appropriate population for<br />
the study <strong>of</strong> part-time workers because <strong>of</strong> the large number <strong>of</strong> nurses who work part time (Armstrong-<br />
Stassen et al., 1998). The part-time employment rate <strong>of</strong> registered nurses (RNs) in Canada is<br />
approximately 30% <strong>of</strong> all RNs employed in the healthcare sector (Pyper, 2004). About one in five <strong>of</strong><br />
those nurses working part time are not doing so by choice, i.e., they would prefer to work full time.<br />
The conceptual framework (see Figure 1) employed in this study is based on the social exchange<br />
theory and the norm <strong>of</strong> reciprocity (Blau, 1964; Gouldner, 1960). Employees will be willing to show<br />
greater commitment to their organization only when they perceive their organization has demonstrated its<br />
commitment to them. We propose that various organizational practices, such as allowing employees to<br />
work their preferred work status, will be perceived by employees as evidence <strong>of</strong> organizations<br />
demonstrating commitment. In turn, employees will reciprocate by showing greater commitment to their<br />
organization, job, and pr<strong>of</strong>ession. Conversely, if organizational practices do not address employees’<br />
needs and preferences, employees will perceive that their organization is not committed to them. In this<br />
case, employees will feel less obligated to commit to their organization, job, and pr<strong>of</strong>ession.<br />
FIGURE 1<br />
CONCEPTUAL FRAMEWORK<br />
Work Status (In)Congruency and Perceived Commitment from the Organization<br />
An organization can demonstrate its commitment to its employees in a variety <strong>of</strong> ways. Our<br />
study focused on organizational actions at the organizational-level (caring and support for employees,<br />
treating employees fairly, providing job security, and ensuring opportunities for career development) and<br />
also at the job-level (providing job challenge, autonomy, and sense <strong>of</strong> empowerment). We propose that<br />
nurses whose work status is congruent (working full-time and prefer to work full-time or working parttime<br />
and prefer to work part-time) will perceive that their organization has demonstrated significantly<br />
more commitment to them than those nurses whose work status is incongruent (working full-time but<br />
prefer to work part-time or working part-time but prefer to work full-time).<br />
7
Perceived organizational support underlies the social exchange between employees and their<br />
organization (Eisenberger, Huntington, Hutchison & Sowa, 1986; Gakovic & Tetrick, 2003). Perceived<br />
organizational support is the general belief that the organization cares about its employees, values their<br />
contribution, and is committed to them (Eisenberger et al., 1986). In a recent study investigating the<br />
mismatch in working hours, van Emmerik and Sanders (2005) recommended that future research consider<br />
perceived organizational support and other components <strong>of</strong> organizational commitment. In their<br />
recommendations for future research, Stamper and Van Dyne (2001) noted that perceived organizational<br />
support may be a key link between work status and employee attitudes and behavior. However, we found<br />
only one published study that has examined the relationship between work status congruency and<br />
perceived organizational support. Armstrong-Stassen et al. (1998) found that Canadian nurses working<br />
full-time but preferring to work part-time in four hospitals that had undergone a recent amalgamation<br />
tended to report lower perceived organizational support than the other three groups but the difference was<br />
not significant. All four groups <strong>of</strong> nurses, including those with a congruent work status, reported<br />
relatively low levels <strong>of</strong> perceived organizational support suggesting that any significant differences across<br />
the work status groups may have been overridden by the downsizing <strong>of</strong> the nursing staff that accompanied<br />
the amalgamation.<br />
We suggest nurses will perceive organizations as not caring and not committed to them if they<br />
experience a mismatch in their preferred working hours. We therefore predicted that full-time nurses who<br />
preferred to work full-time and part-time nurses who preferred to work part-time would report<br />
significantly higher levels <strong>of</strong> perceived organizational support than full-time nurses who preferred to<br />
work part-time and part-time nurses who preferred to work full-time.<br />
Hypothesis 1a: Nurses with a congruent (FT/FT, PT/PT) work status will perceive significantly<br />
higher levels <strong>of</strong> organizational support than nurses with an incongruent (FT/PT and PT/FT) work<br />
status.<br />
There has also been little research on how work status congruency influences employees’<br />
perceptions <strong>of</strong> fair treatment. Some researchers (Feldman, Leanna & Turnley, 1997; Tansky, Gallagher &<br />
Wetzel, 1997) speculated part-time workers to perceive that they are treated less fairly than full-time<br />
workers. Feldman and Doerpinghaus (1992) found that nurses working part-time did express some<br />
concerns about being treated less fairly. This may especially be true for part-time nurses who desire fulltime<br />
status. These nurses may perceive that they receive fewer rewards and less recognition than fulltime<br />
nurses. We therefore predicted that part-time nurses who preferred to work full-time would perceive<br />
that they are treated less fairly than full-time nurses who preferred full-time work and full-time nurses<br />
who preferred part-time work.<br />
Hypothesis 1b: Part-time nurses who prefer to work full-time (PT/FT) will perceive they are<br />
treated significantly less fairly than full-time nurses who prefer to work full-time (FT/FT) and<br />
full-time nurses who prefer to work part-time (FT/PT).<br />
Part-time jobs in general are associated with greater job insecurity than full-time jobs<br />
(Schellenberg, 1997). When compared with full-time workers, part-time workers have been shown to be<br />
less satisfied with their job security (Levanoni & Sales, 1990) and at greater risk <strong>of</strong> being laid <strong>of</strong>f (Barker,<br />
1993; McDonald & Wanner, 1992). There is some empirical evidence that part-time nurses who prefer<br />
full-time employment have the greater perceived job insecurity than nurses with a congruent work status<br />
(FT/FT or PT/PT) and full-time nurses who prefer to work part-time (Armstrong-Stassen et al., 1994;<br />
Burke & Greenglass, 2000; Keil et al., 2000). We therefore expected that community health nurses who<br />
were working part-time but preferred to work full-time would perceive significantly less job security than<br />
nurses with a congruent work status and full-time nurses who preferred part-time employment.<br />
8
Hypothesis 1c: Nurses working part-time but preferring full-time work (PT/FT) will perceive<br />
significantly lower job security than full-time nurses preferring full-time work (FT/FT), part-time<br />
nurses preferring part-time work (PT/PT) and full-time nurses preferring part-time work (FT/PT).<br />
Ng, Butts, Vandenberg, De Joy, and Wilson (2006) noted that providing opportunities for<br />
learning, such as acquiring new knowledge and skills that benefits one’s career development,<br />
demonstrates care and support for employees which in turn would result in the reciprocation <strong>of</strong><br />
commitment by the employees to the organization. Part-time workers in general have less access to<br />
development opportunities (Gallie, White, Cheng & Tomlinson, 1998; MacDermid, Dean Lee, Buck &<br />
Williams, 2001; Skinner, 1999). They have been found to be significantly less satisfied than full-time<br />
employees with development opportunities (Bergmann, Grahn & Wyatt, 1986; Comfort, Johnson &<br />
Wallace, 2003; Edwards & Robinson, 2004) and to feel excluded from development opportunities (Barker,<br />
1993; Edwards & Robinson, 2004). We found no published studies that have investigated the effects <strong>of</strong><br />
work status congruency on how employees perceive their organization’s provision <strong>of</strong> development<br />
opportunities. However, we expected that within the part-time group, nurses who preferred to work fulltime<br />
would be less satisfied with their access to development opportunities than those part-time nurses<br />
who preferred part-time employment. Stratton (1996) found that part-time workers who preferred to<br />
work full-time more closely resembled full-time workers than other part-time workers who preferred to<br />
work part-time. This suggests that part-time nurses who want to work full-time may have a strong<br />
interest in pursuing development opportunities and that they will be more sensitive to the lack <strong>of</strong> career<br />
development opportunities than part-time nurses who prefer to work part-time.<br />
Hypothesis 1d: Nurses working part-time but preferring full-time work (PT/FT) will perceive<br />
significantly fewer development opportunities than full-time nurses preferring full-time work<br />
(FT/FT), part-time nurses preferring part-time work (PT/PT) and full-time nurses preferring parttime<br />
work (FT/PT).<br />
An organization can also demonstrate its commitment to its employees by providing them with an<br />
interesting and challenging role, allowing employees to make decisions about their job, and creating a<br />
sense <strong>of</strong> empowerment to make those decisions. Ng et al. (2006) suggested that such variables as<br />
autonomy and psychological empowerment that allow employees to fulfill their desire to have more<br />
control at work may be especially relevant for enhancing employees’ commitment to the organization.<br />
Part-time workers have been found to perceive less control over their work (Ross & Wright, 1998), to<br />
have more routine and less challenging jobs (Ross & Wright, 1998), to have less decision-making<br />
autonomy (Feldman & Doerpinghaus, 1992; Gallie et al., 1998; Ross & Wright, 1998), and to manifest a<br />
lower sense <strong>of</strong> empowerment (Markey, Hodgkinson & Kowalczyk, 2002). We expected that this may<br />
especially be the case for those nurses who are working part-time but prefer full-time work. Part-time<br />
nurses who prefer to work part-time may have resigned themselves to performing jobs that are routine,<br />
that <strong>of</strong>fer little autonomy, and that provide few opportunities to exercise control. On the other hand, parttime<br />
nurses who prefer to work full-time are likely to view full-time jobs as an opportunity to engage in<br />
more challenging and meaningful work, to have more control over their job, and to have greater<br />
empowerment. This may result in these nurses having a negative perception <strong>of</strong> their current part-time job.<br />
Hypothesis 1e: Part-time nurses who prefer to work full-time (PT/FT) will report significantly<br />
lower levels <strong>of</strong> job challenge, job autonomy, and empowerment than full-time nurses who prefer<br />
full-time work (FT/FT), full-time nurses who prefer part-time work (FT/PT), and part-time nurses<br />
who prefer part-time work (PT/PT).<br />
Perceived Work Status (In)Congruency and Nurses’ Commitment<br />
9
Meyer and Allen (1997) view organizational commitment as a multidimensional construct which<br />
includes affective commitment (emotional attachment), continuance commitment (attachment due to<br />
perceived loss associated with leaving), and normative commitment (attachment through moral<br />
obligation). Most <strong>of</strong> the research comparing work status and work status congruency and organizational<br />
commitment has focused on affective commitment and neglected possible differences for other types <strong>of</strong><br />
commitment (Thorsteinson, 2003). Holtom et al. (2002) found that work status congruency was<br />
significantly positively related to affective commitment suggesting that full-time and part-time workers<br />
with a congruent work status will exhibit greater commitment to their organization than full-time and<br />
part-time workers with an incongruent work status. Krauz, Sagie and Bidermann (2000) found that<br />
organizational commitment was higher for nurses who preferred to work more hours (PT/FT) and lower<br />
for nurses who preferred to work fewer hours (FT/PT). Similar findings were reported for Dutch<br />
Ministry employees by van Emmerick and Sanders (2005). These researchers concluded that<br />
mismatches <strong>of</strong> preferring to work fewer hours are more detrimental to affective commitment than<br />
mismatches <strong>of</strong> preferring to work more hours. We therefore predicted that nurses who were working fulltime<br />
but who preferred to work part-time would express significantly less affective commitment toward<br />
their organization than nurses in the other three groups.<br />
Hypothesis 2a: Nurses working full-time but preferring to work part-time (FT/PT) will report<br />
significantly less affective commitment to their organization than full-time nurses preferring fulltime<br />
work (FT/FT), part-time nurses preferring part-time work (PT/PT), and part-time nurses<br />
preferring full-time work (PT/FT).<br />
The relationship <strong>of</strong> work status with continuance and normative commitment has received sparse<br />
attention and we found no published studies that have examined the relationship <strong>of</strong> work status<br />
congruency with these two dimensions <strong>of</strong> commitment. Conway and Briner (2002) and Hackett and<br />
Bycio (1993) found that full-time nurses expressed significantly higher levels <strong>of</strong> continuance commitment<br />
than part-time nurses. However, it can be argued that nurses with an incongruent work status may exhibit<br />
higher levels <strong>of</strong> continuance commitment than nurses with a congruent work status. It may be that nurses<br />
with an incongruent work status continue to stay with their organization either because <strong>of</strong> the sacrifices<br />
involved in leaving or because there are no other available alternatives. We therefore predicted that<br />
nurses with an incongruent work status would report significantly higher levels <strong>of</strong> continuance<br />
commitment than nurses with a congruent work status.<br />
Hypothesis 2b: Nurses with an incongruent (FT/PT, PT/FT) work status will report significantly<br />
greater continuance commitment than nurses with a congruent (FT/FT, PT/PT) work status.<br />
We found only one study that has examined the effect <strong>of</strong> work status on normative commitment.<br />
Hackett and Bycio (1993) found no significant differences between full-time and part-time nurses for<br />
normative commitment. Affective commitment and normative commitment are highly similar dimensions<br />
<strong>of</strong> commitment (Cooper-Hakim & Viswesvaran, 2005) and should exhibit similar relationships with other<br />
variables. Given that full-time nurses who prefer to work part-time have been shown to express less<br />
affective commitment toward their organization than the other three groups, we extrapolated from this to<br />
propose that nurses working full-time but preferring part-time work would express significantly lower<br />
levels <strong>of</strong> normative commitment than nurses in the other three groups.<br />
Hypothesis 2c: Nurses working full-time but preferring to work part-time (FT/PT) will report<br />
significantly less normative commitment than full-time nurses preferring full-time work (FT/FT),<br />
part-time nurses preferring part-time work (PT/PT), and part-time nurses preferring full-time<br />
work (PT/FT).<br />
10
Another form <strong>of</strong> commitment, job involvement, reflects the extent to which individuals identify<br />
psychologically with their job (Blau & Boal, 1987). There is empirical research comparing the job<br />
involvement <strong>of</strong> full-time and part-time workers but we could find no published studies that have<br />
investigated the relationship between work status congruency and job involvement. Part-time workers in<br />
general report significantly less job involvement than full-time workers (Bergmann, Grahn & Wyatt, 1986;<br />
Hackett & Bycio, 1993; Morrow, McElroy & Elliot, 1994; Sverke, Gallagher & Hellgren, 2000; Wetzel et<br />
al., 1990). Part-time nurses are more likely to feel they are underemployed than full-time nurses.<br />
However, part-time nurses who prefer to work full-time are more likely to perceive that they are<br />
underemployed than part-time nurses who prefer to work part-time. Feldman et al. (1997) noted that<br />
underemployment has been consistently linked with poorer job involvement. We therefore expected that<br />
nurses working part-time who preferred to work full-time would report significantly lower levels <strong>of</strong> job<br />
involvement than nurses in the other three groups.<br />
Hypothesis 2d: Nurses working part-time but preferring full-time work (PT/FT) will report<br />
significantly lower job involvement than full-time nurses preferring full-time work (FT/FT), parttime<br />
nurses preferring part-time work (PT/PT), and full-time nurses preferring part-time work<br />
(FT/PT).<br />
Pr<strong>of</strong>essional commitment is another form <strong>of</strong> commitment that has received little attention in the<br />
work status literature. We found only three published studies that have examined the relationship<br />
between work status and pr<strong>of</strong>essional commitment. Two <strong>of</strong> these studies (Hackett & Bycio, 1993; Wetzel<br />
et al., 1990) found no significant differences between full-time and part-time nurses for pr<strong>of</strong>essional<br />
commitment. Morrow et al. (1994) found that full-time nurses reported higher pr<strong>of</strong>essional commitment<br />
than part-time nurses but there were no significant differences for preferred work status. Unfortunately,<br />
Morrow et al. collapsed across full-time and part-time and this may have masked significant differences<br />
within the full-time and part-time congruent and incongruent groups. We propose that full-time nurses<br />
who prefer to work part-time and part-time nurses who prefer to work full-time will be less committed to<br />
their pr<strong>of</strong>ession because their needs or preferences are not being fulfilled. We therefore expected nurses<br />
with incongruent work status to exhibit lower levels <strong>of</strong> commitment to the nursing pr<strong>of</strong>ession than nurses<br />
with congruent work status.<br />
Hypothesis 2e: Nurses with an incongruent (FT/PT, PT/FT) work status will report significantly<br />
less pr<strong>of</strong>essional commitment than nurses with a congruent (FT/FT, PT/PT) work status.<br />
A second objective <strong>of</strong> our study was to identify the relevant predictors <strong>of</strong> each <strong>of</strong> the forms <strong>of</strong><br />
nurses’ commitment. Based on previous research, we expected the perceived employer commitment<br />
variables to be differentially associated with the five commitment variables. In particular, we expected<br />
the organizational practices that were organization specific (perceived organizational support, fair<br />
treatment, job security and development opportunities) to be important predictors <strong>of</strong> organizational<br />
commitment and the organizational practices that were job-specific (job challenge, job autonomy,<br />
empowerment) to be important predictors <strong>of</strong> job involvement. We anticipated that both types <strong>of</strong><br />
organizational practices would be associated with pr<strong>of</strong>essional commitment.<br />
Participants and Procedure<br />
Method<br />
In 2005, we mailed questionnaire packets to 2,576 registered nurses employed in community<br />
health agencies. The nurses were randomly selected from the College <strong>of</strong> Nurses <strong>of</strong> Ontario registry list.<br />
Seventeen questionnaire packets were returned because the person had moved, 61 nurses indicated they<br />
11
were not eligible to participate for various reasons (e.g., retired, on disability, left the pr<strong>of</strong>ession). A total<br />
<strong>of</strong> 1275 questionnaires was received with two <strong>of</strong> these discarded because they were incomplete giving a<br />
response rate <strong>of</strong> 51%. Ninety-eight percent <strong>of</strong> the respondents were women. The current paper only<br />
deals with the female nurses.<br />
The 1,230 female participants had worked for their respective community health agency an<br />
average <strong>of</strong> 9.91 years (SD = 7.52) and in their current job an average <strong>of</strong> 8.55 years (SD = 6.75). Their<br />
average age was 46.77 years (SD = 9.36) ranging from 26 to 73 years. About 34% worked in public<br />
health, 28% worked in home care, 24% worked in community care access centres, and the remaining 14%<br />
worked in a variety <strong>of</strong> community health settings including clinics, doctors’ <strong>of</strong>fices, and community<br />
mental health. The majority (85%) <strong>of</strong> the respondents was married.<br />
Measures<br />
Work status congruency. Respondents were asked to indicate their current work status (full-time<br />
or part-time) and then to indicate their preferred work status (full-time or part-time). We created a work<br />
status variable <strong>of</strong> four categories: full-time nurses preferring full-time work (FT/FT), full-time nurses<br />
preferring part-time work (FT/PT), part-time nurses preferring full-time work (PT/FT), and part-time<br />
nurses preferring part-time work (PT/PT). Of these four groups, two have a congruent work status<br />
(FT/FT, n = 456 and PT/PT, n = 478) and two have an incongruent work status (FT/PT, n = 212 and<br />
PT/FT, n = 49). There were 35 missing values.<br />
Perceived employer commitment. We assessed perceived organizational support with nine items<br />
from the Survey <strong>of</strong> Perceived Organizational Support scale developed by Eisenberger et al. (1986). We<br />
modified the items to read “my agency” instead <strong>of</strong> “my organization.” Sample items are “My agency<br />
values my contribution to its well-being” and “My agency really cares about my well-being.” The<br />
response categories ranged from 1 (strongly disagree) to 5 (strongly agree). The reliability coefficient<br />
(Cronbach alpha) was .94. Perceptions <strong>of</strong> being treated fairly were adapted from Price and Mueller’s<br />
(1986) 6-item perceived distributive justice scale. Sample items are “To what extent do you feel you are<br />
fairly rewarded for the amount <strong>of</strong> effort that you put forth?” and “To what extent do you feel you are<br />
fairly rewarded considering the responsibilities that you have?” The response categories ranged from 1<br />
(to no extent) to 5 (very great extent). The coefficient alpha was .95. We measured perceived job<br />
security with three items from Hellgren and Sverke (2003). A sample item is “I feel uneasy about losing<br />
my job in the near future” (reverse scored). The response categories ranged from 1 (strongly disagree) to<br />
5 (strongly agree). The coefficient alpha was .83. The10-item measure <strong>of</strong> pr<strong>of</strong>essional development<br />
opportunities was developed for this study. Respondents were asked to rate how well their agency was<br />
doing in providing them with such opportunities as training to update their current job skills or to learn<br />
new skills and expertise, release time with pay for attending pr<strong>of</strong>essional development activities,<br />
opportunities for promotion or advancement, and opportunities to utilize their knowledge and skills in<br />
their current position. The response categories ranged from 1 (very poorly) to 5 (very well). The<br />
coefficient alpha was .92.<br />
We assessed job challenge with five items adapted from Zeitz, Johannesson and Ritchie (1997).<br />
A sample item is “My job requires me to do many different things at work, using a variety <strong>of</strong> skills.” The<br />
coefficient alpha was .83. Job autonomy was measured with four items adapted from Price’s (1997)<br />
measure <strong>of</strong> work autonomy. A sample item is “I have a great deal <strong>of</strong> freedom as to how to do my job.”<br />
The coefficient alpha was .83. We assessed two dimensions <strong>of</strong> empowerment—meaning and outcome<br />
empowerment. The meaning dimension <strong>of</strong> empowerment was assessed using the 3-item Meaning<br />
subscale developed by Spreitzer (1995). A sample item is “My job activities are personally meaningful to<br />
me.” The coefficient alpha was .79. The 3-item measure <strong>of</strong> the outcome dimension <strong>of</strong> empowerment was<br />
adapted from the Outcome Empowerment scale developed by Irvine, Leatt, Evans and Baker (1999).<br />
12
This is similar to Spreitzer’s impact subscale but was specifically developed for nurses. A sample item is<br />
“I feel that I am able to bring about improvements in the way work is done in community health nursing.”<br />
The coefficient alpha was .79. For the four job features measures, the response categories ranged from 1<br />
(strongly disagree) to 5 (strongly agree).<br />
Nurses’ commitment. Organizational commitment consisted <strong>of</strong> affective, continuance and<br />
normative commitment. Affective commitment was assessed with three items from the Affective<br />
Commitment scale developed by Meyer, Allen and Smith (1993). A sample item is “This agency has a<br />
great deal <strong>of</strong> personal meaning for me.” The coefficient alpha was .86. Continuance commitment was<br />
assessed with three items from the Continuance Commitment scale developed by Meyer et al. (1993). A<br />
sample item is “It would be hard for me to leave my agency right now, even if I wanted to.” The<br />
coefficient alpha was .61. Normative commitment was measured with three items from the Normative<br />
Commitment scale developed by Meyer et al. (1993). A sample item is “This agency deserves my<br />
loyalty.” The coefficient alpha was .75. For each <strong>of</strong> the organizational commitment scales, the word<br />
“organization” was replaced with “agency. The response categories for the three organizational<br />
commitment measures ranged from 1 (strongly disagree) to 5 (strongly agree). We measured job<br />
involvement with six items from the Job Involvement-Role (JIR) scale developed by Paullay, Alliger and<br />
Stone-Romero (1994). The JIR items reflect the degree to which people are engaged in the specific tasks<br />
that make up their job. Sample items are “I’ll stay overtime to finish something that I’m working on” and<br />
“I am very much involved personally in the type <strong>of</strong> work that I do in my present job.” The response<br />
categories ranged from 1 (strongly disagree) to 5 (strongly agree). The coefficient alpha was .70. The 5item<br />
measure <strong>of</strong> pr<strong>of</strong>essional commitment was adapted from Reilly and Orsak (1991). Sample items are<br />
“If I could go into a different pr<strong>of</strong>ession other than nursing which paid the same, I would probably do it”<br />
(reverse scored) and “I definitely want a career for myself in the nursing pr<strong>of</strong>ession.” The response<br />
categories ranged from 1 (strongly disagree) to 5 (strongly agree). The coefficient alpha was .87.<br />
Demographic variables. The demographic variables included the workplace setting (public<br />
health, community care access centre, home care, other), length <strong>of</strong> time employed in current workplace,<br />
length <strong>of</strong> time employed in current position, age, gender, and marital status.<br />
Data Analysis<br />
To test our hypotheses, we conducted multivariate analysis <strong>of</strong> covariance (MANCOVA) with<br />
work status (in)congruency as the independent variable and organizational tenure, job tenure and age as<br />
the covariates. Thorsteinson (2003) noted that demographic differences such as age and tenure may<br />
obscure potential differences between full-time and part-time workers’ attitudes. We used Tukey post<br />
hoc comparisons to identify significant differences among the four work status groups. To identify the<br />
relative contribution <strong>of</strong> work status (in)congruency and the perceived employer commitment variables on<br />
the five forms <strong>of</strong> nurses’ commitment, we conducted a series <strong>of</strong> hierarchical regressions. We effects<br />
coded the work status (in)congruency variable and used FT/FT as the comparison group. The control<br />
variables (organizational tenure, job tenure, and age) were entered in the first step, in step 2 we entered<br />
the three dummy coded work status (in)congruency groups (FT/PT, PT/FT and PT/PT), and in step 3 we<br />
entered the perceived employer commitment variables.<br />
Results<br />
There was a significant overall work status (in)congruency effect for both the organizationspecific<br />
aspects (F(12,3030) = 4.02, p < .001) and the job-specific aspects (F(12,3330) = 3.99, p < .001) <strong>of</strong><br />
perceived employer commitment. The means, standard deviations, and ANCOVA F-values are presented<br />
in Table 1. Hypothesis 1a predicted that nurses with a congruent work status would perceive significantly<br />
13
higher levels <strong>of</strong> perceived organizational support than nurses with an incongruent work status. This<br />
hypothesis was supported for nurses working full-time but not for nurses working part-time. Full-time<br />
nurses who preferred to work full-time perceived significantly higher levels <strong>of</strong> organizational support<br />
than did full-time nurses who preferred to work part-time.<br />
Hypothesis 1b predicted that part-time nurses who preferred to work full-time would perceive<br />
being treated significantly less fairly than both groups <strong>of</strong> full-time nurses. This hypothesis was partially<br />
supported. Part-time nurses who preferred full-time work perceived that they were treated less fairly than<br />
full-time nurses who preferred full-time work but there was no significant difference between part-time<br />
nurses who preferred full-time work and full-time nurses who preferred part-time work.<br />
Table 1<br />
Means, SD, and ANCOVA F-values for Perceived Commitment from the Organization Variables<br />
FT/FT FT/PT PT/FT PT/PT F-value<br />
Mean<br />
(SD)<br />
Organizational Features:<br />
Organizational Support 3.<strong>29</strong> a<br />
(.84)<br />
Fair Treatment 2.77 a<br />
(.87)<br />
Job security 3.69 a<br />
(1.02)<br />
Development Opportunities 3.05 a<br />
(.85)<br />
Mean<br />
(SD)<br />
3.06 a<br />
(.79)<br />
2.54<br />
(.81)<br />
3.70 b<br />
(.90)<br />
2.91 b<br />
(.78)<br />
Mean<br />
(SD)<br />
3.21<br />
(.90)<br />
2.47 a<br />
(.86)<br />
2.96 abc<br />
(1.16)<br />
2.57 abc<br />
(.98)<br />
Mean<br />
(SD)<br />
3.17<br />
(.77)<br />
2.69<br />
(.76)<br />
3.60 c<br />
(.94)<br />
2.92 c<br />
(.72)<br />
Job Features:<br />
Job Challenge 4.21 a<br />
4.13 4.12<br />
(.59) (.63) (.53) (.68)<br />
Job Autonomy 3.60 a<br />
3.45 3.62<br />
(.78) (.79) (.76) (.82)<br />
Meaning Empowerment 4.25 4.09<br />
(.55)<br />
a<br />
4.30<br />
(.55)<br />
a<br />
4.14<br />
(.47) (.54)<br />
Outcome Empowerment 3.79 a<br />
3.64 3.65<br />
(.77) (.83) (.87) (.73)<br />
Note. Superscripts denote means that differ significantly across the work status groups.<br />
* p < .05 ** p < .01 *** p < .001<br />
4.02 a<br />
3.45 a<br />
3.58 a<br />
3.73*<br />
4.21*<br />
6.83***<br />
4.81**<br />
8.30***<br />
3.17*<br />
6.24***<br />
5.99***<br />
There was support for hypotheses 1c and 1d. Part-time nurses who preferred full-time work<br />
perceived significantly lower job security and fewer development opportunities than did nurses in the<br />
other three groups.<br />
Hypothesis 1e, which predicted that part-time nurses who preferred full-time work would report<br />
significantly lower levels <strong>of</strong> job challenge, job autonomy, and empowerment received no support. The<br />
significant differences were primarily between the two work status congruent groups (FT/FT and PT/PT).<br />
Full-time nurses who preferred full-time work reported significantly higher levels <strong>of</strong> job challenge, job<br />
autonomy, and outcome empowerment than did part-time nurses who preferred part-time work. Contrary<br />
to our prediction, part-time nurses who preferred full-time work reported a significantly higher sense <strong>of</strong><br />
meaning empowerment than did full-time nurses who preferred part-time work.<br />
14
There was an overall significant work status (in)congruency effect for the nurses’ commitment<br />
variables (F(15,3259) = 9.48, p < .001). The means, standard deviations and ANCOVA F-values are in<br />
Table 2. Hypothesis 2a predicted that full-time nurses who preferred part-time work would report<br />
significantly less affective commitment to their agency than nurses in the other three groups. This<br />
hypothesis was partially supported. Full-time nurses who preferred part-time work reported significantly<br />
less affective commitment to their agency than did part-time nurses who preferred full-time work. Parttime<br />
nurses who preferred part-time work also reported significantly less affective commitment to their<br />
agency than part-time nurses who preferred full-time work.<br />
Hypothesis 2b predicted that nurses with an incongruent work status would report higher<br />
continuance commitment than nurses with a congruent work status. Part-time nurses who preferred fulltime<br />
work reported significantly higher continuance commitment than full-time nurses who preferred fulltime<br />
work and part-time nurses who preferred part-time work. Full-time nurses who preferred part-time<br />
work reported significantly higher levels <strong>of</strong> continuance commitment than part-time nurses who preferred<br />
part-time work.<br />
Table 2<br />
Means, SD, and ANCOVA F-values for Nurses’ Commitment<br />
FT/FT FT/PT PT/FT PT/PT F-value<br />
Mean<br />
(SD)<br />
Mean<br />
(SD)<br />
Mean<br />
(SD)<br />
Mean<br />
(SD)<br />
Affective Commitment 3.41<br />
(.96) (.92) (.99) (.92)<br />
Continuance Commitment 3.11 a<br />
3.32<br />
(.83)<br />
c<br />
3.55<br />
(.79)<br />
ab<br />
(.81) (.78)<br />
Normative Commitment 3.07 2.82<br />
(.88)<br />
a<br />
3.11<br />
(.83)<br />
a<br />
2.94<br />
(.96) (.77)<br />
Job Involvement<br />
a b<br />
4.01 3.84<br />
(.60)<br />
a<br />
4.00<br />
(.66) (.56) (.58)<br />
Pr<strong>of</strong>essional Commitment 3.51 a<br />
a b c<br />
3.07 3.58<br />
(.92) (.91)<br />
b<br />
(.83) (.87)<br />
Note. Superscripts denote means that differ significantly across the work status groups.<br />
* p < .05 ** p < .01 *** p < .001<br />
3.17 a<br />
3.51 ab<br />
3.17 b<br />
2.88 bc<br />
3.83 b<br />
3.45 c<br />
7.06***<br />
21.62***<br />
5.19**<br />
9.68***<br />
11.46***<br />
Hypothesis 2c was only partially supported. Full-time nurses who preferred part-time work<br />
reported significantly lower normative commitment than part-time nurses who preferred part-time work.<br />
There were no significant differences between the FT/PT nurses and nurses with a congruent work status<br />
(FT/FT, PT/PT).<br />
Hypothesis 2d predicted that part-time nurses who preferred full-time work would report<br />
significantly lower job involvement than nurses in the other three groups. This hypothesis received no<br />
support. Full-time nurses who preferred full-time work reported significantly higher job involvement<br />
than full-time nurses who preferred part-time work and part-time nurses who preferred part-time work.<br />
There were no significant differences found for part-time nurses who preferred full-time work.<br />
Hypothesis 2e predicted that nurses with an incongruent work status would report significantly<br />
lower commitment to the nursing pr<strong>of</strong>ession than nurses with a congruent work status. This was<br />
supported for the FT/PT group but not for the PT/FT group. Full-time nurses who preferred part-time<br />
15
work reported significantly lower levels <strong>of</strong> pr<strong>of</strong>essional commitment than full-time nurses who preferred<br />
full-time work, part-time nurses who preferred full-time work and part-time nurses who preferred parttime<br />
work.<br />
The hierarchical regression results are presented in Table 3 (affective and continuance<br />
commitment), Table 4 (normative commitment and job involvement), and Table 5 (pr<strong>of</strong>essional<br />
commitment). The predictor variables accounted for 52% <strong>of</strong> the variance in affective commitment, 15%<br />
<strong>of</strong> the variance in continuance commitment, 48% <strong>of</strong> the variance in normative commitment, 43% <strong>of</strong> the<br />
variance in job involvement, and 30% <strong>of</strong> the variance in pr<strong>of</strong>essional commitment.<br />
The significant predictors <strong>of</strong> affective commitment were perceived organizational support,<br />
meaning empowerment, outcome empowerment, and development opportunities. The significant<br />
predictors <strong>of</strong> continuance commitment were the three work status groups, perceived job security, and<br />
meaning empowerment. The significant predictors <strong>of</strong> normative commitment were perceived<br />
organizational support, outcome empowerment, perceived fair treatment, development opportunities,<br />
meaning empowerment, and FT/PT work status. The significant predictors <strong>of</strong> job involvement were<br />
meaning empowerment, job challenge, outcome empowerment, job autonomy, and PT/PT work status.<br />
The significant predictors <strong>of</strong> pr<strong>of</strong>essional commitment were meaning empowerment, outcome<br />
empowerment, FT/PT work status, and perceived organizational support.<br />
Table 3<br />
Hierarchical Regression Results <strong>of</strong> Affective and Continuance Commitment<br />
Step 1<br />
β<br />
Affective Commitment Continuance Commitment<br />
Step 2<br />
β<br />
Step 3<br />
β<br />
Step 1<br />
β<br />
Step 2<br />
β<br />
Step 3<br />
β<br />
Organizational tenure .23*** .22*** .23*** .00 -.01 .02<br />
Job tenure -.12** -.11** -.06 .12** .15*** .14***<br />
Age .08* .08* -.03 -.03 -.01 .01<br />
R 2 = .05<br />
F = 15.37***<br />
R 2 = .01<br />
F = 4.62**<br />
FT/PT -.10** -.02 .12** .09**<br />
PT/FT .01 .02 .09** .07*<br />
PT/PT -.12** -.05 -.15*** -.19***<br />
R 2 = .06<br />
∆ R 2 = .01<br />
F = 5.15**<br />
R 2 = .07<br />
∆ R 2 = .06<br />
F = 21.32***<br />
Organizational support .50*** -.07<br />
Fair treatment .04 .00<br />
Job security -.01 -.17***<br />
Develop opportunities .08** -.02<br />
Job challenge -.03 -.05<br />
Job autonomy .01 -.04<br />
Meaning empowerment .15*** -.08*<br />
Outcome empowerment .12*** -.02<br />
R 2 = .52<br />
∆ R 2 = .46<br />
F = 116.18***<br />
* p < .05 ** p < .01 *** p < .001<br />
R 2 = .15<br />
∆ R 2 = .08<br />
F = 10.74***<br />
16
Table 4<br />
Hierarchical Regression Results <strong>of</strong> Normative Commitment and Job Involvement<br />
Step 1<br />
β<br />
Normative Commitment Job Involvement<br />
Step 2<br />
β<br />
Step 3<br />
β<br />
Step 1<br />
β<br />
Step 2<br />
β<br />
Organizational tenure .16*** .16*** .17*** .05 .05 .03<br />
Job tenure -.08 -.08 -.02 -.07 -.06 -.04<br />
Step 3<br />
β<br />
Age .11** .11** .00 .25*** .25*** .16***<br />
R 2 = .04<br />
F = 12.05***<br />
R 2 = .06<br />
F = 22.24***<br />
FT/PT -.14*** -.06* -.07* -.03<br />
PT/FT -.03 -.01 .02 .00<br />
PT/PT -.10** -.04 -.16*** -.07*<br />
R 2 = .05<br />
∆ R 2 = .01<br />
F = 6.<strong>29</strong>***<br />
R 2 = .09<br />
∆ R 2 = .03<br />
F = 8.30***<br />
Organizational support .45*** .03<br />
Fair treatment .11*** .01<br />
Job security -.04 -.05<br />
Develop opportunities .08* -.05<br />
Job challenge -.03 .26***<br />
Job autonomy .01 .09**<br />
Meaning empowerment .08** .31***<br />
Outcome empowerment .13*** .10**<br />
R 2 = .48<br />
∆ R 2 = .43<br />
F = 101.24***<br />
* p < .05 ** p < .01 *** p < .001<br />
R 2 = .43<br />
∆ R 2 = .34<br />
F = 72.01***<br />
Table 5<br />
Hierarchical Regression Results <strong>of</strong> Pr<strong>of</strong>essional Commitment<br />
Pr<strong>of</strong>essional Commitment<br />
Step 1<br />
β<br />
Step 2<br />
β<br />
Step 3<br />
β<br />
Organizational tenure .01 .00 .00<br />
Job tenure -.04 -.05 -.02<br />
Age .14*** .14*** .07*<br />
R 2 = .02<br />
F = 6.07***<br />
FT/PT -.18*** -.13***<br />
PT/FT .01 .01<br />
PT/PT -.03 .03<br />
R 2 = .05<br />
∆ R 2 = .03<br />
17
F = 9.99***<br />
Organizational support .10*<br />
Fair treatment .04<br />
Job security .03<br />
Develop opportunities .04<br />
Job challenge -.07<br />
Job autonomy -.02<br />
Meaning empowerment .30***<br />
Outcome empowerment .23***<br />
R 2 = .30<br />
∆ R 2 = .25<br />
F = 43.44***<br />
* p < .05 ** p < .01 *** p < .001<br />
Discussion<br />
Work status research has been criticized in the past for being largely atheoretical (Barling &<br />
Gallagher, 1996). In the present study, we applied social exchange theory to investigate the relationships<br />
between work status (in)congruency with perceived employer commitment and with nurses’ commitment.<br />
For the organization-specific aspects <strong>of</strong> perceived employer commitment, the findings partially supported<br />
our prediction that nurses with an incongruent work status would perceive significantly less<br />
organizational support and feel they are treated less fairly than nurses with a congruent work status. The<br />
findings fully supported our prediction that part-time nurses who preferred full-time employment would<br />
perceive significantly less job security and fewer development opportunities than nurses in the other three<br />
groups. However, there was no support for the predictions regarding work status (in)congruency and the<br />
job-specific aspects <strong>of</strong> perceived employer commitment. In this case, the significant differences were<br />
primarily between the two groups <strong>of</strong> nurses with a congruent work status (FT/FT, PT/PT). Full-time<br />
nurses who preferred to work full-time reported significantly more job challenge, greater job autonomy,<br />
and a greater sense <strong>of</strong> outcome empowerment than part-time nurses who preferred to work part-time.<br />
These findings indicate that work status incongruency (FT/PT, PT/FT) has a greater influence on the<br />
organization-specific aspects <strong>of</strong> perceived employer commitment (perceived organizational support, fair<br />
treatment, job security, and development opportunities) whereas work status congruency (FT/FT, PT/PT)<br />
has a greater influence on the job-specific aspects <strong>of</strong> perceived employer commitment (job challenge, job<br />
autonomy, and outcome empowerment).<br />
Reynolds (2003) noted that researchers tend to focus on employees who are part-time but prefer<br />
full-time work and give little attention to employees who are full-time but prefer part-time employment.<br />
The findings <strong>of</strong> our study show that full-time nurses who preferred to work part-time perceived<br />
significantly less organizational support than the other three groups, reported the lowest sense <strong>of</strong> meaning<br />
empowerment, and expressed the lowest levels <strong>of</strong> affective commitment (along with the PT/PT group),<br />
normative commitment, and pr<strong>of</strong>essional commitment. Compared with the FT/PT group, part-time<br />
nurses who preferred to work full-time (PT/FT) reported less job security and less access to development<br />
opportunities. At the same time, they expressed a greater sense <strong>of</strong> meaning empowerment and higher<br />
levels <strong>of</strong> affective, normative, and pr<strong>of</strong>essional commitment. These findings suggest that employers need<br />
to provide part-time nurses who prefer to work full-time with full-time positions if at all possible.<br />
Otherwise, they may jeopardize the commitment <strong>of</strong> these nurses to both the organization and the nursing<br />
pr<strong>of</strong>ession. But what may be even more critical is to provide full-time nurses who want part-time work<br />
with part-time employment. By not doing so, these nurses may already have decided that their<br />
18
organization has failed to fulfill its obligations to them and, in turn, may reduce their commitment toward<br />
their organization and pr<strong>of</strong>ession.<br />
What is noteworthy about the regression findings is the association <strong>of</strong> the meaning and outcome<br />
dimensions <strong>of</strong> empowerment with nurses’ commitment. The meaning dimension <strong>of</strong> empowerment was<br />
the only perceived employer commitment variable that was a significant predictor <strong>of</strong> all five forms <strong>of</strong><br />
nurses’ commitment and the outcome dimension <strong>of</strong> empowerment significantly predicted four <strong>of</strong> the<br />
nurses’ commitment variables. Nurses who felt their work was personally meaningful and important to<br />
them and who felt that they were instrumental in making improvements to community health nursing<br />
reported significantly greater affective, normative, job, and pr<strong>of</strong>essional commitment than nurses who<br />
lacked a sense <strong>of</strong> empowerment. Nurses who felt their work was personally meaningful and important to<br />
them also were significantly less likely to be staying with their agency because they felt they had to, i.e.,<br />
they expressed lower levels <strong>of</strong> continuance commitment. Consistent with social exchange theory, Liden,<br />
Wayne & Sparrowe (2000) proposed that empowerment may be linked with employee’s sense <strong>of</strong><br />
commitment (affective) to the organization through the process <strong>of</strong> reciprocation. Individuals are<br />
appreciative when an organization provides them with work context that results in a sense <strong>of</strong><br />
empowerment and, in turn, will reciprocate by being more committed to the organization. The findings <strong>of</strong><br />
our study suggest that empowerment not only is associated with greater affective commitment but that it<br />
is strongly related to other forms <strong>of</strong> commitment as well.<br />
As expected, the job-specific aspects <strong>of</strong> perceived employer commitment were significant<br />
predictors <strong>of</strong> job involvement. Nurses with greater job challenge, autonomy, and empowerment reported<br />
significantly higher levels <strong>of</strong> involvement in their job than nurses whose job lacked these aspects <strong>of</strong><br />
perceived employer commitment. For the organization-specific aspects <strong>of</strong> perceived employer<br />
commitment, perceived organizational support was a significant predictor <strong>of</strong> affective commitment,<br />
normative commitment and pr<strong>of</strong>essional commitment. Nurses who perceived their agency cared about<br />
them, valued their contribution, and was committed to them reported significantly greater affective,<br />
normative and pr<strong>of</strong>essional commitment than nurses who lacked this type <strong>of</strong> employer commitment.<br />
Access to development opportunities was a significant predictor <strong>of</strong> affective and normative commitment.<br />
Nurses who felt they had access to such opportunities as training to update current skills and to acquire<br />
new skills, promotion or transfer to a different job, and funding to attend pr<strong>of</strong>essional development<br />
activities reported significantly higher levels <strong>of</strong> affective and normative commitment to their agency than<br />
nurses who lacked access to these development opportunities. Perceived fair treatment was a significant<br />
predictor <strong>of</strong> normative commitment. Nurses who perceived that they were receiving adequate rewards<br />
and recognition reported significantly higher levels <strong>of</strong> normative commitment than nurses who perceived<br />
they were being treated unfairly.<br />
The predictor variables accounted for the least amount <strong>of</strong> variance in continuance commitment.<br />
All three work status (in)congruency variables remained significant predictors <strong>of</strong> continuance<br />
commitment even after the perceived employer commitment variables had been entered. Apart from the<br />
work status (in)congruency variables, only perceived job security and the meaning dimension <strong>of</strong><br />
empowerment were significant predictors <strong>of</strong> continuance commitment. Nurses who perceived less job<br />
security and who did not feel that their job was personally meaningful or important to them reported<br />
significantly higher levels <strong>of</strong> continuance commitment. Allen and Meyer (1996) noted that continuance<br />
commitment has few significant relationships with work-related attitude measures and suggested that<br />
continuance commitment may be an affectively neutral construct. The findings <strong>of</strong> the present study lend<br />
some support to this conjecture. On the other hand, both the ANCOVA and hierarchical regression<br />
findings suggest that work status (in)congruency has important implications for continuance commitment.<br />
Nurses with an incongruent work status reported significantly greater continuance commitment than<br />
nurses with a congruent work status. This means that community health nurses who are being denied<br />
19
their preferred work hours are remaining with their agency because they feel they have to, not because<br />
they feel an emotional attachment or a moral obligation to remain.<br />
Along with the limitations associated with the cross-sectional research design and the selfreported<br />
data, the findings <strong>of</strong> this study may not be generalizable beyond the nursing population.<br />
Although the overall sample size was relatively large with over 1,200 respondents, the number <strong>of</strong> nurses<br />
in the PT/FT group was small compared to the number <strong>of</strong> nurses in the FT/FT, FT/PT, and PT/PT groups.<br />
This reduces the likelihood <strong>of</strong> finding significant differences between the PT/FT group and the other three<br />
groups. All <strong>of</strong> the measures except for continuance commitment had reliability coefficients <strong>of</strong> .7 or above.<br />
We have used this same measure <strong>of</strong> continuance commitment in previous studies and the coefficient<br />
alphas have been in the acceptable range (above .7). However, other researchers have also reported low<br />
reliability coefficients for the continuance commitment scale. For example, Ko, Price and Mueller (1997)<br />
reported a coefficient alpha <strong>of</strong> .58 for the 6-item version <strong>of</strong> Meyer et al.’s (1993) continuance<br />
commitment subscale. The continuance commitment subscale has not been without controversy,<br />
especially in regard to whether it is unidimensional or consists <strong>of</strong> multiple dimensions. The items may be<br />
less applicable to certain work contexts or interpreted differently by different populations. In our case, it<br />
was community health nurses, and in the case <strong>of</strong> Ko et al., it was employees in South Korea.<br />
The present findings have management implications for community health nursing. Our study<br />
suggests that when agencies fail to meet the needs <strong>of</strong> their nurses (i.e., failing to provide nurses with their<br />
preferred work status), nurses may reciprocate by exhibiting reduced organizational commitment, job<br />
involvement, and pr<strong>of</strong>essional commitment. It is important for employers to accommodate the needs and<br />
provide nurses with the flexibility to match their preferred work status to avoid the expenses associated<br />
with work dissatisfaction (Havlovic, Lau & Pinfield, 2002). Campbell, Fowles, and Weber (2004) have<br />
stressed the importance <strong>of</strong> increasing the attractiveness <strong>of</strong> nurses’ jobs to retain a highly motivated<br />
workforce. Thus, the need for agencies to accommodate the needs and preferences <strong>of</strong> its nurses is<br />
becoming more crucial to develop and maintain commitment in the workplace.<br />
Maynard, Thorsteinson and Parfyonova (2006) noted that work status congruency models are<br />
valuable in extending research past simple full-time/part-time differences. However, Maynard et al. also<br />
claimed that even work status congruency models may, by themselves, still be too simplistic. These<br />
researchers argued for the need to incorporate reasons why people engage in full-time or part-time work.<br />
We suggest that future research not only look at the reasons for working full-time or part-time but also the<br />
reasons why employees prefer their current work status or why they prefer a different work status.<br />
Researchers also need to identify the reasons why employees are not able to work their preferred work<br />
status.<br />
The present study examined the relationship <strong>of</strong> work status in(congruency) with organizationspecific<br />
and job-specific aspects <strong>of</strong> perceived employer commitment and with nurses’ commitment to<br />
their organization, to their job, and to their pr<strong>of</strong>ession. In general, the pattern <strong>of</strong> findings reported here<br />
show that nurses with an incongruent work status perceived less commitment from the organization than<br />
nurses with a congruent work status. Thus, it is vital for employers to try and accommodate the needs <strong>of</strong><br />
their employees with regards to work status preferences. Furthermore, our findings suggest that<br />
employers should try to develop methods <strong>of</strong> increasing employees’ empowerment (meaning and outcome)<br />
as having a sense <strong>of</strong> empowerment was found to be the strongest predictor <strong>of</strong> nurses’ commitment.<br />
20
References<br />
Allen, N.J., & Meyer, J.P, “Affective, continuance, and normative commitment to the organization: an<br />
examination <strong>of</strong> construct validity,” Journal <strong>of</strong> Vocational Behavior, 49, (1996), 252-276.<br />
Armstrong-Stassen, M., Al-Ma’aitah, R., Cameron, S.J., & Horsburgh, M.E, “The relationship between<br />
work status congruency and the job attitudes <strong>of</strong> full-time and part-time Canadian and Jordanian<br />
nurses,” The International Journal <strong>of</strong> Human Resource Management, 9, (1998), 41-57.<br />
Armstrong-Stassen, M., Horsburgh, M.E., & Cameron, S.J, “The reactions <strong>of</strong> full-time and part-time<br />
nurses to restructuring in the Canadian health care system,” Paper presented at the annual<br />
convention <strong>of</strong> the Academy <strong>of</strong> Management, Dallas, TX, (1994, August).<br />
Barker, K, “Changing assumptions and contingent solutions: The costs and benefits <strong>of</strong> women working<br />
full- and part-time,” Sex Roles, 28, (1993), 47-71.<br />
Barling, J., & Gallagher, D.G, “Part-time employment,” In C.L. Cooper & I.T. Robertson (Eds.),<br />
International review <strong>of</strong> industrial and organizational psychology Vol. 11 (pp. 243-277).<br />
Chichester, England: John Wiley & Sons, 1996<br />
Bergmann, T.J., Grahn, J.L., & Wyatt, R, “Relationship <strong>of</strong> employment status to employee job<br />
satisfaction,” Akron Business and Economic Review, 17(2), (1986), 45-50.<br />
Blau, G.J., & Boal, K.R, “Conceptualizing how job involvement and organizational commitment affect<br />
turnover and absenteeism,” Academy <strong>of</strong> Management Review, 12, (1987), 288-300.<br />
Blau, P.M., Exchange and power in social life. New York, NY: Wiley, 1964.<br />
Burke, R.J., & Greenglass, E.R, “Work status congruence, work outcomes and psychological well-being,”<br />
Stress Medicine, 16, (2000), 91-99.<br />
Campbell, S. L., Fowles, E. R., & Weber, B. J., “Organizational structure and job satisfaction in public<br />
health nursing,” Public Health Nursing, 21(6), (2004), 564-571.<br />
Comfort, D., Johnson, K., & Wallace, D, “Part-time work and family-friendly practices in Canadian<br />
workplaces,” http://www.statcan.ca, [Accessed 9 July 2003].<br />
Conway, N., & Briner, R.B., “Full-time versus part-time employees: Understanding the links between<br />
work status, the psychological contract, and attitudes,” Journal <strong>of</strong> Vocational Behavior, 61,<br />
(2002), 279-301.<br />
Cooper-Hakim, A., & Viswesvaran, C., “The construct <strong>of</strong> work commitment: Testing an integrative<br />
framework,” Psychological Bulletin, 131, (2005), 241-259.<br />
Edwards, C., & Robinson, O., “Evaluating the business case for part-time working amongst qualified<br />
nurses,” British Journal <strong>of</strong> Industrial Relations, 42, (2004), 167-183.<br />
Eisenberger, R., Huntington, R., Hutchison, S., & Sowa, D., “Perceived organizational support,” Journal<br />
<strong>of</strong> Applied Psychology, 71, (1986), 500-507.<br />
21
Feldman, D.C., & Doerpinghaus, H.I., “Missing persons no longer: Managing part-time workers in the<br />
‘90s,” Organizational Dynamics, Summer, (1992), 59-72.<br />
Feldman, D.C., Leana, C.R., & Turnley, W.H., “A relative deprivation approach to understanding<br />
underemployment,” Trends in Organizational Behavior, 4, (1997), 43-60.<br />
Gakovic, A., & Tetrick, L.E., “Perceived organizational support and work status: a comparison <strong>of</strong> the<br />
employment relationships <strong>of</strong> part-time and full-time employees attending university classes,”<br />
Journal <strong>of</strong> Organizational Behavior, 24, (2003), 649-666.<br />
Gallie, D., White, M., Cheng, Y., & Tomlinson, M., Restructuring the employment relationship. Oxford:<br />
Clarendon Press, 1998.<br />
Gouldner, A.W., “The norm <strong>of</strong> reciprocity: A preliminary statement,” American Sociological Review, 25,<br />
(1960), 161-178.<br />
Hackett, R.D., & Bycio, P., “An examination <strong>of</strong> the job related attitudes <strong>of</strong> part-time versus full-time<br />
employed nurses,” Paper presented at the annual conference <strong>of</strong> the Administrative Sciences<br />
Association <strong>of</strong> Canada, Lake Louise, (1993, June).<br />
Havlovic, S. J., Lau, D. C., & Pinfield, L. T., “Repercussions <strong>of</strong> work schedule congruence among fulltime,<br />
part-time, and contingent nurses,” Health Care Management Review, 27(4), (2002), 30-41.<br />
Hellgren, J., & Sverke, M., “Does job insecurity lead to impaired well-being or vice versa? Estimation <strong>of</strong><br />
cross-lagged effects using latent variable modeling,” Journal <strong>of</strong> Organizational Behavior, 24,<br />
(2003), 215-236.<br />
Holtom, B.C., Lee, T.W., & Tidd, S.T., “The relationship between work status congruence and workrelated<br />
attitudes and behaviors,” Journal <strong>of</strong> Applied Psychology, 87, (2002), 903-915.<br />
Irvine, D., Leatt, P., Evans, M.G., & Baker, R.G., “Measurement <strong>of</strong> staff empowerment within health<br />
service organizations,” Journal <strong>of</strong> Nursing Measurement, 7, (1999), 79-95.<br />
Keil, J.M., Armstrong-Stassen, M., Cameron, S.J., & Horsburgh, M.E., “Part-time nurses: The effect <strong>of</strong><br />
work status congruency on job attitudes,” Applied Psychology: An International Review, 49,<br />
(2000), 227-236.<br />
Ko, J., Price, J.L., & Mueller, C.W. (1997). Assessment <strong>of</strong> Meyer and Allen’s three-component model <strong>of</strong><br />
organizational commitment in South Korea. Journal <strong>of</strong> Applied Psychology, 84, 961-973.<br />
Krausz, M., Sagie, A., & Bidermann, Y., “Actual and preferred work schedules and scheduling control as<br />
determinants <strong>of</strong> job-related attitudes,” Journal <strong>of</strong> Vocational Behavior, 56, (2000), 1-11.<br />
Liden, R.C., Wayne, S.J., & Sparrowe, R.T. “An examination <strong>of</strong> the mediating role <strong>of</strong> psychological<br />
empowerment on the relations between the job, interpersonal relationships, and work outcomes,”<br />
Journal <strong>of</strong> Applied Psychology, 85, (2000), 407-416.<br />
Levanoni, E., & Sales, C.A., “Differences in job attitudes between full-time and part-time Canadian<br />
employees,” Journal <strong>of</strong> Social Psychology, 130, (1990), 231-237.<br />
22
MacDermid, S.M., Dean Lee, M., Buck, M., & Williams, M.L., “Alternative work arrangements among<br />
pr<strong>of</strong>essionals and managers: Rethinking career development and success,” The Journal <strong>of</strong><br />
Management Development, 20, (2001), 305-317.<br />
Markey, R., Hodgkinson, A., & Kowalczyk, J., “Gender, part-time employment and employee<br />
participation in Australian workplaces,” Employee Relations, 24, (2002), 1<strong>29</strong>-150.<br />
Maynard, D.C., Thorsteinson, T.J., & Parfyonova, N.M., “Reasons for working part-time: Subgroup<br />
differences in job attitudes and turnover intentions,” Career Development International, 11,<br />
(2006), 145-162.<br />
McDonald, P.L., & Wanner, R.A., “Part-time work and the older worker in Canada,” In B.O. Warme,<br />
K.L.P. Lundy & L.A. Lundy (Eds.), Working part-time: Risks and opportunities (pp. 187-203).<br />
New York: Praeger, 1992.<br />
Meyer, J.P., & Allen, N.J., Commitment in the workplace: Theory, research, and application. Thousand<br />
Oaks, CA: Sage Publications, 1997.<br />
Meyer, J.P., Allen, N.J., & Smith, C.A., “Commitment to organizations and occupations: Extension and<br />
test <strong>of</strong> a three-component conceptualization,” Journal <strong>of</strong> Applied Psychology, 78, (1993), 538-<br />
551.<br />
Morrow, P.C., McElroy, J.C., & Elliot, S.M., “The effect <strong>of</strong> preference for work status, schedule, and<br />
shift on work-related attitudes,” Journal <strong>of</strong> Vocational Behavior, 45, (1994), 202-222.<br />
Ng, T.W.H., Butts, M.M., Vandenberg, R.J., DeJoy, D.M., & Wilson, M.G., “Effects <strong>of</strong> management<br />
communication, opportunity for learning, and work schedule flexibility on organizational<br />
commitment,” Journal <strong>of</strong> Vocational Behavior, 68, (2006), 474-489.<br />
Paullay, I.M., Alliger, G.M., & Stone-Romero, E.F., “Construct validation <strong>of</strong> two instruments designed to<br />
measure job involvement and work centrality,” Journal <strong>of</strong> Applied Psychology, 79, (1994), 224-<br />
228.<br />
Price, J.L., “Handbook <strong>of</strong> organizational measurement,” International Journal <strong>of</strong> Manpower, 18, (1997),<br />
305-558.<br />
Price, J.L., & Mueller, C.W., Handbook <strong>of</strong> organizational measurement. Marshfield, Mass: Pitman<br />
Publishing Inc., 1986.<br />
Pyper, W., “Employment trends in nursing,” Perspectives on Labour and Income, 5(11), (2004), 5-17.<br />
Reilly, N.P., & Orsak, C.L., “A career stage analysis <strong>of</strong> career and organizational commitment in<br />
nursing,” Journal <strong>of</strong> Vocational Behavior, 39, (1991), 311-330.<br />
Reynolds, J., “You can’t always get the hours you want: Mismatches between actual and preferred work<br />
hours in the U.S.,” Social Forces, 81, (2003), 1171-1199.<br />
Ross, C.E., & Wright, M.P., Women’s work, men’s work, and the sense <strong>of</strong> control,” Work &<br />
Occupations, 25, (1998), 333-355.<br />
23
Schellenberg, G., The changing nature <strong>of</strong> part-time work. Ottawa: The Centre for International Statistics,<br />
Canadian Council on Social Development, 1997.<br />
Skinner, D., “The reality <strong>of</strong> equal opportunities: The expectations and experiences <strong>of</strong> part-time staff and<br />
their managers,” Personnel Review, 28, (1999), 425-438.<br />
Spreitzer, G.M., “Psychological empowerment in the workplace: Dimensions, measurement, and<br />
validation,” Academy <strong>of</strong> Management Journal, 38, (1995), 1442-1465.<br />
Stamper, C.L., & Van Dyne, L., “Work status and organizational citizenship behavior: a field study <strong>of</strong><br />
restaurant employees,” Journal <strong>of</strong> Organizational Behavior, 22, (2001), 517-536.<br />
Stratton, L.S., “Are “involuntary” part-time workers indeed involuntary?” Industrial and Labor Relations<br />
Review, 49, (1996), 522-536.<br />
Sverke, M., Gallagher, D.G., & Hellgren, J., “Alternative work arrangements: Job stress, well-being and<br />
pro-organizational attitudes among employees with different employment contracts,” In K.<br />
Isaksson, C. Hogstedt, C. Eriksson, & T. Theorell (Eds.), Health effects on the new labour market<br />
(pp. 145-168). New York: Plenum, 2000.<br />
Tanksy, J.W., Gallagher, D.G., & Wetzel, K., “The effects <strong>of</strong> demographics, work status and relative<br />
equity on attitudes <strong>of</strong> full and part-time workers,” Canadian Journal <strong>of</strong> Administrative Sciences,<br />
14, (1997), 315-326.<br />
Thorsteinson, T.J., “Job attitudes <strong>of</strong> part-time vs. full-time workers: A meta-analytic review,” Journal <strong>of</strong><br />
Occupational and Organizational Psychology, 76, (2003), 151-177.<br />
Van Emmerik, I.J.H., & Sanders, K., “Mismatch in working hours and affective commitment,” Journal <strong>of</strong><br />
Managerial Psychology, 20, (2005), 712-726.<br />
Wetzel, K., Soloshy, E.D., & Gallagher, D.G., “The work attitudes <strong>of</strong> full-time and part-time registered<br />
nurses,” Health Care Management Review, 3(15), (1990), 79-85.<br />
Zeitz, G., Johannesson, R., & Ritchie, J.E. Jr., “An employee survey measuring total quality management<br />
practices and culture. Group & organization Management, 22, (1997), 414-444.<br />
24
ASAC <strong>2008</strong> Yoshio Yanadori<br />
Halifax, Nova Scotia Sauder School <strong>of</strong> Business<br />
University <strong>of</strong> British Columbia<br />
DO FIRM COMPENSATION STRATEGIES MATTER TO ALL EMPLOYEES? THE SCOPE<br />
OF FIRM COMPENSATION STRATEGIES IN HIGH TECHNOLOGY FIRMS<br />
This study evaluated the extent and nature <strong>of</strong> firm compensation<br />
strategies using employee compensation data from U.S. high-technology<br />
firms. I found that firm pay level strategies influenced the pay level for<br />
strategic employee groups (i.e., managers, R&D employees) to a greater<br />
extent than the pay level for other employee groups. My analyses also<br />
revealed that firms applied consistent pay level and pay mix strategies<br />
across different employee groups.<br />
“(Strategy is defined as) the determination <strong>of</strong> the basic long-term goals and objectives <strong>of</strong> an enterprise,<br />
and the adoption <strong>of</strong> courses <strong>of</strong> action and the allocation <strong>of</strong> resources necessary for carrying out these<br />
goals” (Chandler, 1962: 13).<br />
As suggested by Chandler, resource allocation presents a key element <strong>of</strong> corporate strategy. Granted, the<br />
way firms manage employee compensation will form one important dimension <strong>of</strong> corporate strategy as it<br />
is directly associated with the allocation <strong>of</strong> compensation budget, which accounts for more than 50<br />
percent <strong>of</strong> operating expenses in most firms (Milkovich & Newman, <strong>2008</strong>). Reported relationships<br />
between compensation management and subsequent firm performance endorse the strategic nature <strong>of</strong><br />
compensation management (Gerhart, 2000).<br />
If compensation management is really strategic, firms will exhibit differences in this respect.<br />
Indeed, compensation researchers have long recognized that aspects <strong>of</strong> compensation differ systematically<br />
across firms (Gerhart & Rynes, 2003). Senior managers make strategic decisions when determining their<br />
employee compensation taking various contextual factors into consideration to achieve organizational<br />
goals. Differences in strategic decisions yield inter-firm differences in aspects <strong>of</strong> compensation such as<br />
pay level (e.g., Groshen, 1991) and pay mix (e.g., Gerhart & Milkovich, 1990). Empirical studies have<br />
shown that inter-firm differences in these aspects lead to inter-firm differences in performance (Gerhart,<br />
2000). Building on this notion, compensation textbooks elaborate how senior managers determine these<br />
aspects and their implications to employee behavior and organizational outcomes (Martocchio, 2006;<br />
Milkovich & Newman, <strong>2008</strong>).<br />
Despite the recognition <strong>of</strong> the role firm compensation strategies play in shaping firm<br />
compensation systems, current theory and research <strong>of</strong>fer ambiguous predictions regarding the scope <strong>of</strong><br />
compensation strategies. On the one hand, some studies focused on the compensation for specific<br />
employee groups, typically managerial employees (e.g., Kerr, 1985), and found that firm compensation<br />
strategies shape the compensation for these employees. Implicit in this argument is that the scope <strong>of</strong> firm<br />
compensation strategies is limited; they may not influence compensation for other employee groups. On<br />
the other hand, there are studies that examined employee compensation in firms and reported that firm<br />
compensation strategies influence compensation systems in organizations as a whole (e.g., Boyd &<br />
Salamin, 2001). These studies suggest that the influence <strong>of</strong> firm compensation strategies pierces<br />
throughout the firms. Thus, although compensation researchers agree on the importance <strong>of</strong> firm<br />
compensation strategies, some ambiguity remains about how they influence compensation for different<br />
employee groups within firms in different ways.<br />
25
In this study, I seek to address this issue using employee compensation data from US hightechnology<br />
firms. The data include compensation information for employees below the executive level<br />
holding different jobs (i.e., research and development (R&D), technical, administrative) with different<br />
responsibilities (i.e., middle-level managers, non-managerial employees). I examine how firm<br />
compensation strategies shape compensation systems for these employees by focusing on two dimensions:<br />
pay level and pay mix, which have been rigorously studied by prior researchers.<br />
Effects <strong>of</strong> Firm Compensation Strategies<br />
Theory Development<br />
Organizational variation in compensation management forms one key research area for labor<br />
economists and management researchers. Among various aspects <strong>of</strong> compensation management, much <strong>of</strong><br />
research effort has been devoted to understand inter-firm differences in pay level and pay mix (Gerhart,<br />
2000).<br />
While pay rates are determined to a large extent by the supply and demand in the labor market,<br />
inter-firm differences in pay rates have consistently been reported (Gerhart & Rynes, 2003). It is in part<br />
attributable to the differences in employees’ human capital retained, but inter-firm differences still persist<br />
after controlling for such worker effects (Groshen, 1991). Empirical investigations revealed that some<br />
firm-level characteristics such as firm size, pr<strong>of</strong>itability, and industry membership are systematically<br />
associated with firm pay level (Gerhart & Milkovich, 1990). Beyond these firm characteristics, Gerhart<br />
and Milkovich (1990) explicitly claimed that firm compensation strategies play a role in the determination<br />
<strong>of</strong> firm pay level. Examining managerial pay, they reported that firm pay rates are systematically different<br />
after controlling for employee human capital, job characteristics, and firm characteristics. The authors<br />
contended that firm-specific effects reflected firm compensation strategies.<br />
Employee compensation consists <strong>of</strong> various pay programs (e.g., base pay, short-term incentives,<br />
long-term incentives, and benefits) and anecdotal evidence indicates that firms have discretion in mixing<br />
different programs (e.g., Lincoln Electric, Starbucks, SAS Institute). Similar to pay level, Gerhart and<br />
Milkovich (1990) reported that inter-firm differences in pay mix for managerial employees, measured as<br />
the ratio <strong>of</strong> bonus to base pay, are still apparent after controlling for observable firm characteristics (i.e.,<br />
firm size and performance), as well as worker characteristics and job characteristics. Their empirical<br />
analyses revealed that firm specific effects accounted for a greater proportion <strong>of</strong> the variance in pay mix<br />
than they accounted for the proportion <strong>of</strong> the variance in pay level. Based on this result, the authors<br />
claimed that firms have more discretion in pay mix than in pay level.<br />
This study’s first set <strong>of</strong> hypotheses intends to confirm the effects <strong>of</strong> firm compensation strategies<br />
on employee compensation in high-technology firms. It is claimed that high-technology workers are less<br />
hesitant to change their employers in pursuit <strong>of</strong> better working conditions (Cappelli, 1999). Due to a<br />
higher degree <strong>of</strong> worker mobility, firms may have difficulty in deviating from market pay rates. For pay<br />
mix, most firms in the high-technology industry had adopted stock-based compensation, and it is regarded<br />
as the norm in this industry (National Center for Employee Ownership, 2001). Taken together, firms may<br />
have limited discretion in shaping employee compensation in high-technology industry. Given these<br />
industry characteristics, it is important to test the effects <strong>of</strong> firm compensation strategies on employee<br />
compensation. If firm compensation strategies really play a role in shaping employee compensation, firms<br />
will exhibit systematic differences in pay level and pay mix after controlling for observable job<br />
characteristics and firm characteristics, reflecting the differences in firm compensation strategies.<br />
26
Hypothesis 1a: Employee pay level is systematically different across high-technology firms.<br />
Hypothesis 1b: Employee pay mix is systematically different across high-technology firms.<br />
While this study’s primary interest is to test whether or not high-technology firms exhibit<br />
systematic differences in pay level and pay mix, it is also interested in evaluating the degree to which<br />
firm-specific effects account for the variance in employee pay level and mix. I will also compare firmspecific<br />
effects on pay level and those on pay mix to see if firm has more discretion in determining pay<br />
mix than in determining pay level, as reported by Gerhart and Milkovich (1990).<br />
Scope <strong>of</strong> Firm Compensation Strategies<br />
Within-firm differences. Much <strong>of</strong> prior research on firm compensation strategies examined<br />
their effects on compensation for a particular employee group, rather than compensation for the entire<br />
employees in the firm. The majority <strong>of</strong> studies examined the effects <strong>of</strong> firm compensation strategies on<br />
managerial compensation (Kerr, 1985; Gerhart & Milkovich, 1990).Yet, some studies analyzed other<br />
employee groups such as research and development employees (R&D) in high-technology firms<br />
(Yanadori & Marler, 2006), faculty members in research oriented universities (Gomez-Mejia & Balkin,<br />
1992a), and customer service representatives in call centers (Batt, 2001).<br />
This line <strong>of</strong> research claims that firm decision makers pay closer attention to the compensation<br />
for “strategic employee groups,” employee groups that make critical contributions to firm competitive<br />
advantage (Gomez-Mejia & Balkin, 1992b: 101; Milkovich, 1988: 266). Given the fact that firm success<br />
hinges on the performance <strong>of</strong> strategic employee groups, senior managers elaborate firm compensation<br />
strategies in such a way that they encourage their strategic employee groups to exhibit desirable behaviors.<br />
As a result, firm compensation strategies primarily influence the compensation for strategic employee<br />
groups. To the author’s knowledge, Yanadori and Marler (2006) is the only empirical study that explicitly<br />
drew upon the notion <strong>of</strong> strategic employee groups (i.e., R&D employees in high-technology firms);<br />
nevertheless, aforementioned studies focused on specific employee groups (i.e., R&D employees, faculty<br />
members, customer service representatives) as they played a critical role in firm effectiveness and thus<br />
firm compensation strategies were more apparent in their compensation.<br />
Traditionally strategic employee groups referred to managers as their behavior directly influences<br />
firm performance irrespective <strong>of</strong> industry or firm business strategy (Milkovich, 1988). Accordingly<br />
researchers have typically examined the effects <strong>of</strong> firm compensation strategies on executive<br />
compensation (Berkema & Gomez-Mejia, 1998). Recently, researchers have extended this model to<br />
middle-level managers (e.g., Kerr, 1986) as their contributions to firm effectiveness have been well<br />
established. Several empirical studies reported that not only senior executives but also middle-level<br />
managers’ compensation is associated with firm performance (e.g., Gerhart & Milkovich, 1990). If<br />
managers make up a part <strong>of</strong> strategic employee groups, firm compensation strategies will influence their<br />
pay level and pay mix to a greater extent than other employees’ pay level and pay mix. Consequently, I<br />
hypothesize that firm compensation strategies influence managerial employees’ compensation to a greater<br />
extent than they influence non-managerial employees’ compensation.<br />
Hypothesis 2a: Firm-specific effects on pay level are greater for managerial employees’<br />
compensation than for non-managerial employee compensation in high-technology firms.<br />
Hypothesis 2b: Firm-specific effects on pay mix are greater for managerial employees’ compensation<br />
than for non-managerial employee compensation in high-technology firms.<br />
As I will subsequently discuss in the method section, this study’s empirical approach does not<br />
directly measure firm compensation strategies. Rather, consistent with prior studies (Gerhart & Milkovich,<br />
1990; Gerlach & Stephan, 2006), I estimate firm-specific effects, which I regard reflect firm<br />
compensation strategies.<br />
27
On top <strong>of</strong> middle-level managers, research and development (R&D) employees will make up the<br />
strategic employee groups in high technology firms as firm success <strong>of</strong>ten hinges on innovation (Gomez-<br />
Mejia & Balkin, 1992b). Admittedly the contributions R&D employees make to their firms could vary<br />
depending on firm business strategy (e.g., concentrate on innovation, or emphasize marketing);<br />
nevertheless, research shows that how firms manage their employees that engage in developing<br />
innovation influences high-technology firms’ performance (Collins & Smith, 2006). Granted, firms will<br />
carefully develop firm compensation strategies for R&D employees’ compensation and thus firm effects<br />
will be more apparent for their pay level and mix. It follows that firm effects on pay level and mix for<br />
R&D employees will be greater than firm effects on pay level and mix for non-R&D employees.<br />
Hypothesis 3a: Firm-specific effects on pay level are greater for R&D employees’ compensation<br />
than for other employee groups’ compensation in high-technology firms.<br />
Hypothesis 3b: Firm-specific effects on pay mix are greater for R&D employees’ compensation than<br />
for other employee groups’ compensation in high-technology firms.<br />
Nature <strong>of</strong> within-firm differences. The research that focused on the compensation for specific<br />
employee groups appears to assume that firm compensation strategies have a limited influence on the<br />
compensation for other employee groups. Yet there are also studies that analyzed compensation for all<br />
employees and identified the influence <strong>of</strong> firm compensation strategies throughout the firms (Balkin &<br />
Gomez-Mejia, 1990; Boyd & Salamin, 2001). Hence, extant empirical evidence <strong>of</strong>fers an ambiguous<br />
picture regarding how firm compensation strategies influence compensation for different employee<br />
groups in different manners.<br />
More broadly, the field <strong>of</strong> human resource management faces a similar theoretical challenge as to<br />
whether firms apply similar practices to different employee groups or not. A line <strong>of</strong> strategic human<br />
resource management research has recognized distinct intra-firm differences in human resource practices,<br />
which are grounded in the differences in employee contributions to firm competitiveness (Lepak & Snell<br />
1999, Osterman, 1987). It essentially argues that, given the differences in the type and level <strong>of</strong><br />
contributions different employee groups make to firm competitive advantage, employers form different<br />
types <strong>of</strong> employment relationship, adopting different human resource practices. In spite <strong>of</strong> this argument,<br />
the majority <strong>of</strong> strategic human resource management studies captured the entire employees in firms for<br />
their unit <strong>of</strong> analysis (e.g., Arthur, 1994; Datta, Guthrie, & Wright, 2005). These studies implicitly<br />
assumed that similar practices were applied to employees throughout the firms. In short, similar to firm<br />
compensation strategies, current theory and research are ambiguous as to how firm human resource<br />
strategies shape human resource practices for different employee groups in different managers.<br />
In the context <strong>of</strong> compensation management, assuming that firm compensation strategies<br />
influence the compensation for strategic employee groups, I contend that firms apply consistent strategies<br />
to compensation for other employee groups. Equity theory (Adams, 1965) <strong>of</strong>fers a primary underpinning<br />
for this expectation. This theory claims that individuals compare their pay with their comparators’ and<br />
they choose various persons for their comparators including their colleagues holding different jobs in the<br />
same firms. If some groups <strong>of</strong> employees are paid above the market, while other groups are paid below<br />
the market, underpaid employee groups may find the situation inequitable and exhibit negative reactions.<br />
Concerning pay mix, different pay mix arrangements are associated with different level <strong>of</strong> employee<br />
income risk (Eisenhardt, 1988). If employees in one group find their compensation is more variable (so<br />
risky), they may find their firms’ decisions unfair. However, if income risk level is shared in a consistent<br />
manner across employee groups, such the sense <strong>of</strong> inequity may be minimized.<br />
Hence, I propose that firms are likely to apply consistent pay level and mix strategies across<br />
employee groups. High-technology firms determine compensation strategies for managerial employees<br />
28
and R&D employees, who are strategic employee groups, and in order to maintain equity across<br />
employee groups, consistent pay level and mix strategies will be applied to other employees (i.e., nonmanagerial<br />
employees, non-R&D employees).<br />
Hypothesis 4a: Firm-specific effects on managerial employees’ pay level are positively associated<br />
with firm-specific effects on non-managerial employees’ pay level in high-technology firms.<br />
Hypothesis 4a: Firm-specific effects on managerial employees’ pay mix are positively associated with<br />
firm-specific effects on non-managerial employees’ pay mix in high-technology firms.<br />
Hypothesis 5a: Firm-specific effects on R&D employees’ pay level are positively associated with<br />
firm-specific effects on other employee groups’ pay level in high-technology firms.<br />
Hypothesis 5a: Firm-specific effects on R&D employees’ pay mix are positively associated with firmspecific<br />
effects on other employee groups’ pay mix in high-technology firms.<br />
Overall, I claim that firms differentiate pay level and pay mix strategies among employee groups<br />
in the sense that these strategies influence different employee groups’ compensation to a different extent.<br />
Yet, as firms apply consistent pay level and pay mix strategies to their employees within organizations,<br />
inter-firm differences in pay level and mix are still apparent.<br />
Data and Variables<br />
Methods<br />
Data. This study’s analysis drew upon an employee-level compensation data set collected as a<br />
part <strong>of</strong> an annual compensation survey administered to US high technology firms by Clark Consulting, a<br />
Boston-based consulting company. Clark Consulting aggregates individual employee compensation<br />
information from participating firms, and provides them with market pay information.<br />
I analyzed the compensation data in 1997 and 1998. The data in both years include information<br />
on individual employee compensation from approximately 100 firms, both public and private, operating<br />
in the high-technology industry. Because the survey is administered annually, firms and employees<br />
included vary between these two years. I restricted our sample to publicly-traded firms whose firm<br />
characteristics information is available. The 1997 data set includes compensation information for 82,365<br />
employees from 73 firms and the 1998 data set includes compensation information for 102,881<br />
employees from 94 firms. Firm characteristics information was collected from Standard & Poor’s<br />
COMPUSTAT.<br />
Table 1 summarizes information about the firms in my data sets. It presents the means <strong>of</strong> the<br />
following statistics: number <strong>of</strong> employees, firm sales (in million dollars), return on assets (ROA, in<br />
percent), and research and development (R&D) intensity (in percent). The R&D intensity for a firm was<br />
calculated by dividing R&D expenditure by sales. The high-technology category includes firms with an<br />
R&D intensity measure that meets the 5.0 percent threshold or exceeds it. As these figures suggest, the<br />
majority <strong>of</strong> firms in our dataset were established high-technology firms.<br />
<strong>29</strong>
TABLE 1<br />
Descriptive Statistics<br />
1997 1998<br />
data data<br />
Sales $7,277 million $7,859 million<br />
Number <strong>of</strong> employees* 32,279 33,823<br />
ROA 2.90% 3.63%<br />
R&D intensity* 12.65% 13.39%<br />
* N = 73 for the 1997 data set and 94 for the 1998 data set. This information is not available<br />
for some firms. Mean statstics are calculated among firms with information.<br />
Dependent variables. I evaluated the extent and nature <strong>of</strong> firm compensation strategies in two<br />
aspects <strong>of</strong> compensation management: pay level and pay mix. For the measure <strong>of</strong> pay level, I analyzed<br />
base pay and total pay. Total pay is the sum <strong>of</strong> base pay, payment from financial incentive plans (i.e.,<br />
pr<strong>of</strong>it sharing, stock awards), and various employee benefits. Monetary values <strong>of</strong> stock awards were<br />
calculated using Black-Scholes model, and the monetary values <strong>of</strong> employee benefits represented<br />
hypothetical values that an employee would need to purchase equivalent plans in the marketplace.<br />
With respect to pay mix, I examined the ratio <strong>of</strong> financial incentives to base pay and the ratio <strong>of</strong><br />
long-term pay to short-term pay. The relative emphasis on financial incentives in employee compensation<br />
has been studied rigorously and their inter-firm differences are well acknowledged (Gerhart & Milkovich,<br />
1990). Researchers also recognize that firms mix pay components with different time horizons (i.e., shortterm<br />
pay vs. long-term pay, Gomez-Mejia & Balkin, 1992b; Yanadori & Marler, 2006) in different<br />
manners. Short-term pay includes base pay, pr<strong>of</strong>it sharing, and other short-term incentives, and long-term<br />
pay includes stock awards. I hereafter refer to these pay mix variables as incentive intensity and long-term<br />
to short-term ratio, respectively. I applied logarithmic transformation to all pay variables as their<br />
distributions were skewed.<br />
Independent variables. I estimated my four pay variables (i.e., base pay, total pay, incentive<br />
intensity, and long-term to short-term ratio) using six variables. The first variable is region. Many firms in<br />
the high-technology industry are clustered in specific geographical locations. In this study, I focused on<br />
the compensation information <strong>of</strong> employees that worked in one <strong>of</strong> the following three locations: Silicon<br />
Valley in California, Austin Texas, and the Route 128 area in Massachusetts. The second variable is job<br />
level. Our data sets include 8 job levels. Levels 1-5 are pr<strong>of</strong>essional contributors (= non-managerial<br />
employees), and level 6-8 are middle-level managers. A higher number represents a higher job level. The<br />
third variable is job. My data sets consist <strong>of</strong> employees holding different jobs. The 1997 data set consists<br />
<strong>of</strong> employees holding 49 jobs, and the 1998 data set consists <strong>of</strong> employees holding 55 jobs. The fourth<br />
variable is firm. The 1997 data set includes pay information from 73 firms, and the 1998 data set includes<br />
pay information from 94 firms. All these four variables were treated as categorical variables.<br />
The remaining two variables were related to observable firm characteristics. Past research has<br />
consistently reported that firm size and performance are key influences on firm pay level and pay mix. I<br />
used firm sales 1 and return on assets (ROA) as the proxies <strong>of</strong> firm size and firm performance, respectively.<br />
Logarithmic transformation was applied to firm sales as its distribution was skewed.<br />
1 This study did not use the number <strong>of</strong> employees for the proxy <strong>of</strong> firm size as Compustat data base does not include<br />
this information for some firms. We believe that using the number <strong>of</strong> employee in our analysis would not change our<br />
results because firm sales and the number <strong>of</strong> employees was highly correlated in our sample (r = 0.96 for both the<br />
1997 data set the 1998 data set).<br />
30
Analysis<br />
Firm-specific effects on pay level and pay mix. I empirically evaluated the extent to which<br />
firm explained the variance in pay level and pay mix after controlling for other relevant factors (i.e.,<br />
region, job level, job, firm, firm size, and firm performance). Following previous studies (e.g., Gerhart &<br />
Milkovich, 1990), I regarded firm-specific effects as the influences <strong>of</strong> firm compensation strategies.<br />
In order to understand the firm-specific effects on compensation management, I ran two models.<br />
The first model estimated fixed firm effects, which are firm-specific effects after controlling for region,<br />
job level, and job. They essentially represented the degree to which a firm’s pay level or mix deviated<br />
from a standard firm’s pay level and pay mix. The second model estimated idiosyncratic effects, the<br />
portion <strong>of</strong> the fixed firm effects that was not explained by observable firm-level characteristics (i.e., firm<br />
size and performance; Gerlach & Stephan, 2006). For estimating fixed firm effects, I used the following<br />
model:<br />
PDijkm = µ + αi + βj + χk + δm + εijkm (1)<br />
PDijkm denotes the pay variable (i.e., base pay, total pay, incentive intensity, or long-term to shortterm<br />
pay) in region i (i = Silicon Valley, Austin, or the Route 128 area), <strong>of</strong> job level j (j = 1, 2, 3,..8), job<br />
k (k = 1, 2, 3,…49 for the 1997 data set and 1, 2, 3, …55 for the 1998 data set), in firm m (m = 1, 2,<br />
3, ..73 for the 1997 data set and 1, 2, 3, ..94 for the 1998 data set). µ is the sample mean, and α, β, χ, and<br />
δ represent the effects <strong>of</strong> region, job family, job level, and firm, respectively. I calculated the increases in<br />
R-squared by adding variables in the following hierarchical order: region, job level, job, and firm. The<br />
change in R-squared when firm was added to the model was captured as the proportion <strong>of</strong> the variance<br />
explained by fixed firm effects.<br />
For estimating idiosyncratic effects, I used the same model but substituted firm size and firm<br />
performance variable for the firm categorical variable. Similar to the analysis on fixed firm effects, I<br />
added independent variables in the order <strong>of</strong> region, job level, job, and firm size and performance. The<br />
change in R-squared was calculated when I added firm size and performance, which I regarded the part <strong>of</strong><br />
fixed firm effects that was accounted for by firm size and performance. I then subtracted the change in Rsquared<br />
attributed to firm size and firm performance from the change in R-squared attributed to fixed firm<br />
effects. The difference represents the proportion <strong>of</strong> the variance explained by idiosyncratic effects (i.e.,<br />
fixed firm effects unexplained by firm size or performance).<br />
I first estimated fixed firm effects and idiosyncratic effects using all observations that included<br />
employees holding different jobs in different job levels. Next, as I hypothesized the effects <strong>of</strong> firm<br />
compensation strategies will be different across employee groups, I estimated firm-specific effects<br />
according to different employee groups separately. Employees were sorted into either middle-level<br />
managers (Levels 6 to 8) or non-managerial employees (Levels 1 to 5). Employee jobs (49 jobs in the<br />
1997 data set and 55 jobs in the 1998 data set) were categorized into three large job families: R&D,<br />
technical, and administrative. As a result, I estimated firm effects according to six employee groups: R&D<br />
managers, R&D non-managerial employees, technical managers, technical non-managerial employees,<br />
administrative managers, and administrative non-managerial employees.<br />
I expected that the effects <strong>of</strong> firm compensation strategies to be bigger on managerial<br />
compensation than on non-managerial employees’ compensation. Likewise, I hypothesized that the<br />
effects to be bigger on R&D employees’ compensation than on other employees’ compensation. In order<br />
to test these hypotheses, I estimated the correlation coefficients between compensation variables and firm<br />
effects by taking the square root <strong>of</strong> the R-squared calculated in the previous step. I then applied Fisher’s<br />
z-test (Cohen, Cohen, West, & Aiken, 2003) to examine if the correlation coefficients were significantly<br />
31
different between managers and non-managerial employees or between R&D employees and other<br />
employees.<br />
Consistency across jobs. In order to evaluate the consistency <strong>of</strong> firm compensation strategies<br />
across employee groups, I first estimated each firm’s fixed firm effects and idiosyncratic effects on this<br />
study’s pay variables according to six employee groups. Specifically, fixed firm effects were estimated<br />
using the model specified in (1). Then I regressed firm size and firm performance on the estimated fixed<br />
firm effects. The residuals <strong>of</strong> this regression, interpreted as the portion <strong>of</strong> fixed firm effects unexplained<br />
by firm size or performance, were used as the proxy for each firm’s idiosyncratic firm effects. In this<br />
regression, I took a one-year lag for these two firm characteristics.<br />
I then calculated the correlation coefficients between the firm effects for managers and nonmanagers<br />
within job families (i.e., R&D managers vs. R&D non-managerial employees, technical<br />
managers vs. technical non-managerial employees, and administrative managers vs. administrative nonmanagers)<br />
to see if firm compensation strategies for managers are associated with those for nonmanagerial<br />
employees. With respect to consistency across job families, I calculated correlation<br />
coefficients among R&D employees, technical employees, and administrative employees to see if firm<br />
compensation strategies for R&D employees are associated with those for technical employees or<br />
administrative employees.<br />
Results<br />
Tables 2a and 2b present the amount <strong>of</strong> the variance (in percent) in this study’s pay variables<br />
explained by six independent variables. For each pay variable, I first present the result <strong>of</strong> the analysis that<br />
examined all employees, and then I display the results <strong>of</strong> the analyses on different employee groups. The<br />
numbers in the column <strong>of</strong> “Firm” represent the proportion <strong>of</strong> the variance explained by fixed firm effects.<br />
The numbers in the column <strong>of</strong> “Firm effects not explained by firm size or performance” represent the<br />
proportion <strong>of</strong> the variance attributable to idiosyncratic firm effects. The interpretation <strong>of</strong> the results does<br />
not change according to firm effects used (i.e., fixed-firm effects or idiosyncratic effects). Thus, for the<br />
ease <strong>of</strong> discussion, I present my interpretation <strong>of</strong> the results by focusing on the results <strong>of</strong> idiosyncratic<br />
firm effects (i.e., column “Firm effects not explained by firm size or performance”).<br />
The analyses using all employees show that firms were distinctly different in their pay level and<br />
pay mix. While the degree varied across variables and between years, idiosyncratic firm effects were all<br />
significant at the p < 0.001 level. These results supported Hypotheses 1a and 1b. Both pay level and pay<br />
mix were systematically different across firms, indicating that firms developed distinct compensation<br />
strategies for their pay level and pay mix. The comparison across pay variables revealed that idiosyncratic<br />
firm effects explained the variance in pay mix variables (i.e., incentive intensity and long-term to shortterm<br />
ratio) to a greater extent that they explained the variance in pay level variables (i.e., base pay and<br />
total pay). This result was consistent with Gerhart and Milkovich (1990). I also found that the proportion<br />
<strong>of</strong> the variance in total pay explained by idiosyncratic firm effects were greater than the proportion <strong>of</strong> the<br />
variance in base pay explained by idiosyncratic firm effects. This result may suggest that firms in my<br />
sample took market pay rates into account more seriously when they determined employee base pay level<br />
than when they determined total pay level.<br />
Next, I examined the degree to which idiosyncratic firm effects on compensation management<br />
varies across employee groups. Overall, my hypotheses were supported for the determination <strong>of</strong> employee<br />
pay level, but they were not supported for the determination <strong>of</strong> employee pay mix. Idiosyncratic firm<br />
effects influenced managerial employees’ pay level to a greater extent than they influenced nonmanagerial<br />
employees’ pay level. For instance, idiosyncratic effects accounted for 19.54% <strong>of</strong> the variance<br />
in R&D managers’ base pay in 1997, whereas they accounted for 7.24% <strong>of</strong> the variance in R&D non-<br />
32
managerial employees’ base pay in the same year (z = 13.69, p < 0.001). Likewise, idiosyncratic effects<br />
accounted for 25.35% <strong>of</strong> the variance in R&D managers’ total pay in 1997, whereas they accounted for<br />
12.91% <strong>of</strong> the variance in administrative non-managerial employees’ total pay in the same year (z = 12.24,<br />
p < 0.001). This pattern was consistent for the comparisons between technical managers and technical<br />
non-managerial employees, and comparisons between administrative managers and administrative nonmanagerial<br />
employees. Idiosyncratic firm effects accounted for significantly greater variance in<br />
managerial employees’ pay level than they did non-managerial employees’ pay level. The results were<br />
essentially the same in the analyses <strong>of</strong> the 1998 data set. Thus, Hypotheses 2a was supported.<br />
As we hypothesized, idiosyncratic firm effects influenced R&D employees’ pay level to a greater<br />
extent than other employees’ pay level. For instance, idiosyncratic firm effects accounted for a greater<br />
proportion <strong>of</strong> R&D manager’s base pay than technical managers’ base pay (19.54% vs. 13.27%, z = 4.34,<br />
p < 0.001) and administrative managers’ base pay (19.54% vs. 8.86%, z = 8.11, p < 0.001) in 1997.<br />
Idiosyncratic firm effects explained a greater proportion <strong>of</strong> R&D non-managerial employees’ base pay<br />
than technical non-managerial employees’ base pay (7.24% vs. 3.18%, z = 11.07 , p < 0.001) and<br />
administrative non-managerial employees’ base pay (7.24% vs. 3.26%, z = 8.75, p < 0.001) in 1997. This<br />
pattern was consistent for total pay in 1997. The results were essentially consistent for the analyses <strong>of</strong> the<br />
1998 data set except that the comparison <strong>of</strong> idiosyncratic effects on total pay between R&D managers and<br />
those technical managers (<strong>29</strong>.51% vs. 28.23%, z = 0.63, p = 0.53). Hence, Hypothesis 3a was supported.<br />
In contrast, I failed to obtain the strong evidence that shows idiosyncratic firm effects accounted<br />
for managers’ pay mix than non-managerial employees’ pay mix. In my analyses on the 1997 data set,<br />
idiosyncratic firm effects generally accounted for the variance in non-managerial employees’ incentive<br />
intensity to a greater extent than they accounted for the variance in managers’ incentive intensity (22.96%<br />
for managers vs. 25.87% for non-managerial employees in R&D job family, 22.83% vs. 26.28% within<br />
technical job family, and 18.77% vs. 31.<strong>29</strong>% within administrative job family). Although idiosyncratic<br />
firm effects influenced managers’ pay mix more than non-managers’ pay mix in some occasions (e.g.,<br />
42.73% <strong>of</strong> R&D managers’ long-term to short-term ratio vs. 32.51% <strong>of</strong> R&D non-managerial employees’<br />
long-term to short-term ratio in 1998, z = 9.86, p < 0.000), my analyses revealed no consistent patterns.<br />
Thus, I concluded that Hypothesis 2b was not supported.<br />
Similarly, I did not find the strong evidence for the greater influence <strong>of</strong> idiosyncratic firm effects<br />
on R&D employees’ pay mix than on other employees’ pay mix. Hypothesis 3b was not supported.<br />
Finally, I examined the correlations among fixed firm effects and idiosyncratic firm effects for<br />
different employee groups. Tables 3A and 3B shows the relationships between firm effects on pay level<br />
among three job families. Most combinations yielded highly significant correlations. The only exception<br />
was the correlation between the firm effects on R&D non-managerial employees’ base pay and those on<br />
administrative non-managerial employees’ base pay in 1998. Tables 4A and 4B present the relationship<br />
between firm effects on pay mix. All combinations were statistically significant at the p < 0.001 level.<br />
Table 5 summarizes the correlations between firm effects on managerial employees’ pay level and mix<br />
and firm effects on non-managerial employees pay level and mix. Correlation coefficients were all highly<br />
significant, too. From these results, I concluded that firm strategies on pay mix and pay level were<br />
consistently applied across employee groups. These results <strong>of</strong>fered the support for Hypotheses 4a, 4b, 4c,<br />
and 4d. Note that my analyses did not establish causal relationship (i.e., compensation strategies for<br />
managers influence compensation strategies for non-managerial employees, compensation strategies for<br />
R&D employees influence compensation strategies). My analyses simply revealed that firm compensation<br />
strategies were applied consistently across employee groups.<br />
33
Pay variable<br />
TABLE 2A<br />
Firm Effects on Compnesation Management in 1997<br />
Percent <strong>of</strong> variance attributable to:<br />
Region Job level Job family Firm<br />
Firm size and<br />
performance<br />
Firm effects not<br />
explained by<br />
firm size or<br />
performance<br />
Error Total<br />
Base pay All employees 7.47% 65.25% 7.20% 3.22% 0.07% 3.15% 16.86% 100.00%<br />
R&D Managers 16.37% 34.51% 1.20% 19.73% 0.19% 19.54% 28.19% 100.00%<br />
R&D Non-managers 11.61% 61.94% 0.70% 7.98% 0.74% 7.24% 17.78% 100.00%<br />
Technical Managers 8.02% 36.93% 6.<strong>29</strong>% 14.68% 1.41% 13.27% 34.08% 100.00%<br />
Technical Non-managers 10.00% 61.85% 3.36% 3.26% 0.08% 3.18% 21.53% 100.00%<br />
Administrative Managers 7.84% 47.27% 5.05% 9.87% 1.00% 8.86% <strong>29</strong>.98% 100.00%<br />
Administrative Non-managers 4.34% 52.06% 15.70% 3.61% 0.35% 3.26% 24.30% 100.00%<br />
Total pay All employees 7.70% 55.86% 5.25% 11.88% 4.22% 7.66% 19.32% 100.00%<br />
R&D Managers 9.70% 23.30% 2.22% 36.44% 11.09% 25.35% 28.34% 100.00%<br />
R&D Non-managers 11.91% 48.57% 0.68% 18.59% 5.68% 12.91% 20.25% 100.00%<br />
Technical Managers 5.32% 26.61% 7.30% 27.77% 6.32% 21.45% 33.00% 100.00%<br />
Technical Non-managers 9.18% 49.68% 3.98% 14.41% 5.72% 8.69% 22.76% 100.00%<br />
Administrative Managers 8.62% 36.80% 6.54% 23.67% 8.70% 14.97% 24.37% 100.00%<br />
Administrative Non-managers 7.49% 47.72% 13.07% 13.01% 5.15% 7.86% 18.70% 100.00%<br />
Incentive All employees 10.92% 6.79% 4.65% 39.11% 16.01% 23.09% 38.53% 100.00%<br />
intensity R&D Managers 6.66% 4.38% 5.40% 38.89% 15.92% 22.96% 44.67% 100.00%<br />
R&D Non-managers 10.61% 3.73% 5.26% 42.18% 16.31% 25.87% 38.21% 100.00%<br />
Technical Managers 7.78% 4.79% 3.97% 35.10% 12.27% 22.83% 48.36% 100.00%<br />
Technical Non-managers 11.14% 2.08% 5.52% 45.27% 18.99% 26.28% 35.99% 100.00%<br />
Administrative Managers 13.60% 10.09% 2.63% 32.95% 14.18% 18.77% 40.74% 100.00%<br />
Administrative Non-managers 18.96% 2.14% 4.52% 48.40% 17.10% 31.<strong>29</strong>% 25.99% 100.00%<br />
Long-term All employees 2.84% 6.58% 5.77% 40.20% 17.93% 22.27% 44.61% 100.00%<br />
to short-term R&D Managers 2.03% 2.92% 4.73% 50.08% 21.25% 28.83% 40.24% 100.00%<br />
ratio R&D Non-managers 3.72% 1.65% 5.20% 45.54% 22.52% 23.02% 43.89% 100.00%<br />
Technical Managers 3.43% 2.39% 9.54% 35.44% 8.70% 26.74% 49.20% 100.00%<br />
Technical Non-managers 2.45% 0.94% 8.99% 41.85% 16.18% 25.67% 45.77% 100.00%<br />
Administrative Managers 4.75% 10.25% 2.74% 40.00% 12.74% 27.25% 42.26% 100.00%<br />
Administrative Non-managers 5.33% 2.02% 7.60% 51.56% 18.11% 33.45% 33.49% 100.00%<br />
N = 82,365 for overall, 5,452 for R&D managers, 36,322 for R&D non-managers, 3,634 for technical managers, 21,252 for technical non-managers, 4,094 for administrative<br />
managers, and 11,601 for administrative non-managers<br />
34
Pay variable<br />
TABLE 2B<br />
Firm Effects on Compnesation Management in 1998<br />
Percent <strong>of</strong> variance attributable to:<br />
Region Job level Job family Firm<br />
Firm size and<br />
performance<br />
Firm effects not<br />
explained by<br />
firm size or<br />
performance<br />
Error Total<br />
Base pay All employees 6.95% 67.45% 6.42% 2.45% 0.02% 2.43% 16.73% 100.00%<br />
R&D Managers 18.65% 28.08% 1.43% 21.92% 0.21% 21.72% <strong>29</strong>.91% 100.00%<br />
R&D Non-managers 12.87% 65.90% 0.36% 5.05% 0.38% 4.67% 15.81% 100.00%<br />
Technical Managers 6.66% 34.88% 6.96% 16.93% 0.93% 16.00% 34.56% 100.00%<br />
Technical Non-managers 6.18% 66.87% 2.27% 3.57% 0.24% 3.32% 21.11% 100.00%<br />
Administrative Managers 7.39% 41.56% 8.12% 14.50% 1.73% 12.77% 28.43% 100.00%<br />
Administrative Non-managers 6.55% 53.68% 12.16% 3.72% 0.57% 3.15% 23.88% 100.00%<br />
Total pay All employees 7.32% 57.19% 4.88% 9.88% 0.98% 8.90% 20.73% 100.00%<br />
R&D Managers 9.46% 19.41% 1.50% 34.97% 5.81% <strong>29</strong>.15% 34.67% 100.00%<br />
R&D Non-managers 13.01% 51.57% 1.24% 13.43% 1.39% 12.05% 20.74% 100.00%<br />
Technical Managers 5.13% 22.94% 3.51% 31.32% 3.09% 28.23% 37.09% 100.00%<br />
Technical Non-managers 7.26% 54.88% 2.75% 10.92% 0.48% 10.44% 24.18% 100.00%<br />
Administrative Managers 6.52% 30.96% 8.66% 26.69% 4.14% 22.55% 27.17% 100.00%<br />
Administrative Non-managers 9.06% 48.02% 11.09% 12.19% 2.28% 9.92% 19.64% 100.00%<br />
Incentive All employees 9.43% 7.53% 4.12% 39.94% 4.53% 35.41% 38.98% 100.00%<br />
intensity R&D Managers 2.03% 6.38% 8.60% 45.58% 7.34% 38.24% 37.41% 100.00%<br />
R&D Non-managers 12.56% 4.87% 3.63% 42.03% 4.45% 37.58% 36.91% 100.00%<br />
Technical Managers 4.76% 5.17% 3.72% 40.89% 4.01% 36.88% 45.46% 100.00%<br />
Technical Non-managers 8.54% 1.65% 6.21% 43.30% 4.10% 39.20% 40.<strong>29</strong>% 100.00%<br />
Administrative Managers 8.98% 8.36% 2.64% 38.66% 7.23% 31.44% 41.36% 100.00%<br />
Administrative Non-managers 11.71% 2.06% 3.73% 49.78% 5.82% 43.96% 32.73% 100.00%<br />
Long-term All employees 5.68% 7.65% 4.88% 33.87% 4.79% <strong>29</strong>.08% 47.91% 100.00%<br />
to short-term R&D Managers 5.19% 3.10% 2.91% 47.52% 4.79% 42.73% 41.<strong>29</strong>% 100.00%<br />
ratio R&D Non-managers 8.06% 2.15% 5.08% 40.58% 8.07% 32.51% 44.13% 100.00%<br />
Technical Managers 5.<strong>29</strong>% 2.53% 3.93% 40.94% 1.42% 39.52% 47.31% 100.00%<br />
Technical Non-managers 3.57% 1.22% 7.49% 34.22% 2.96% 31.26% 53.50% 100.00%<br />
Administrative Managers 5.67% 6.80% 3.95% 35.72% 2.28% 33.43% 47.86% 100.00%<br />
Administrative Non-managers 6.60% 3.21% 6.16% 38.88% 4.48% 34.40% 45.16% 100.00%<br />
N = 82,365 for overall, 5,452 for R&D managers, 36,322 for R&D non-managers, 3,634 for technical managers, 21,252 for technical non-managers, 4,094 for administrative<br />
managers, and 11,601 for administrative non-managers<br />
35
TABLE 3A TABLE 3B<br />
Conherence across Job Families Conherence <strong>of</strong> Pay Level Strategy<br />
Managerial Pay Level Non-managerial Pay Level<br />
Job family A R&D Technical Admin. Job family A R&D Technical Admin.<br />
Job family B Job family B<br />
R&D Fixed firm effects Base pay - 0.59*** 0.67*** R&D Fixed firm effects Base pay - 0.50*** 0.00<br />
Total pay - 0.81*** 0.79*** Total pay - 0.74*** 0.59***<br />
Idiosyncratic firm Base pay - 0.59*** 0.68*** Idiosyncratic firm Base pay - 0.50*** 0.10<br />
effects Total pay - 0.78*** 0.77*** effects Total pay - 0.74*** 0.64***<br />
Technical Fixed firm effects Base pay 0.70*** - 0.65*** Technical Fixed firm effects Base pay 0.66*** - 0.47***<br />
Total pay 0.68*** - 0.78*** Total pay 0.90*** - 0.80***<br />
0.50***<br />
Idiosyncratic firm Base pay 0.73*** - 0.64*** Idiosyncratic firm Base pay 0.68*** - 0.81***<br />
effects Total pay 0.71*** - 0.76*** effects Total pay 0.90*** - -<br />
Admin. Fixed firm effects Base pay 0.41** 0.34** - Admin. Fixed firm effects Base pay 0.33** 0.59*** -<br />
Total pay 0.85*** 0.59*** - Total pay 0.79*** 0.91*** -<br />
Idiosyncratic firm Base pay 0.42** 0.31* - Idiosyncratic firm Base pay 0.44*** 0.63*** -<br />
effects Total pay 0.84*** 0.54*** - effects Total pay 0.83*** 0.92*** -<br />
Values below the diagonal are based on the analysis on the 1997 data set and above are based on the analysis on the 1998 data set.<br />
N ranges from 54 to 64 in the 1997 data set and from 67 to 88 in the 1998 data set, depending on combinations.<br />
*** p < 0.001, ** p < 0.01, * < 0.05.<br />
TABLE 4A TABLE 4B<br />
Conherence across Job Families Conherence <strong>of</strong> Pay Level Strategy<br />
Managerial Pay Mix Non-managerial Pay Mix<br />
Job family A R&D Technical Admin. Job family A R&D Technical Admin.<br />
Job family B Job family B<br />
R&D Fixed firm effects Incentive intensity - 0.82*** 0.86*** R&D Fixed firm effects Incentive intensity - 0.87*** 0.82***<br />
Long-term to - 0.84*** 0.86*** Long-term to - 0.88*** 0.83***<br />
short-term ratio short-term ratio<br />
Idiosyncratic firm Incentive intensity - 0.81*** 0.86*** Idiosyncratic firm Incentive intensity - 0.86*** 0.82***<br />
effects Long-term to - 0.82*** 0.85*** effects Long-term to - 0.87*** 0.81***<br />
short-term ratio short-term ratio<br />
Technical Fixed firm effects Incentive intensity 0.69*** - 0.84*** Technical Fixed firm effects Incentive intensity 0.91*** - 0.94***<br />
Long-term to 0.80*** - 0.83*** Long-term to 0.93*** - 0.93***<br />
short-term ratio short-term ratio<br />
Idiosyncratic firm Incentive intensity 0.67*** - 0.84*** Idiosyncratic firm Incentive intensity 0.90*** - 0.94***<br />
effects Long-term to 0.79*** - 0.83*** effects Long-term to 0.92*** - 0.92***<br />
short-term ratio short-term ratio<br />
Admin. Fixed firm effects Incentive intensity 0.86*** 0.78*** - Admin. Fixed firm effects Incentive intensity 0.90*** 0.97*** -<br />
Long-term to 0.87*** 0.83*** - Long-term to 0.89*** 0.96*** -<br />
short-term ratio short-term ratio<br />
Idiosyncratic firm Incentive intensity 0.84*** 0.74*** - Idiosyncratic firm Incentive intensity 0.89*** 0.97*** -<br />
effects Long-term to 0.85*** 0.82*** - effects Long-term to 0.87*** 0.95*** -<br />
short-term ratio short-term ratio<br />
Values below the diagonal are based on the analysis on the 1997 data set and above are based on the analysis on the 1998 data set.<br />
N ranges from 54 to 64 in the 1997 data set and from 67 to 88 in the 1998 data set, depending on combinations.<br />
*** p < 0.001<br />
36
TABLE 5<br />
Conherence between Managers and Non-managers<br />
Pay level Pay mix<br />
Job family Base pay Total pay Incentive intensity Long-term to<br />
short-term ratio<br />
1997 1998 1997 1998 1997 1998 1997 1998<br />
R&D Fixed firm effects 0.55*** 0.66*** 0.73*** 0.80*** 0.80*** 0.87*** 0.83*** 0.89***<br />
Idiosyncratic firm 0.55*** 0.66*** 0.73*** 0.78*** 0.78*** 0.86*** 0.81*** 0.88***<br />
effects<br />
Technical Gross firm effects 0.38** 0.36** 0.65*** 0.67*** 0.77*** 0.79*** 0.75*** 0.75***<br />
Idiosyncratic firm 0.40** 0.37*** 0.63*** 0.64*** 0.73*** 0.78*** 0.71*** 0.75***<br />
effects<br />
Admin. Gross firm effects 0.41*** 0.48*** 0.66** 0.73*** 0.79*** 0.83*** 0.75*** 0.75***<br />
Idiosyncratic firm 0.40** 0.45*** 0.62*** 0.73*** 0.77*** 0.83*** 0.71*** 0.74***<br />
effects<br />
N ranges from 54 to 64 for the 1997 data set and 74-80 for the 1998 data set, depending on combinations.<br />
*** p < 0.001, ** p < 0.01.<br />
Discussion and Conclusion<br />
Consistent with prior compensation studies, this study confirmed firm specific effects on<br />
employee pay level and pay mix. After controlling for job characteristics (i.e., job level, job) and firm<br />
characteristics (i.e., size and performance), firm exhibited distinct differences in their employee pay level<br />
and mix. This study argues that the observed firm-specific effects reflected firm compensation strategies.<br />
This study further found that firm specific effects had a greater influence on the pay level for<br />
strategic employee groups, managers and R&D employees in our sample, than other employee groups.<br />
This supported this study’s one <strong>of</strong> key arguments that firm compensation strategies will influence<br />
compensation for strategic employee groups to a greater extent than they will influence compensation for<br />
other employee groups. In contrast, I failed to obtain the strong evidence to support the argument that<br />
firm compensation strategies influence strategic employees’ pay mix to a greater extent that they do other<br />
employees’ pay mix.<br />
Compensation research, as well as strategic human resource management research, has been<br />
ambiguous as to how firm compensation strategies influence different employee groups in different<br />
manners. My empirical analyses revealed that firms developed consistent pay level and pay mix strategies<br />
across employee groups. If a firm pays above the market pay rates for one employee group (e.g., R&D<br />
employees), the firm also pays above the market above pay rates for other employee groups (e.g.,<br />
technical employees, administrative employees). Likewise, if a firm emphasizes financial incentives for<br />
one employee group, the firm also emphasizes financial incentives for other employee groups. While my<br />
analyses were not sufficient to establish the causal direction, this study’s findings are consistent with my<br />
contention that firms determine the pay level strategies for strategic employee groups first, and then they<br />
mirror them to develop the pay level strategies for other employees. Although my analyses showed the<br />
strong correlations among pay mix strategies for different employee groups within firms, I obtained weak<br />
evidence to support my contention that firm pay mix strategies will influence strategic employee groups<br />
the most, and thus I am unable to interpret that firm pay mix strategies will influence strategic employee<br />
groups first and then they will apply similar pay mix strategies to other employee groups.<br />
37
In this manner, this study’s model was well supported for pay level, whereas it received mixed<br />
support for pay mix. This may suggest that the way firms develop and apply compensation strategies may<br />
be different between pay level and pay mix. Given the fact that firm effects are greater for pay mix (e.g.,<br />
23.09% for idiosyncratic firm effects on incentive intensity in 1997) than pay level (e.g., 7.66% for<br />
idiosyncratic firm effects on total pay in the same year), pay mix may be more strategic and thus warrant<br />
more firm discretion. As a result, firms may develop their firms’ overall pay mix strategies first, and<br />
apply the strategies to all employee groups consistently. In contrasts, pay level is traditionally determined<br />
by comparing internal structure with relevant market pay rates, and this comparison is typically conducted<br />
according to employee groups (Milkovich & Newman, <strong>2008</strong>). As a result, firm pay level strategies tend to<br />
be segmented according to job families. In fact, the correlation coefficients among firm pay mix strategies<br />
for different employee groups were generally higher than those among firm pay level strategies for<br />
different employee groups (Tables 4a, 4b, and 5). Admittedly, this is nothing but one possibility, and thus<br />
more careful empirical analyses will be necessarily to better understand the extent and nature <strong>of</strong> firm pay<br />
level strategies and pay mix strategies.<br />
In conclusion, this study makes the first step to understand how firm compensation strategies<br />
influence employee pay level and mix for different employee groups in different manners. While its<br />
analyses revealed some important phenomena, they also posed a new question, which is concerned with<br />
the possible differences in the nature and extent between pay level strategies and pay mix strategies. More<br />
empirical studies are needed to answer this question.<br />
References<br />
Arthur, J .B. (1998). Effects <strong>of</strong> human resource systems on manufacturing performance and turnover.<br />
Academy <strong>of</strong> Management Journal, 37: 670-687.<br />
Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (ed.) Advances in experimental social<br />
psychology. New York: Academic Press.<br />
Balkin D. B., & Gomez-Mejia, L. R. (1987). Toward a contingency theory <strong>of</strong> compensation strategy.<br />
Strategic Management Journal, 8: 169-182.<br />
Balkin D. B., & Gomez-Mejia, L. R. (1990). Matching compensation and organizational strategies.<br />
Strategic Management Journal, 11: 153-169.<br />
Batt, R. (2001). Explaining wage inequality in telecommunications services: customer segmentation,<br />
human resource practices, and union decline. Industrial and Labor Relations Review, 54: 425-449.<br />
Berkema, H. G., & Gomez-Mejia, L .R. (1998). Managerial compensation and firm performance: a<br />
general research framework. Academy <strong>of</strong> Management Journal, 41: 135-145.<br />
Boyd, B. K., & Salamin, A. (2001). Strategic reward systems: A contingency model <strong>of</strong> pay system design.<br />
Strategic Management Journal, 22: 777-792.<br />
Cappelli, Peter. (1999). Career jobs are dead. California Management Review, 42: 146-167.<br />
Chandler, A. D., Jr. (1962). Strategy and structure. Cambridge, MA: MIT Press.<br />
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis<br />
for the behavioral sciences (3 rd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.<br />
38
Collins, C. J., & Smith, K. G. (2006). Knowledge exchange and combinations: the role <strong>of</strong> human resource<br />
practices in the performance <strong>of</strong> high-technology firms. Academy <strong>of</strong> Management Journal, 49: 544-<br />
560.<br />
Datta, D. K., Guthrie, J. P., & Wright, P. M. (2005). Human resource management and labor productivity:<br />
does industry matter? Academy <strong>of</strong> Management Journal, 48: 135-145.<br />
Eisenhardt, K. M. (1989). Agency- and institutional-theory explanations: The case <strong>of</strong> retail sales<br />
compensation. Academy <strong>of</strong> Management Journal, 31: 488-511.<br />
Gerhart, B. (2000). Compensation strategy and organizational performance. In S. Rynes and B. Gerhart<br />
(Eds.), Compensation in organizations. San Francisco, CA: Jossey-Bass.<br />
Gehart, B., & Milkovich, G. T. (1990). Organizational differences in managerial compensation and<br />
financial performance. Academy <strong>of</strong> Management Journal, 33: 663-691.<br />
Gerhart, B., & Rynes, S. L. (2003). Compensation: Theory, evidence, and strategic implications.<br />
Thousand Oak, CA: Sage.<br />
Gomez-Mejia, L.R., & Balkin, D. B. (1992a). Determinants <strong>of</strong> faculty pay: An agency theory perspective.<br />
Academy <strong>of</strong> Management Journal, 35: 921-955.<br />
Gomez-Mejia, L.R., & Balkin, D. B. (1992b). Compensation, organizational strategy, and firm<br />
performance. South-Western: Cincinnati.<br />
Groshen, E. (1991). Sources <strong>of</strong> intra-industry wage dispersion: how much do employers matter?<br />
Quarterly Journal <strong>of</strong> Economics, 106: 869-885.<br />
Gerlach, K, & Stephan, S. (2006). Pay Policies <strong>of</strong> Firms and Collective Wage Contracts – An Uneasy<br />
Partnership? Industrial Relations, 45: 47-67.<br />
Kerr, J. (1985). Diversification strategies and managerial rewards: an empirical study. Academy <strong>of</strong><br />
Management Journal, 16: 155-179.<br />
Lepak, D. P., & Snell, S. A. (1999). The human resource architecture: Toward a theory <strong>of</strong> human capital<br />
allocation and development. Academy <strong>of</strong> Management Review, 24: 31-48.<br />
Milkovich, G. T. (1988). A strategic perspective on compensation management. In G. R. Ferris, & K. M.<br />
Rowland (Eds.).Research in Personnel and Human Resources Management, Vol. 6: 263-288.<br />
Greenwich CT: JAI Press.<br />
Milkovich, G. T., & Newman, J. (<strong>2008</strong>). Compensation (9 th ed.). New York: McGraw-Hill.<br />
National Center for Employee Ownership (2001). Current practices in stock option plan design. Oakland,<br />
CA.<br />
Osterman, P. (1987). Choice <strong>of</strong> employment systems in internal labor markets. Industrial Relations, 26:<br />
46-67.<br />
Yanadori, Y., & Marler, J. H. (2006). Compensation strategy: does business strategy influence<br />
compensation in high-technology firms? Strategic Management Journal, 27: 559-570.<br />
39
ASAC <strong>2008</strong> Janice Foley<br />
Halifax, Nova Scotia<br />
Faculty <strong>of</strong> Business Administration<br />
University <strong>of</strong> Regina<br />
CAN THE DIVERSITY LITERATURE HELP BUSINESS SCHOOLS<br />
ADDRESS THEIR RELEVANCE CRISIS?<br />
According to <strong>Volume</strong> 50, Issues 4 - 6 <strong>of</strong> the Academy <strong>of</strong><br />
Management Journal, some influential members <strong>of</strong> the Academy<br />
perceive that business schools are currently facing a severe crisis <strong>of</strong><br />
legitimacy in the eyes <strong>of</strong> their stakeholders. In this paper I propose<br />
that the diversity literature may be <strong>of</strong> some assistance in<br />
determining how to move forward.<br />
Introduction<br />
Some <strong>of</strong> the most prestigious <strong>of</strong> the Academy’s members weighed in on the Editors’<br />
Forums “Research With Relevance to Practice” (Issue 4), “The Research-Practice Gap in Human<br />
Resource Management” (Issue 5) and “AMJ Turns 50! Looking Back and Looking Ahead,” all<br />
suggesting that all is not well in business schools today. Legitimacy has long been a problem for<br />
business schools (Pfeffer and Wong, 2002) but what appears to have provoked the current rash <strong>of</strong><br />
commentary is the perception that business schools have failed in their mission to improve<br />
management practice through their research and education activities. As a consequence they<br />
have become increasingly vulnerable to encroachments into their “turf” by competitors from<br />
outside the Academy. Osborne and Cowen (1995: 28) commented over a decade ago that<br />
business schools were heading for a “train-wreck” and might eventually become “expensive<br />
engines pulling passengerless trains.” While this has not yet come to pass, some concern about<br />
the future <strong>of</strong> business schools may be warranted.<br />
The concerns expressed in the Editors’ Forums can be summarized as: business schools’<br />
failure to meet the expectations <strong>of</strong> their stakeholders, comprised <strong>of</strong> students, employers, and the<br />
broader society, with respect to delivering relevant management education and producing<br />
research insights that improve management practice; and a deficient academic system that is<br />
institutionalized and therefore resistant to change, that promotes a rigid view <strong>of</strong> what constitutes<br />
acceptable scholarship, thus contributing to teaching and research outputs that may not serve the<br />
public good.<br />
These problems are not likely to bring about the imminent demise <strong>of</strong> business schools.<br />
But commentators warn that it is likely that business schools are going to have to start proving<br />
their worth to an increasingly skeptical public if they are to continue to enjoy their current status<br />
(McGrath, 2007). That means they will have to hone their capacity to change as required to meet<br />
expectations (Osborne and Cowen, 1995; Trank and Rynes, 2003). The pressure for business<br />
schools to become learning organizations that welcome and encourage diversity is growing.<br />
40
Canadian business schools will not be exempt from these pressures, and a proactive approach to<br />
head <strong>of</strong>f emergent problems seems appropriate.<br />
Hence, this paper will summarize what has given rise to the concerns expressed, and the<br />
proposed solutions. Then insights from the diversity literature will be drawn upon to assess<br />
business schools’ current ability to change, and the obstacles that are likely to present themselves<br />
if they make the attempt.<br />
A Brief Description <strong>of</strong> the Perceived Problems facing today’s Business Schools<br />
As indicated, there are two categories <strong>of</strong> problems that are perceived to be contributing to<br />
the legitimacy crisis currently facing business schools. The first is the failure to meet stakeholder<br />
expectations.<br />
Failure to Meet Stakeholder Expectations<br />
According to the current and past contributors to the discussion on this topic, blame has<br />
been laid on business schools for first, failing to deliver a management curriculum to students<br />
that makes them ethical and effective managers, and further, for selling MBA degrees as a way<br />
to enhance personal career success rather than to acquire or enhance management skills (Pfeffer<br />
and Fong, 2002; 2004). Although the evidence does not support the myth <strong>of</strong> career-enhancement,<br />
the marketing ploy has been so successful that “learning” does not appear to motivate the typical<br />
MBA student (Pfeffer and Fong, 2002). Business schools have given in to student demands,<br />
making curricular changes that are indefensible on pedagogical grounds, such as dumbing down<br />
course content and raising average grades in the interests <strong>of</strong> improving school or individual<br />
pr<strong>of</strong>essor rankings (Zell, 2001). As a consequence they can be criticized for failing to protect the<br />
public welfare by churning out incompetent managers, and not acting in the long-term best<br />
interests <strong>of</strong> students.<br />
Other criticisms have also been levied on business school curricula (Mintzberg, 2004;<br />
Porter and McKibbin, 1988), for instance excess emphasis on analysis rather than “problemfinding,<br />
lack <strong>of</strong> integration across disciplinary lines, and inadequate attention to implementation<br />
or the development <strong>of</strong> “s<strong>of</strong>t” skills. The claim has also been made that “bad” theory is taught<br />
(Ghoshal, 2005) that reinforces a highly negative view <strong>of</strong> human behavior, encourages<br />
pathological management practice and discourages ethical management practice. Thus he and<br />
others (Mitr<strong>of</strong>f, 2004) contend that business schools have contributed to the corporate scandals<br />
that have recently come to light. Hence what is, or is not, being taught in business schools is<br />
having serious repercussions for students, for their employers, and ultimately for society.<br />
From the perspective <strong>of</strong> practicing managers, business school research is generally<br />
irrelevant with few, if any, really interesting management advances initiated by business school<br />
faculty (Cohen 2007). Rarely are research topics <strong>of</strong> great interest to practitioners (Pfeffer, 2007),<br />
but even where they are, the academic penchant for conducting “rigorous” scientific studies does<br />
not produce timely findings, or findings that resonate with practitioners’ experience. Nor do the<br />
academic journals effectively communicate findings to practitioners, who are reported to be<br />
more influenced by simpler, less rigourous studies conducted by non-academics (Cohen, 2007;<br />
41
Vermeulen, 2007) than they are by studies produced by more credible, academic researchers.<br />
They are also influenced by emotional appeals (Bartunek, 2007), which are absent from most<br />
academic work. Theories published that do bridge the academic-practitioner divide are<br />
essentially untested (Hambrick, 2007). While they do inform practice, the existence <strong>of</strong><br />
competing theories carrying contradictory prescriptions for action lessen their utility (Ghoshal,<br />
2005).<br />
Perhaps because <strong>of</strong> these difficulties, as well as business schools’ failure to articulate and<br />
defend the notion that a particular body <strong>of</strong> knowledge underpins the practice <strong>of</strong> management<br />
(Pfeffer and Fong, 2004; Trank and Rynes, 2003), practitioners have grown more inclined to see<br />
business education as a commodity that can be easily packaged for purchase or sale (Trank and<br />
Rynes, 2003). As schools have grown increasingly dependent on corporate funding for their<br />
operational needs, this idea may have contributed to the growing sense that business schools are<br />
not essential to the delivery <strong>of</strong> business training or to the improvement <strong>of</strong> management practice<br />
through research.<br />
The Failures <strong>of</strong> the Academic System<br />
The general accusation lodged against the academic system is that it discourages the<br />
search for relevance, which results in the failure to meet stakeholder expectations (Cornuel,<br />
2005). One contributing factor is the increasing funding shortfalls facing universities as a result<br />
<strong>of</strong> government cutbacks, which have brought about many changes which have had severe<br />
repercussions for business programs and faculty members. A second is the way doctoral students<br />
are selected, socialized and rewarded.<br />
The immediate result <strong>of</strong> the funding shortfalls has been that business schools have been<br />
forced to seek out external funding sources to cover operating expenses, and the most lucrative<br />
source has been business corporations. As a result <strong>of</strong> the infusion <strong>of</strong> funds from corporations,<br />
business schools have been exposed to increasing pressure to meet corporate demands for<br />
relevance, for commodification <strong>of</strong> educational programming, and for the development <strong>of</strong><br />
vocational rather than pr<strong>of</strong>essional skills in the classroom (Trank and Rynes, 2003). Schools<br />
have also expanded graduate programs to generate additional tuition revenues, which has<br />
diverted resources away from undergraduate and doctoral training (Trank and Rynes, 2003) and<br />
introduced students into graduate programs who arguably lack the academic preparation to be<br />
there. This has exerted downward pressure on academic standards, depr<strong>of</strong>essionalized course<br />
content and inflated grades (Zell, 2001), all in the interests <strong>of</strong> meeting student demands and<br />
influencing media , so as not to jeopardize enrollment numbers and tuition revenues. Student<br />
evaluations which are being taken seriously by Deans when evaluating academics’ performance,<br />
are reinforcing these trends.<br />
What is important to notice in all this is that business schools’ responses to these funding<br />
shortfalls have not been guided by any concern over the outcomes <strong>of</strong> the initiatives introduced<br />
(Pfeffer and Fong, 2002), for instance the caliber <strong>of</strong> the students that have been credentialed, or<br />
the social consequences <strong>of</strong> having poorly trained or unethical managers running business<br />
enterprises. Even AACSB accreditation does not ensure that a business school delivers quality<br />
education (McGrath, 2007; Trank and Rynes, 2003) because there is no consensus within the<br />
42
management academy about what “quality” means, or what constitutes the body <strong>of</strong> pr<strong>of</strong>essional<br />
management knowledge that students need to master prior to graduation. It is felt that this has<br />
put business schools even more at the mercy <strong>of</strong> market pressures (Trank and Rynes, 2003).<br />
A second failure <strong>of</strong> the academic system, as mentioned, relates to the way doctoral<br />
students are selected, socialized to the job, and rewarded. For example, doctoral students are<br />
rarely recruited from practitioner ranks, and may have little if any work experience. As a<br />
consequence, they may not have much understanding <strong>of</strong>, or interest in, the practical issues that<br />
managers face (Pfeffer and Fong, 2002). Their primary mandate, throughout their training, is to<br />
learn how to do research that will meet the rigorous standards set by the top management<br />
journals (Agarwal and Roetker, 2007; Ghoshal, 2005; Rynes, 2007a). Developing teaching skills<br />
is deemed <strong>of</strong> secondary importance.<br />
The research they produce is expected to conform to the North American model (Ghoshal,<br />
2005; Markides, 2007; Pfeffer, 2007) in order to be publishable in the top management journals.<br />
This model emphasizes the scientific method and a logical positivist research epistemology, and<br />
is grounded in the literature <strong>of</strong> a single discipline. Students are taught to avoid research that is<br />
not geared to theory building, that is multi-disciplinary, that covers topics that might be<br />
controversial, or that might have managerial relevance (Mitr<strong>of</strong>f, 2004; Vermeulen, 2007). The<br />
fact that qualitative methods produce more “interesting” findings (Rynes, 2007b), that this model<br />
may be unsuited to the development <strong>of</strong> social theory (Ghoshal, 2005), and that it may inhibit<br />
theory-building that is context-sensitive, which is deemed to be increasingly necessary in a<br />
global world (Tsui, 2007), has not lessened the model’s influence within the Academy.<br />
Vermeulen (2007) suggests that research irrelevance may be attributable to the belief held by the<br />
“silent majority” that relevance compromises scientific objectivity.<br />
Pursuing the other forms <strong>of</strong> scholarship identified by Boyer (1990), such as the<br />
integration or application <strong>of</strong> research findings, and their dissemination through teaching, bars<br />
academics from the “high table” (Ghoshal, 2005) as they have value only on the fringes <strong>of</strong> the<br />
Academy. So desires for status and publishing opportunities create tremendous pressures for<br />
students seeking jobs and faculty members seeking tenure, to adopt and conform to the North<br />
American model, which is the dominant model internationally, throughout the early stages <strong>of</strong><br />
their careers. If socialization is successful, by the time tenure is achieved this model has been<br />
deeply entrenched.<br />
The result is that academics currently place the ability to use mathematical models to<br />
produce testable propositions above the need to generate “truth” (Ghoshal, 2005: 81). Their<br />
“irrelevant” but scientifically rigorous research produces theories that are rarely, if ever, tested<br />
(Hambrick, 2007; Pfeffer, 2007). The same phenomena carry different labels and <strong>of</strong>fer different<br />
prescriptions for practice (Ghoshal, 2005). The importance <strong>of</strong> teaching is deemphasized – at least<br />
until instructors run afoul <strong>of</strong> their student “customers.” And a narrow view <strong>of</strong> scholarship is<br />
absorbed that makes it unacceptable to spend time in ways that would benefit practitioners.<br />
It is little wonder then that students are able to question the return on their investment in a<br />
business degree, and that practicing managers turn to non-academic sources to gain insight into<br />
how to address their problems.<br />
43
The Proposed Solutions to the Crisis<br />
Many <strong>of</strong> the solutions proposed to correct these problems were fairly straightforward. As<br />
lack <strong>of</strong> relevance in both teaching and research was largely attributed to the academic system and<br />
the behaviours it encourages, suggestions were made to change the way doctoral students were<br />
recruited, for instance by selecting them from the ranks <strong>of</strong> practicing manages (Ghoshal, 2005;<br />
Pfeffer and Fong, 2002). In terms <strong>of</strong> doctoral training, there were recommendations that the<br />
Academy should broaden its views <strong>of</strong> what contributes to knowledge (Ghoshal, 2005; March,<br />
2005), and what are acceptable forms <strong>of</strong> scholarship (Bartunek, 2007; Markides, 2007;<br />
Vermeulen, 2007) to include the development <strong>of</strong> practical understanding based on “disciplined<br />
imagination” rather than “formalized falsification” (Ghoshal, 2005). Reward systems should be<br />
revised to encourage “conceptual diversity” (March, 2005; Pfeffer and Fong, 2002). To increase<br />
relevance, impact on managerial practice could be included in the evaluation criteria for tenure<br />
decisions (McGrath, 2007).<br />
The research enterprise was singled out for major overhaul. It was seen as necessary for<br />
management researchers to recommit to the disinterested search for knowledge and an improved<br />
understanding <strong>of</strong> management phenomena (Pfeffer and Fong, 2004; Starkey et al., 2004).<br />
Research should focus on important problems (Pfeffer and Fong, 2002) and practitioners should<br />
be included in the research process, from the definition <strong>of</strong> research questions, through to the<br />
discussion <strong>of</strong> implications for practice (Bartunek, 2007; Ghoshal, 2005; Vermeulen, 2007).<br />
Theory-testing as well as theory-building needed to be embraced (Hambrick, 2007; March, 2004).<br />
Journals had to become more open to publishing “interesting facts” for which no explanatory<br />
theory was available but that might have potential consequence (Hambrick, 2007), and to<br />
publishing controversial finding and studies that found no significant effects (Pfeffer, 2007).<br />
Journal articles could feature expanded “Implications for Practice” sections.<br />
The value <strong>of</strong> studies conducted outside the traditions <strong>of</strong> the North American model<br />
needed to be recognized. That model had to become less restrictive, to allow deep<br />
contextualizing <strong>of</strong> findings (Tsui, 2007) and to increase the legitimacy <strong>of</strong> field research<br />
(McGrath, 2007) as well as encourage cross-disciplinary work (Mitr<strong>of</strong>f, 2004; Osborne and<br />
Cowen, 1995). Pluralism with respect to theory, methods, and research questions needed to be<br />
relegitimized (Ghoshal, 2005; Tsui, 2007). Novel ways <strong>of</strong> disseminating research findings, such<br />
as publishing in practitioner journals and books, writing textbooks and case studies, and<br />
participating in talk shows or making videos were proposed as ways to broaden researchers’<br />
impact on practitioners (Bartunek, 2007; McGrath, 2007).<br />
One <strong>of</strong> the suggestions advanced for changing the research enterprise also addressed the<br />
problems associated with management education. Klimoski (2007) suggested that the Academy<br />
should encourage research regarding the desired outcomes <strong>of</strong> management education, and what<br />
kinds and amounts <strong>of</strong> learning students needed to perform well in class and on the job. Many <strong>of</strong><br />
the contributors felt it was necessary to articulate the field <strong>of</strong> knowledge that defined<br />
pr<strong>of</strong>essional management (McGrath, 2007; Pfeffer and Fong, 2004; Trank and Rynes, 2003) in<br />
order to fulfill what some felt was the primary mission <strong>of</strong> business schools, the passing on <strong>of</strong><br />
such knowledge (McGrath, 2007; Pfeffer and Fong, 2004). Curricular reform could then be<br />
guided by an understanding <strong>of</strong> what is necessary to improve pr<strong>of</strong>essional practice (Pfeffer and<br />
44
Fong, 2002) rather than by the dictates <strong>of</strong> the market (Trank and Rynes, 2003). There was some<br />
agreement on the need for training in ethics and corporate social responsibility (Ghoshal, 2005;<br />
McGrath, 2007; Osborne and Cowen, 1995; Trank and Rynes, 2003; Zell, 2001), and Cornuel<br />
(2005) suggested that business education should also prepare students to become innovative<br />
leaders. The need to specify metrics to permit assessment <strong>of</strong> learning outcomes was noted<br />
(Klimoski, 2007; Pfeffer and Fong, 2002; Trank and Rynes, 2003).<br />
Some general suggestions were made for business schools. One was that they should<br />
clarify their core purpose (Pfeffer amd Fong, 2004) and stop participating in the oversimplification<br />
<strong>of</strong> business knowledge in the interests <strong>of</strong> pr<strong>of</strong>it (Trank and Rynes, 2003). They<br />
could start serving niche markets, to reduce competition between school (Zell, 2001). They were<br />
advised to model themselves on other pr<strong>of</strong>essional schools and return to their roots as university<br />
departments with educational and research responsibilities, as well as responsibilities to the<br />
public to exercise “due diligence” (Trank and Rynes, 2003) concerning the social consequences<br />
<strong>of</strong> management practice (Pfeffer and Fong, 2002; 2004) in order to differentiate themselves from<br />
their competitors. One blunt recommendation was that business schools had to “become learning<br />
organizations - or else” (Osborne and Cowen, 1995: 28).<br />
Several contributors directly advanced the notion that greater diversity within business<br />
schools was needed in order to serve all stakeholders better. Greenberg et al. (2007: 454)<br />
proposed that students would benefits from “diverse pr<strong>of</strong>essors who are guided by diverse<br />
interpretive frames <strong>of</strong> their roles,” while Bennis and O’Toole (2005) emphasized the need for<br />
skill variety and different perspectives. In regard to the improvement <strong>of</strong> research, March (2005:<br />
16) argued against the requirement to conform to a particular research paradigm, noting that<br />
“…it is what [scholars] do not share that makes [discourse] valuable.” Emphasizing that point,<br />
Agarwal and Hoetker (2007: 1318) observed that “…diversity <strong>of</strong> thought… <strong>of</strong> design… theories<br />
and methodologies introduced by scholars from related disciplines” were all valuable to the<br />
evolution <strong>of</strong> particular disciplines and the field <strong>of</strong> management. But they went on to<br />
acknowledge one <strong>of</strong> the major problems with implementing any <strong>of</strong> the recommendations<br />
proposed above, that “the extent to which the value is captured depends on the ability <strong>of</strong> our<br />
discipline to assimilate relevant perspectives, which itself relates to whether consensus exists on<br />
the value <strong>of</strong> a disciplined integration <strong>of</strong> multiple perspectives.”<br />
According to Lawler (2007) and others (Bartunek, 2007; Hambrick, 2007; Pfeffer, 2007),<br />
this consensus does not currently exist within the Academy and business schools have shown<br />
little sign <strong>of</strong> wanting to change their approach to research. Members <strong>of</strong> the “academic high table”<br />
(Ghoshal, 2005) have internalized the existing rules (Pfeffer, 2007) and are likely to be heavily<br />
represented on journal review boards and tenure committees, and within the senior faculty,<br />
where they help to reproduce the academic system by socializing new entrants into conformity<br />
with the North American research model. Any change in the status quo is likely to impose costs<br />
on the senior faculty (Markides, 2007), and the schools that are currently winning the ratings<br />
game have no incentive to change (Pfeffer and Fong, 2002).<br />
Other problems have been identified as well. The demand for relevance in research and<br />
teaching is a new burden for a pr<strong>of</strong>essoriate that is already over-extended (Pfeffer and Fong,<br />
2002). The resources needed to support a change in emphasis (Ghoshal, 2005) are unlikely to<br />
45
materialize when the funding shortfalls that gave voice to demands for relevance still exist, and<br />
external competition for business education dollars is increasing. There is a deep-seated bias<br />
against pursuing relevance as it is perceived to be antithetical to the advancement <strong>of</strong> scientific<br />
knowledge (Vermeulen, 2007). Opposing viewpoints exist regarding whether there is an<br />
excessive focus on theory-building (Hambrick, 2007), whether it would be desirable to<br />
incorporate multiple perspectives when building theory (Agarwal and Hoetker, 2007), whether<br />
adopting a pluralistic approach to scholarship would be acceptable, what constitutes<br />
“pr<strong>of</strong>essional” management education, and what responsibility the pr<strong>of</strong>essoriate has to prevent<br />
unscrupulous management practice (Trank and Rynes, 2003: 199).<br />
Osborne and Cowen (1995: <strong>29</strong>) articulate the need for businesses to change their<br />
emphases from bottom-line thinking and concern for shareholders to “the entire ecology <strong>of</strong><br />
human enterprise – energy, conservation, environment, and justice efficacies.” Business schools<br />
have a role to play in ensuring these outcomes materialize. The competencies embodied in the<br />
Academy, such as how to encourage organizational innovation (Pfeffer, 2007), how to establish<br />
educational standards, and how to create and evaluate “truth” (McGrath, 2007), constitute potent<br />
resources to draw upon in ensuring better outcomes materialize. But creativity is needed to find<br />
ways to reconcile the wide variety <strong>of</strong> opinions regarding what is desirable, in order to establish a<br />
shared vision for the future.<br />
Currently, according to Ghoshal (2005: 82), non-paradigm-following scholars have been<br />
“[eliminated] from our milieu or, at best, [accommodated] at the periphery.” March (2005: 20)<br />
believes that the future <strong>of</strong> the field lies with those “ambivalent scholars” at the periphery,<br />
possibly because they have less invested in the dominant research paradigm. Therefore,<br />
capitalizing on the diverse views that already exist within the pr<strong>of</strong>essoriate, recognizing and<br />
legitimizing the work <strong>of</strong> scholars who currently operate outside the existing research paradigm,<br />
and validating the generation <strong>of</strong> common sense-based knowledge and critical theories (March,<br />
2005) would appear to be helpful steps. Long-range thinking is being encouraged to ensure that<br />
the best interests <strong>of</strong> students, business and society are met (Ghoshal, 2005; McGrath, 2007;<br />
Pfeffer and Fong, 2002; 2004; Trank and Rynes, 2003), and that business schools enhance their<br />
perceived usefulness.<br />
Since the general diversity literature explores how the benefits <strong>of</strong> diversity can be<br />
captured and the potential costs avoided, and also identifies the contextual factors that influence<br />
outcomes, it may provide some insights regarding the likely outcomes <strong>of</strong> attempting to introduce<br />
any changes into the Academy and the types <strong>of</strong> roadblocks that might be encountered.<br />
Overview <strong>of</strong> Relevant Diversity Literature<br />
Many different aspects <strong>of</strong> diversity have been examined within the diversity literature.<br />
The type <strong>of</strong> diversity that is <strong>of</strong> concern in this section is diversity as it relates to the ideas,<br />
approaches and activities related to teaching and research that are found within business schools.<br />
This is fairly close to Thomas and Ely’s (1996: 80) definition <strong>of</strong> organizational diversity as “the<br />
varied perspectives and approaches to work that members <strong>of</strong> different identity groups bring.”<br />
46
Historically, the interest in diversity arose from the changes in the demographic make-up<br />
<strong>of</strong> the North American workforce, and the introduction <strong>of</strong> team-based work methods (Page,<br />
2007), the belief in the moral incorrectness <strong>of</strong> discriminating against minority group members<br />
(Von Bergen et al., 2005), the obligation to comply with affirmative action and employment<br />
equity laws, and the instrumental notion that, if an organization’s workforce reflected the<br />
demographic makeup <strong>of</strong> its current or future customer base, organizational pay-<strong>of</strong>fs would result.<br />
In none <strong>of</strong> these cases was diversity valued for its ability to capture the full range <strong>of</strong> employee<br />
capabilities, in order to benefit them, the organization, and society. Instead, individuals were<br />
expected to assimilate into the mainstream organization, adopting its values, culture, priorities,<br />
and the like (Dass and Parker, 1999), making their differences disappear.<br />
Then a qualitative study done by Thomas and Ely (2001: 265) produced the proposition<br />
that organizations could only tap into the true benefits <strong>of</strong> diversity if they viewed differences as<br />
learning opportunities and fully integrated diverse employee perspectives into how the<br />
organization’s “markets, products, strategies, missions, business practices, and even cultures”<br />
were defined. Although their study examined diversity with respect to race, they theorized that it<br />
was the differences in “life experiences, knowledge and insights” associated with racial<br />
differences that were <strong>of</strong> value. Klein and Harrison (2007) subsequently proposed that variety<br />
with respect to knowledge, ideas, values, attitudes and beliefs would confer positive benefits on<br />
organizations.<br />
The positive benefits <strong>of</strong> diversity have been identified as: greater innovation, better<br />
problem-solving and better predictions (Page, 2007); greater pr<strong>of</strong>itability, learning, creativity,<br />
flexibility, and individual and organizational growth (Thomas and Ely, 1996); internal resiliency<br />
in the face <strong>of</strong> change, greater meaningfulness <strong>of</strong> work and the advancement <strong>of</strong> social change<br />
(Svyantek and Bott, 2004); capitalizing on the full potential <strong>of</strong> employees to encourage<br />
organizational and national success (Oliver, 2005); greater efficiency, customer satisfaction,<br />
employee development and social responsibility (Dass and Parker, 1999); better decision making<br />
(Hartel, 2004); and greater recruitment success (Jayne and Dipboye, 2004). These benefits,<br />
however, have not always materialized during empirical tests and much <strong>of</strong> the literature raises<br />
doubts about the value <strong>of</strong> differences as they seem to be frequently accompanied by interpersonal<br />
and communication difficulties that impede organizational functioning. Some researchers have<br />
suggested that the process-oriented difficulties that arise within a diverse workforce might<br />
outweigh any associated gains (Jayne and Dipboye, 2004; Pitts and Jarry, 2007).<br />
Several studies have tried to reconcile the inconsistent findings, focusing on contextual<br />
factors that seem to affect outcomes. There is greater evidence supporting the performance<br />
effects <strong>of</strong> job-related rather than demographic differences (Svyantek and Bott, 2004). Task<br />
complexity and organizational strategy have been singled out, with the rationale that creativity<br />
and other benefits <strong>of</strong> diversity are more likely to materialize if the task is difficult and the<br />
company has adopted a growth strategy that makes creativity and innovation necessary (Bell and<br />
Berry, 2007; Dass and Parker, 1999). The number <strong>of</strong> diverse individuals and/or perspectives<br />
matters, as does whether or not power relations between the dominant and minority groups are<br />
sufficiently equalized to allow diverse perspectives to emerge (Klein and Harrison, 2007;<br />
Thomas and Ely, 2001). Thomas and Ely (1996; 2001) also found that the presence <strong>of</strong> common<br />
values across groups, including the valuing <strong>of</strong> diversity, is highly significant. The degree to<br />
47
which diversity is, in fact, integrated into the core activities <strong>of</strong> the organization (Dass and Parker,<br />
1999) and whether organizational socialization tactics encourage homogeneity or diversity<br />
(McMillan-Capehart, 2006), are also relevant to outcomes.<br />
The issues around shared values and ways <strong>of</strong> doing things, for example how new entrants<br />
are socialized, suggest that whether or not the organizational culture values diversity (Thomas<br />
and Ely, 1996; 2001) and encourages diversity openness (Hartel, 2004) is critical. Such a culture<br />
must specify effective means <strong>of</strong> reconciling differences in order to address the conflicts and<br />
misunderstandings to which differences can give rise (Jackson et al., 2003). Hartel (2004)<br />
noted that group processes tend to be more problematic in organizations that are less open to<br />
diversity. As Dass and Parker (1999) observed, conflict and debate, if badly handled, can be<br />
destructive to organizational functioning. Hence how diversity initiatives are implemented is<br />
important.<br />
This implicates the leadership in the success <strong>of</strong> any diversity initiative. Klein and<br />
Harrison (2007) observed that diversity that resulted in polarized opinions would not be helpful.<br />
Therefore, if leaders do not provide time for dialogue that allows hidden assumptions, values and<br />
beliefs to be revealed, and do not manage the revealed differences in such a way that different<br />
viewpoints can be reconciled to forge a sense <strong>of</strong> similarity that crosses cultural, functional and<br />
hierarchical lines, positive outcomes are unlikely to materialize (Dass and Parker, 1999; Thomas<br />
and Ely, 2001). In addition, lack <strong>of</strong> diversity within the top management team (Dass and Parker,<br />
1999) and lack <strong>of</strong> demonstrated commitment to diversity at that level, reflects and influences the<br />
prevailing organizational climate. Thomas and Ely (2001) noted that the leaders’ vision <strong>of</strong> the<br />
purpose <strong>of</strong> a diversified workforce is <strong>of</strong> fundamental importance. So, the optimum benefits <strong>of</strong><br />
diversity will never be realized if leaders feel diversity is only important because <strong>of</strong> employment<br />
laws, or because it allows niche markets to be better exploited, and refuse to allow diversity to<br />
influence core organizational functioning or to look favourably on differences.<br />
Can the Academy Become More Relevant?<br />
It might be argued that the Academy’s current insistence on homogenization is morally<br />
wrong, given that an increasing proportion <strong>of</strong> its members are internationally-trained, or<br />
internationally-based, that the North American model privileges “Anglophone experiences,<br />
researchers, and worldviews… [and] undermines the effectiveness <strong>of</strong> scholarly development and<br />
the fairness <strong>of</strong> scholarly competition (March, 2005: 7). That it provides little insight “into the<br />
nature <strong>of</strong> management in novel or emerging-economy contexts” (Tsui, 2007: 1356) might be<br />
considered morally objectionable, as well as detrimental to the advancement <strong>of</strong> management<br />
knowledge. However, the diversity literature suggests that justifying diversity on the basis <strong>of</strong><br />
moral arguments is not sufficient to reap its optimal benefits. Encouraging assimilation, which is<br />
what the current system accomplishes, at least at its core, is similarly ineffective. Valuing<br />
diversity for its capacity to release the full capabilities <strong>of</strong> employees or, in the case <strong>of</strong> the<br />
Academy, its members, with the inevitable positive impact this would have on management<br />
knowledge, would yield the best outcomes. But the Academy’s current interest in diversity arises<br />
from a perception <strong>of</strong> crisis, not from an acceptance <strong>of</strong> the value <strong>of</strong> diversity as such, which<br />
predicts sub-optimal outcomes.<br />
48
There are other indicators that diversity initiatives are unlikely to succeed. Relegitimizing<br />
pluralism into the Academy, as recommended by Ghoshal (2005) and Tsui (2007), will only be<br />
effective if, in the process, the core activities such as how the publication process works, how<br />
new entrants are socialized, and how the pr<strong>of</strong>essoriate is rewarded, are fundamentally altered.<br />
This, in turn, can only materialize with the support <strong>of</strong> the Academy’s leadership, which might be<br />
loosely construed to consist <strong>of</strong> administrators, senior faculty, tenure committee members, top<br />
journal reviewers, and “high table” members. For instance, practitioner-academic partnerships<br />
will not materialize unless the reward system is changed. Expanding the “Implications for<br />
Practice” sections in journal articles must be authorized by journal editors. Reassuming<br />
responsibility for studying the negative as well as positive impacts <strong>of</strong> business practice for<br />
groups other than shareholders might produce outputs that would be eyed with suspicion by<br />
tenure committees, and might also be discouraged by Deans if it threatened the flow <strong>of</strong> corporate<br />
funds. Deans might also have a great deal to say about the nature and extent <strong>of</strong> curricular and<br />
program reform that would be tolerated, for the same reason.<br />
Introducing more fundamental changes, like converting business schools into learning<br />
organizations as Osborne and Cowen (1995) proposed, would require huge amounts <strong>of</strong> time and<br />
effort, and would have to be implemented and facilitated by the formal leadership. Hence the<br />
Deans <strong>of</strong> the business schools would have to be sold on the idea and willing to provide the<br />
resources and support necessary. They would probably face resistance from faculty members<br />
who are rewarded for their individual research outputs rather than for their citizenship<br />
behaviours, particularly when they have benefited from the current situation and might not see<br />
any immediate need for change.<br />
Even if such an initiative did get <strong>of</strong>f the ground, the skewed power relations within the<br />
Academy, with students at the bottom, internationally-respected scholars at the top, and<br />
peripheral scholars somewhere in the middle, would make it difficult to elicit and then integrate<br />
the full range <strong>of</strong> opinions and suggestions present, which Senge (2006) identifies as critical to the<br />
creation <strong>of</strong> learning organizations. In addition, it seems reasonable to depict the homogenizing<br />
culture <strong>of</strong> the Academy as diversity-closed, and in such cultures, group process difficulties are<br />
more likely to arise (Hartel, 2004). Because there is no tradition, and no individual or institution<br />
held responsible for reconciling differing views and expectations within the Academy, the skills<br />
necessary to resolve conflict may be lacking. Given all these circumstances, establishing a<br />
common belief in the value <strong>of</strong> diversity would not come easily.<br />
However, the literature also suggests that encouraging greater diversity might be a fitting<br />
solution to the problems perceived to be bedeviling the Academy. First, the type <strong>of</strong> diversity<br />
being encouraged, <strong>of</strong> ideas, approaches and activities related to teaching and research, is the jobrelated<br />
diversity that the diversity literature has found to be most strongly associated with<br />
improved performance outcomes. Second, modifying how core activities are performed within<br />
the Academy can be considered a complex task requiring innovative solutions, which is the kind<br />
<strong>of</strong> task most likely to benefit from diverse inputs. Third, creativity is going to be necessary if<br />
some kind <strong>of</strong> consensus on how to move forward is to be forged across quite significant status,<br />
disciplinary and values divides, which again is most likely to be forthcoming when the combined<br />
talents <strong>of</strong> a highly intelligent, diverse community are brought to bear on the problem.<br />
49
It is difficult to imagine what might act as a catalyst to change the current situation.<br />
Lawler (2007) suggested that the AACSP’s draft report on the Academy’s research impact might<br />
stimulate change. A decline in MBA enrollments as prospective students come to realize the<br />
degree no longer is associated with career success, or general enrollment declines and loss <strong>of</strong><br />
tuition dollars as competitors make further inroads into the field <strong>of</strong> management education might<br />
serve as wake-up calls. Finally, the emergence <strong>of</strong> leaders within the Academy who see the need<br />
and have both the inclination and the ability to forge a consensus on issues and potential<br />
solutions, would be a hopeful development.<br />
Conclusion<br />
If it is true that the future <strong>of</strong> business schools will be jeopardized by continued inaction<br />
on the relevance front, and if the diversity literature is correct, we can conclude that a culture that<br />
values diversity and finds ways to make it work will produce the best alternatives, predictions,<br />
problem solutions and outcomes for the Academy. Although it has been argued that the<br />
Academy’s current diversity-closed culture is not particularly helpful, it should be noted that<br />
some substantial diversity already exists at the periphery <strong>of</strong> the North American research model,<br />
and within the international management research community where different research traditions<br />
prevail. This may be adaptive for the research community overall. Those scholars currently<br />
working outside the dominant paradigm have adapted to their marginalized status within the<br />
Academy, finding alternative research communities and funding sources, and different places to<br />
publish and otherwise disseminate their work. Some <strong>of</strong> them may even have chosen to be on the<br />
periphery so that they could do relevant research and teaching.<br />
So, to end this paper on a positive note, it is possible that the looming crisis the Academy<br />
fears at worst may be overstated, and at best, may never materialize.<br />
50
References<br />
Agarwal, R. & Hoetker, G. (2007). A Faustian bargain? The growth <strong>of</strong> management and its<br />
relationship with related disciplines. Academy <strong>of</strong> Management Journal, 50(6), 1304-1322.<br />
Bartunek, J. (2007). Academic-practitioner collaboration need not require joint or relevant<br />
research: Toward a relational scholarship <strong>of</strong> integration. Academy <strong>of</strong> Management Journal,<br />
50(6), 1323-1333.<br />
Bell, M. & Berry, D. (2007). Viewing diversity through different lenses: Avoiding a few blind<br />
spots. Academy <strong>of</strong> Management Perspectives, 21(4), 21-25.<br />
Bennis, W. & O’Toole, J. (2005). How business schools lost their way. Harvard Business<br />
Review, 98-104.<br />
Boyer, E. (1990). Scholarship reconsidered: Priorities <strong>of</strong> the pr<strong>of</strong>essoriate. Princeton, NJ:<br />
Carnegie Foundation for the Advancement <strong>of</strong> Teaching.<br />
Cohen, D. (2007). The very separate worlds <strong>of</strong> academic and practitioner periodicals in Human<br />
Resource Management. Academy <strong>of</strong> Management Journal, 50(5), 1013-1019.<br />
Cornuel, E. (2005). The role <strong>of</strong> business schools in society. Journal <strong>of</strong> Management<br />
Development, 24(9), 819-8<strong>29</strong>.<br />
Dass, P. & Parker, B. (1999). Strategies for managing human resource diversity: From<br />
resistance to learning. Academy <strong>of</strong> Management Executive, 13(2), 68-80.<br />
Ghoshal, S. (2005). Bad management theories are destroying good management practices.<br />
Academy <strong>of</strong> Management Learning and Education, 4(1), 75-91.<br />
Greenberg, D., Clair, J., & MacLean, T. (2007). Enacting the role <strong>of</strong> management pr<strong>of</strong>essor:<br />
Lessons from Athena, Prometheus, and Asclepius. Academy <strong>of</strong> Management Learning and<br />
Education, 6(4), 439-457.<br />
Hambrick, D. (2007). The field <strong>of</strong> management’s devotion to theory: Too much <strong>of</strong> a good thing?<br />
Academy <strong>of</strong> Management Journal. 50(6), 1346-1352.<br />
Hartel, C. (2004). Towards a multicultural world: Identifying work systems, practices and<br />
employee attitudes that embrace diversity. Australian Journal <strong>of</strong> Management, <strong>29</strong>(2), 189-<br />
200.<br />
Jackson, S., Joshi, A. & Erhardt, N. (2003). Recent research on team and organizational<br />
diversity: SWOT analysis and implications. Journal <strong>of</strong> Management, <strong>29</strong>(6), 801-830.<br />
Jayne, M. & Dipboye, R. (2004). Leveraging diversity to improve business performance:<br />
Research findings and recommendations for organizations. Human Resource Management,<br />
43(4), 409-424.<br />
51
Klein, K. & Harrison, D. (2007). On the diversity <strong>of</strong> diversity: Tidy logic, messier realities.<br />
Academy <strong>of</strong> Management Perspectives, 21(4), 26-33.<br />
Klimoski, R. (2007). From the editor: Becoming a prophet in our own land. Academy <strong>of</strong><br />
Management Learning and Education, 6(4), 433-438.<br />
Lawler, E. (2007). Why HR practices are not evidence-based. Academy <strong>of</strong> Management<br />
Journal, 50(5) 1033-1036.<br />
March, J. (2005). Parochialism in the evolution <strong>of</strong> a research community: The case <strong>of</strong><br />
organization studies. Management and Organization Review, 1(1), 5-22.<br />
Markides, C. (2007). In search <strong>of</strong> ambidextrous pr<strong>of</strong>essors. Academy <strong>of</strong> Management Journal,<br />
50(4), 762-768.<br />
McGrath, R. (2007). No longer a stepchild: How the management field can come into its own.<br />
Academy <strong>of</strong> Management Journal, 50(6), 1365-1378.<br />
McMillan-Capehart, A. (2006). Heterogeneity <strong>of</strong> homogeneity. Socialization makes the<br />
difference in firm performance. Performance Improvement Quarterly, 19(1), 83-98.<br />
Mintzberg, H. (2004). Managers not MBAs: A hard look at the s<strong>of</strong>t practice <strong>of</strong> managing. San<br />
Francisco: Berrett-Koehler.<br />
Mitr<strong>of</strong>f, I. (2004). An open letter to the Deans and the faculties <strong>of</strong> American business schools.<br />
Journal <strong>of</strong> Business Ethics, 54, 185-189.<br />
Oliver, D. (2005). Achieving results through diversity: A strategy for success. Ivey Business<br />
Journal, March/April, 1-6.<br />
Osborne, R. & Cowen, S. (1995). Business schools must become learning organizations – or<br />
else. Learning Organization, 2(2), 28-31.<br />
Page, S. (2007). Making the difference: Applying a logic <strong>of</strong> diversity. Academy <strong>of</strong><br />
Management Perspectives, 21(4), 6-20.<br />
Pfeffer, J. (2007). A modest proposal: How we might change the process and product <strong>of</strong><br />
managerial research. Academy <strong>of</strong> Management Journal, 50(6), 1334-1345.<br />
Pfeffer. J. & Fong, C. (2002). The end <strong>of</strong> business schools? Less success than meets the eye.<br />
Academy <strong>of</strong> Management Learning and Education, 1(1), 78-95.<br />
Pfeffer, J. & Fong, C. (2004). The business school ‘business’: Some lessons from the US<br />
experience. Journal <strong>of</strong> Management Studies, 41(8), 1501-1520.<br />
52
Pitts, D. & Jarry, E. (2007). Ethnic diversity and organizational performance: Assessing<br />
diversity effects at the managerial and street levels. International Public Management<br />
Journal, 10(2), 233-254.<br />
Porter, L. &.McKibbin, L. (1988). Management education and development: Drift or thrust<br />
into the 21st century. New York: McGraw-Hill.<br />
Rynes, S. (2007a). AMJ turns 50! Looking back and looking forward. Academy <strong>of</strong><br />
Management Journal, 50(6), 1277-1279.<br />
Rynes, S. (2007b). Time flies when you’re having fun: AMJ, 2005-2007. Academy <strong>of</strong><br />
Management Journal, 50(6), 1273-1276.<br />
Senge, P. (2006). The fifth discipline. The art & practice <strong>of</strong> the learning organization. New<br />
York: Doubleday.<br />
Starkey, K., Hatchuel, A. & Tempest, S. (2004). Rethinking the business school. Journal <strong>of</strong><br />
Management Studies, 41(8), 1521-1531.<br />
Svyantek, D. & Bott, J. (2004). Received wisdom and the relationship between diversity and<br />
organizational performance. Organizational Analysis, 12(3), <strong>29</strong>5-317.<br />
Thomas, D.& Ely, R. (1996). Making differences matter: A new paradigm for managing<br />
diversity. Harvard Business Review, September/October, 79-90.<br />
Thomas, D. & Ely, R. (2001). Cultural diversity at work: The effects <strong>of</strong> diversity perspectives<br />
on work group processes and outcomes. Administrative Science Quarterly, 46(2), 2<strong>29</strong>-273.<br />
Trank, C. & Rynes, S. (2003). Who moved our cheese: Reclaiming pr<strong>of</strong>essionalism in business<br />
education. Academy <strong>of</strong> Management Learning and Education, 2(2), 189-205.<br />
Tsui, A. (2007). From homogenization to pluralism: International management research in the<br />
Academy and beyond. Academy <strong>of</strong> Management Journal, 50(6), 1353-1364.<br />
Vermeulen, F. (2007). “I will not remain insignificant”: Adding a second loop to matter more.<br />
Academy <strong>of</strong> Management Journal, 50(4), 754-761.<br />
Von Bergen, C., Soper, B., Parnell, J. (2005). Workforce diversity and organizational<br />
performance. Equal Opportunities International. 24(3/4), 1-16.<br />
Zell, D. (2001), The market-driven business school. Has the pendulum swung too far? Journal<br />
<strong>of</strong> Management Inquiry, 10(4), 324-338.<br />
53
ASAC <strong>2008</strong> Christopher R Perryer<br />
Halifax, Nova Scotia Steven L McShane<br />
UWA Business School<br />
The University <strong>of</strong> Western Australia<br />
THE INFLUENCE OF TRAINING TRANSFER CLIMATE AND<br />
INDIVIDUAL TRAINEE CHARACTERISTICS ON CUSTOMER<br />
ORIENTATION<br />
This study tests hypotheses regarding the factor structure <strong>of</strong> a<br />
prominent training transfer climate instrument, the issue <strong>of</strong><br />
moderating rather than just direct effects <strong>of</strong> training transfer climate,<br />
and the effects <strong>of</strong> two trainee characteristics -- organizational<br />
commitment and cynicism about organizational change -- on the<br />
transfer <strong>of</strong> training.<br />
Increasingly, scholars and practitioners alike recognize that human capital is a critical<br />
source <strong>of</strong> organizational competitive advantage (Appelbaum, Bailey, Berg, & Kalleberg,<br />
2000; Benson, Young, & Lawler III, 2006; Pfeffer, 1998). Furthermore, developing<br />
employee competencies is considered one <strong>of</strong> the core components <strong>of</strong> high performance work<br />
practices that, when combined with other such practices, can leverage the value <strong>of</strong> human<br />
capital (Becker & Huselid, 2006; Combs, Liu, Hall, & Ketchen, 2006; Lawler III, Mohrman,<br />
& Ledford Jr., 1998). Training is one <strong>of</strong> the primary tools for developing employee<br />
competencies, yet much <strong>of</strong> the value <strong>of</strong> this training isn’t transferred to the job or deteriorates<br />
within a short time after being transferred. For example, 150 Canadian human resource<br />
executives recently estimated that less than two-thirds <strong>of</strong> employees use their training back<br />
on the job, and this diminishes to 44% and 34% after six months and one year, respectively<br />
(Saks & Belcourt, 2006). Others provide more dismal figures, estimating that as little as 10-<br />
15% <strong>of</strong> what is learned in formal training programs is transferred or remains in use one year<br />
later (Cromwell & Kolb, 2004; Newstrom, 1986).<br />
Given the substantial investment in workplace training -- approximately $850 per<br />
employee in Canada and much higher in the United States (Grant & Hughes, 2007) -- it is<br />
little wonder that scholars have been eager to identify factors that facilitate or hinder transfer<br />
<strong>of</strong> training (Burke & Hutchins, 2007). Indeed, some scholars identify transfer <strong>of</strong> training as<br />
the critical point through which training influences organizational effectiveness (Kozlowski,<br />
Brown, Weissbein, Cannon-Bowers, & Salas, 2000). Baldwin and Ford (1988) laid the<br />
foundation for current thinking about training transfer predictors, and many others have<br />
either built models or empirically tested some <strong>of</strong> these factors. However, much <strong>of</strong> that<br />
research has concentrated on trainee characteristics, such as trainee ability and motivation, or<br />
training design issues, such as the relevance <strong>of</strong> program content to trainees or instructional<br />
techniques. Only recently have scholars taken a more integrated approach to the subject by<br />
examining the unique and combined effects <strong>of</strong> trainee characteristics and workplace factors<br />
on training transfer (Lim & Morris, 2006; Tziner, Fisher, Senior, & Weisberg, 2007).<br />
54
This paper hopes to shed further light on the transfer <strong>of</strong> training in three ways. First, we<br />
examine the construct validity (specifically the factor structure) <strong>of</strong> a training transfer climate<br />
instrument that has gained prominence over the past decade. Second, we test the moderator<br />
effects, not just the direct effects, <strong>of</strong> training transfer climate on the training outcome variable,<br />
which in this study is employee customer orientation. Third, together with the training<br />
transfer climate dimensions, this study examines the predictive effects <strong>of</strong> two trainee<br />
characteristics, organizational commitment and cynicism <strong>of</strong> organizational change. The<br />
significance and rationale for these research objectives are explained over the next few pages.<br />
Training Transfer Climate<br />
Several training transfer studies employ the term “training transfer climate” to describe<br />
work environment conditions that influence the generalization and maintenance <strong>of</strong><br />
knowledge and skills learned during training (E. F. III Holton, Bates, Seyler, & Carvalho,<br />
1997; Machin & Fogarty, 2004; Tracey, Tannenbaum, & Kavanagh, 1995; Washington,<br />
2002). Significant relationships have been reported between training transfer climate and the<br />
degree <strong>of</strong> learning transferred from the training program to the workplace, but these findings<br />
are sometimes inconsistent. One concern recently identified by Burke and Hutchins (2007) is<br />
that “measures <strong>of</strong> transfer climate are in a state <strong>of</strong> transition.” We take a more critical<br />
position: the problem <strong>of</strong> conceptualizing and measuring training transfer climate has existed<br />
from its earliest days. Holton et al (1997) similarly state that, notwithstanding general<br />
agreement regarding the importance <strong>of</strong> training transfer climate, there is no clear, shared<br />
understanding <strong>of</strong> the transfer climate construct.<br />
The confusion in defining and measuring training transfer climate is apparent while<br />
sifting through the more than 60 published studies on the subject. A dozen <strong>of</strong> these studies<br />
operationalized training transfer climate as a single composite measure without conducting<br />
any construct validation to support a single-factor approach. Almost four-dozen studies took<br />
a multi-dimensional approach to training transfer climate by examining several variables or<br />
dimensions that the researchers claim are within the construct definition. However, most <strong>of</strong><br />
these studies did not attempt to identify and measure the construct’s full conceptual domain<br />
and, even where so attempted, did not rely on a theoretical framework to justify inclusion <strong>of</strong><br />
these dimensions <strong>of</strong> training transfer climate (Detailed tabulations <strong>of</strong> these studies are<br />
available from the authors on request.).<br />
A small number <strong>of</strong> publications have studied training transfer climate with a theoretical<br />
foundation, typically adopting the behaviourist and/or social learning model. The most<br />
widely cited <strong>of</strong> these is Rouiller & Goldstein (1993), who suggest that transfer climate<br />
consists <strong>of</strong> situational and consequence cues. Their work was further developed by Thayer<br />
and Teachout (1995), who adopted and refined Rouiller and Goldstein’s measure <strong>of</strong> these<br />
cues. Most recently, Machin and Fogarty (2004) further refined the training transfer climate<br />
measure developed and revised in the earlier two studies. One <strong>of</strong> our objectives is to test the<br />
reliability and factor structure <strong>of</strong> this emerging instrument, thereby hopefully moving closer<br />
to establishing a conceptual and operational foundation for training transfer climate.<br />
Hypothesis 1: Training transfer climate consists <strong>of</strong> six factors, goal cues, social<br />
cues, task cues, positive reinforcement, negative reinforcement and punishment,<br />
and extinction.<br />
55
Hypothesis 2: Employee perceptions <strong>of</strong> the training transfer climate are positively<br />
related to employee customer orientation.<br />
Moderator Effects <strong>of</strong> Training Transfer Variables<br />
Another issue investigated by this study is the direct versus moderator effects <strong>of</strong> training<br />
transfer climate variables on training outcomes. Although Baldwin and Ford (1988)<br />
encouraged further research on the interaction effects <strong>of</strong> training transfer variables, only a<br />
handful <strong>of</strong> studies have done so (Burke & Baldwin, 1999; Mathieu, Tannenbaum, & Salas,<br />
1992; Richman-Hirsch, 2001; Smith-Jentsch, Salas, & Brannick, 2001). This is surprising<br />
because variables that facilitate or hinder training transfer are, by definition, moderating the<br />
relationship between the employee’s attendance and learning in the training program and<br />
training outcomes such as on-the-job attitudes and behaviour. Moderator variables affect the<br />
strength or direction <strong>of</strong> the relationship between predictor and criterion variables (Baron &<br />
Kenny, 1986; Mason, Tu, & Cauce, 1996; Sharma, Durand, & Gur-Arie, 1981). Training<br />
transfer variables are moderators because they purportedly affect the strength <strong>of</strong> the<br />
relationship between training program attendance or learning and on-the-job attitudes or<br />
behaviour.<br />
Consider the following example: Suppose that a training program taught employees the<br />
procedures for driving forklifts more safely, such as looking, signaling, and visually checking<br />
the equipment before driving. Compared to typical forklift driving, these safe-driving<br />
behaviours slightly increase the time required to complete tasks. Let’s further suppose that<br />
employees with light or moderate workloads apply the learned forklift safety behaviours<br />
more <strong>of</strong>ten than do employees with heavy workloads. In this example, workload moderates<br />
the relationship between forklift safety training attendance or learning and on the job<br />
behaviour. The reason for this moderator is that the predictive effect <strong>of</strong> the training program<br />
varies with (i.e. depends or is contingent upon) the level <strong>of</strong> workload. A high correlation<br />
would be observed between training program attendance or learning and on-the-job<br />
behaviour when workload is low or medium – employees who attended the training program<br />
or learned more in that program engage in more forklift safety behaviours than those who did<br />
not attend or learned very little in the program. The correlation decreases as workload<br />
increases – the incidence <strong>of</strong> forklift safety behaviour varies minimally by attendance or<br />
learning in the training program when employees experience high workloads. Moderator<br />
variables <strong>of</strong>ten have a direct association with the dependent variable, but they also have<br />
additional explained variance as an interaction term with the predictor variable (training<br />
attendance or learning in our example).<br />
As mentioned, a few studies have examined the moderator effects <strong>of</strong> some training<br />
transfer variables. At least two studies found no moderator effect (Rouiller & Goldstein,<br />
1993; Tracey et al., 1995). Others have reported a moderator effect, but many <strong>of</strong> these have<br />
limitations. At least one study claimed to investigate the moderator effect, but did not<br />
conduct any moderated analysis (Kontoghiorghes, 2001). Burke and Baldwin (1999) reported<br />
a significant moderator effect size, but relied on a small sample and their dependent variable<br />
was the trainees’ use <strong>of</strong> transfer strategies, rather than transfer <strong>of</strong> the training content itself.<br />
Richman-Hirsch’s (2001) study <strong>of</strong> goal-setting training found a significant interaction<br />
between a supportive work environment and post-training behaviour. However, goal setting<br />
56
is also a training transfer variable, which may have confounded the results. Finally, none <strong>of</strong><br />
these or other training transfer studies conducted a preliminary test to see whether the<br />
significant moderator effect is spurious due to curvilinear relationships between predictor and<br />
dependent variables (Lubinski & Humphreys, 1990). Consequently, the present study<br />
hypothesizes that the effect <strong>of</strong> training program attendance on the dependent variable will be<br />
moderated by training transfer climate variables. This hypothesis will include a test for<br />
quadratic (curvilinear) effects:<br />
Hypothesis 3: Training transfer climate variables will moderate the relationship<br />
between participation in the training intervention and employee customer<br />
orientation.<br />
Trainee Characteristics and Transfer <strong>of</strong> Learning<br />
Notwithstanding the importance <strong>of</strong> training transfer climate, trainee characteristics also<br />
have an influence on the degree to which training concepts and practices are transferred to<br />
the workplace (Burke & Hutchins, 2007; Lim & Morris, 2006; Tziner et al., 2007).<br />
Personality, motivation, ability, and self-efficacy have been widely studied as predictors <strong>of</strong><br />
training transfer. There are, however, other individual trainee characteristics that have<br />
received far less attention from researchers. One <strong>of</strong> these, organizational commitment, is the<br />
relative strength <strong>of</strong> an individual’s identification with, and involvement in an organization<br />
(Mowday, Steers, & Porter, 1979). Meyer and Allen (1997) maintain that organizational<br />
commitment has a positive influence on the training process. Some research supports this<br />
assertion (Naquin & Holton III, 2002; Tracey, Hinkin, Tannenbaum, & Mathieu, 2001),<br />
whereas other studies report that organizational commitment has a nonsignificant influence<br />
on training transfer (Facteau, Dobbins, Russell, Ladd, & Kudisch, 1995). The lack <strong>of</strong> clarity<br />
surrounding the effect <strong>of</strong> this variable warrants further investigation.<br />
This study also introduces cynicism about organizational change as an individual<br />
characteristic that we hypothesize will have an inverse effect on training transfer. Cynicism<br />
about organizational change is defined as “a pessimistic outlook for successful change, and<br />
blame placed on ‘those responsible’ for lacking the motivation and/or the ability to effect<br />
successful change” (Reichers, Wanous, & Austin, 1997). We believe that this construct has<br />
particular relevance to training transfer, because training is an important strategy to support<br />
the change process. To date no studies have attempted to examine the relationship between<br />
training transfer climate and cynicism about organizational change. It is, however, a variable<br />
that shows great explanatory promise, and consequently warrants further study, leading to the<br />
fourth research question.<br />
Hypothesis 4: Organizational commitment is positively related to employee<br />
customer orientation.<br />
Hypothesis 5: Cynicism about organizational change is negatively related to<br />
employee customer orientation.<br />
57
Research Setting<br />
Methods<br />
The research setting for this study is Centrelink, which was formed by the Australian<br />
government in the late 1990s to centralize a range <strong>of</strong> social services and payments previously<br />
distributed by dozens <strong>of</strong> government departments and agencies. This innovative “one-stop<br />
shop” currently provides services to approximately 6.5 million customers each year, which<br />
includes more than 6 billion customer record transactions and distribution <strong>of</strong> AUD $66<br />
billion on behalf <strong>of</strong> the 31 government policy departments (Centrelink, 2007; Halligan, 2007).<br />
Centrelink’s leaders have focused on customer service as a key driver <strong>of</strong> organizational<br />
change. One intervention introduced to support this cultural transformation is the Value<br />
Creation Workshop (VCW) (Vardon, 1998). VCWs are one-day workshops conducted by<br />
trained facilitators held away from Centrelink <strong>of</strong>fices with up to 20 customer and 20<br />
employee participants. During the morning session, employees listen (without speaking) to<br />
customers describing their experiences and expectations from Centrelink. The morning<br />
session also involves technology that allows employee participants to predict how customers<br />
will respond to issues, and to assess the accuracy <strong>of</strong> their predictions. Customers and<br />
employees mingle during lunch, after which customers leave. In the afternoon session,<br />
employees build on the customer feedback to develop strategies for customer service<br />
improvement (Australian National Audit Office, 2005; Bennington & Cummane, 1998).<br />
To date there have been no empirical studies <strong>of</strong> the Value Creation Workshop, even<br />
though it has been applied to a few organizations. One published paper (Bennington &<br />
Cummane, 1998) details the methodology <strong>of</strong> the intervention, and suggests that it is a useful<br />
market research tool that has the added benefit <strong>of</strong> improving service quality. However, in the<br />
absence <strong>of</strong> any rigorous research, the usefulness <strong>of</strong> the intervention is largely based on the<br />
claims <strong>of</strong> the promoters <strong>of</strong> the program (Bennington, 1999). While this study relies on selfreports<br />
<strong>of</strong> the degree to which employees apply the training, it will add to the limited existing<br />
knowledge about the intervention.<br />
Sample and Data Collection<br />
Data were collected from Centrelink customer service employees in two Australian states.<br />
The population consists <strong>of</strong> a mixture <strong>of</strong> people who have been exposed to the training<br />
intervention and those who have not. Data were initially collected through an intranet survey,<br />
with invitations and web site link delivered by email to employees with primarily customer<br />
service duties. Due to a low response rate through the online survey, hard copies <strong>of</strong> the<br />
questionnaire were also distributed in person by the first author to customer services<br />
employees located at Centrelink’s main metropolitan centres in Perth. Those who completed<br />
the hard copy version indicated that they had not completed the online survey. In total, online<br />
and hard copy questionnaires were returned by 260 employees, although missing data<br />
reduced the final data set to 234 cases. The response rate is slightly below the average <strong>of</strong> 21<br />
percent reported in a study <strong>of</strong> response rates in the 1990s (Paxson, Dillman, & Tarnai, 1995),<br />
and should be considered a limitation <strong>of</strong> this study. However, the sample size is adequate for<br />
inferential data analysis (Guadagnoli & Velicer, 1988; Hair, Anderson, Tatham, & Black,<br />
58
1998). Also, statistical power for these data is well above the .80 level necessary to detect<br />
medium to large effects in regression-based research (Cohen, 1988)<br />
Measures<br />
Customer Orientation. The dependent variable – a training outcome – in this study is<br />
the employee’s customer orientation, which is defined as “an employee’s tendency or<br />
predisposition to meet customer needs in an on-the-job context” (Brown, Mowen, Donavan,<br />
& Licata, 2002, p. 111). This variable was measured using the 22-item (α = .97) scale<br />
developed by Kelly (1992). A typical item is “It is important to have operating hours that are<br />
convenient for customers”. Other customer orientation scales were not used because they<br />
emphasized selling (rather than service) or after sales service, or were less parsimonious with<br />
no apparent advantage.<br />
Training Transfer Climate. This study adopted the 56-item scale refined and tested by<br />
Machin and Fogarty (2004). This instrument was originally developed by Rouiller and<br />
Goldstein (1993) based on their behaviourist and social learning conceptual foundation <strong>of</strong><br />
training transfer variables. The instrument was later refined and adapted for military use by<br />
Thayer & Teachout (1995). The current version by Machin and Fogarty is a further<br />
development <strong>of</strong> the previous two versions. The instrument consists <strong>of</strong> six subscales, three<br />
relating to workplace cues (goal cues, social cues and task cues), and three relating to<br />
workplace consequences (positive reinforcement, negative reinforcement and punishment,<br />
and extinction). Machin and Fogarty (2004) reported good reliabilities for five subscales and<br />
marginally acceptable (α = .66) for the negative reinforcement and punishment subscale. The<br />
present study reports similar results; reliability exceeded .80 on each the first five subscales,<br />
but was exceptionally low (α = .27) on the sixth dimension. As with most training transfer<br />
instruments (for an exception see: Holton et al (1997)), this scale has received limited<br />
construct validation <strong>of</strong> its assumed underlying factor structure.<br />
Organizational Commitment. This variable was measured using the 15-item (α = .91)<br />
scale developed by (Mowday et al., 1979). A typical item is “I could just as well be working<br />
for a different organization as long as the type <strong>of</strong> work was similar”. This scale is widely<br />
used when the research focuses only on affective attachment (Mowday, 1998).<br />
Cynicism about Organizational Change. This variable was measured using the 8-item<br />
(α = .93) scale developed by Wanous, Reichers, & Austin (2000). This scale consists <strong>of</strong> four<br />
items measuring the level <strong>of</strong> pessimism regarding the likely success <strong>of</strong> change, and four<br />
items measuring dispositional attribution about those people responsible for effecting<br />
successful change. A typical item is “The people responsible for making things better around<br />
here do not care enough about their jobs.”<br />
Training Attendance. Data were also obtained relating to the number <strong>of</strong> times<br />
respondents had been exposed to the training intervention, and the most recent attendance at<br />
a workshop. In this study, whether or not the respondent had attended the Value Creation<br />
Workshop was coded as a dichotomous variable (1=Yes, 0=No).<br />
59
Results<br />
To test the hypotheses, two separate statistical analyses were conducted. The first<br />
hypothesis was tested through factor analysis <strong>of</strong> the training transfer climate instrument. The<br />
remaining hypotheses were tested through hierarchical multiple regression.<br />
Factor Analysis <strong>of</strong> Training Transfer Climate Scale<br />
Maximum likelihood factor analysis with promax rotation was conducted on the 56-item<br />
training transfer climate scale, which produced twelve factors with eigenvalues greater than 1<br />
and explained 66% <strong>of</strong> the variance. These results were not readily interpretable due to few<br />
items on each factor and several items with cross loading above .40. Next, items were forced<br />
to load on six factors, as specified in the original studies. These results (available from the<br />
authors) did not support the hypothesized subscales due to high cross-loading <strong>of</strong> some items<br />
and low eigenvalues for some items. Given problems with the original six-factor model, a<br />
scree plot was conducted on the training transfer climate items. This plot indicated that a<br />
four-factor structure would be the most appropriate fit for this data. The original 56 items<br />
were forced to load on four factors, and items loading below .50 were deleted. The<br />
procedure was repeated two more times. This analytic process resulted in a 35-item<br />
instrument with all items loading cleanly on no more than one factor. The four factors –<br />
which we labeled support, extinction, control and outcomes (reinforcement) -- have<br />
reliabilities ranging from .73 to .94 and the factor structure accounts for 48% <strong>of</strong> variance.<br />
The support factor, which has 20 items, includes all statements relating to physical and<br />
informational resources (e.g., “Necessary tools/equipment are available”, “Supervisors<br />
answer VCW questions”) as well as items pertaining to training transfer goals (e.g.,<br />
“Supervisors help staff to set goals”). The extinction factor includes seven items, including<br />
all items found in the original extinction subscale. These items represent the supervisor’s<br />
indifference to the training program and its application (e.g. “Supervisors don’t notice use <strong>of</strong><br />
VCW ideas”, “Supervisors don’t care about VCW”). The outcomes factor consists <strong>of</strong> four<br />
items that combine positive reinforcement and punishment. Two items refer to prospects <strong>of</strong><br />
promotion for those who apply their training (e.g., “No promotion without using VCW<br />
ideas”). The other two items refer to how staff are treated when the training is applied on the<br />
job. The control factor consists <strong>of</strong> four items, three <strong>of</strong> which refer to time limitations <strong>of</strong><br />
applying the training (e.g., “No time to do the job using VCW ideas”).<br />
Based on the revised training transfer climate instrument, Table 1 presented the means,<br />
standard deviations, internal consistency reliabilities and zero-order correlations for all<br />
variables in this study. These data indicate that all variables have good reliability and, except<br />
for customer orientation, are only slightly skewed from normality. The high mean score for<br />
customer orientation is expected, because the organization specifically attracts and selects<br />
people with a strong customer orientation. Although not ideal for multivariate research, the<br />
distribution for this variable was not normalized because <strong>of</strong> warnings about such a procedure<br />
(Cohen & Cohen, 1983) and because attitude data are typically negatively skewed<br />
(McDougall & Beattie, 1998). The skewed distribution likely suppresses the true effect sizes<br />
<strong>of</strong> variables associated with customer orientation.<br />
60
Most variables also have low or moderate intercorrelations. The highest intercorrelation<br />
is between organizational commitment and cynicism about organizational change (r=-.67),<br />
which is consistent with the relationship between these two variables in previous studies.<br />
Support and extinction also have a high zero-order correlation (r=.65). However, both <strong>of</strong><br />
these correlations are below the .70 limit suggested by Tabachnick and Fidell (1996).<br />
Examination <strong>of</strong> the correlation diagnostics established that no condition indexes were greater<br />
than 30, indicating that multicollinearity was not a significant problem. However, inspection<br />
<strong>of</strong> the variance inflation factors (VIFs) suggests that multicollinearity may in fact be present,<br />
posing a potential, but not serious, problem for the interpretation <strong>of</strong> these results. All<br />
independent variables were centered prior to regression analyses to minimize<br />
multicollinearity issues between first order and higher order (interaction) terms (Aiken &<br />
West, 1991; Cohen & Cohen, 1983; MacCallum & Mar, 1995).<br />
Table 1<br />
Means, standard deviations, reliabilities and bivariate correlations<br />
Variable α Mean SD 1 2 3 4 5 6 7<br />
1. Customer<br />
orientation<br />
.96 6.24 .80<br />
2. Training<br />
attendance y/n<br />
- - - .14*<br />
3. Commitment .90 4.12 1.16 .18** -.04<br />
4. Cynicism about<br />
change<br />
.92 3.41 1.31 -.20** .02 -.67**<br />
5. Support .94 3.96 1.09 .22** .03 .51** -.63**<br />
6. Extinction .86 4.<strong>29</strong> 1.31 .05 -.00 .52** -.60** .65**<br />
7. Outcomes .75 3.00 1.11 -.30 -.06 .15* -.07 .26** .06<br />
8. Control .73 2.96 1.28 -.18** -.04 .42** -.40** .27** .45** .15*<br />
**Correlation is significant at the 0.01 level.<br />
*Correlation is significant at the 0.05 level.<br />
Hierarchical Regression Analysis<br />
The remaining hypotheses postulated that there would be a significant relationship<br />
between each <strong>of</strong> the independent variables, including the interaction terms, and the dependent<br />
variable customer orientation. These hypotheses were tested using hierarchical regression<br />
analysis, shown in Table 2. Independent variables were entered in a logical sequence that<br />
paralleled the transfer experience. As per moderated regression procedures, the higher order<br />
(interaction) terms were entered in later steps. The results showed a good fit <strong>of</strong> the data to the<br />
model. The overall model produced an R² <strong>of</strong> .30, and an adjusted R² <strong>of</strong> .26.<br />
61
Table 2<br />
Hierarchical Regression Analysis <strong>of</strong> Customer Orientation<br />
Variable B SE B β R² Adj. R² ∆ R² ∆ F<br />
Step 1 .02 .01 .02 4.32*<br />
Training .11 .05 .14*<br />
Step 2 .05 .04 .03** 8.32**<br />
Training .11 .05 .14*<br />
Commitment .15 .05 .19**<br />
Step 3 .06 .05 .01 2.70<br />
Training .11 .05 .14*<br />
Commitment .07 .07 .09<br />
Cynicism -.11 .07 -.14<br />
Step 4 .17 .15 .11** 7.49**<br />
Training .10 .05 .12<br />
Commitment .14 .07 .18*<br />
Cynicism -.12 .08 -.15<br />
Support .17 .07 .21*<br />
Extinction -.11 .07 -.13<br />
Outcomes -.05 .05 -.06<br />
Control -.24 .06 -.30**<br />
Step 5 .19 .16 .02 2.70<br />
Training -.20 .17 -.25**<br />
Commitment .16 .07 .20*<br />
Cynicism -.08 .08 -.10<br />
Support .21 .08 .26**<br />
Extinction -.12 .07 -.15<br />
Outcomes -.05 .05 -.07<br />
Control -.24 .06 -.30**<br />
Train X Commit. .01 .06 .01<br />
Train X Cynicism .09 .05 .40<br />
Step 6 .30 .26 .11** 8.46**<br />
Training -.54 .18 -.67**<br />
Commitment .18 .07 .23**<br />
Cynicism -.07 .07 -.09<br />
Support -.17 .07 .21*<br />
Extinction -.12 .07 -.15<br />
Outcomes -.04 .05 -.05<br />
Control -.24 .05 -.30**<br />
Train X Commit. -.11 .06 -.13<br />
Train X Cynicism .19 .05 .86**<br />
Train X Support .21 .07 .23**<br />
Train X Extinction .01 .08 .01<br />
Train X Outcomes -.04 .05 -.05<br />
Train X Control .26 .06 .34**<br />
* p ≤ .05 ** p ≤ .01 B values are unstandardized coefficients; β values are standardized coefficients<br />
62
Participation in the Value Creation Workshop was entered in step 1 <strong>of</strong> the hierarchical<br />
regression analysis. Alone, this variable had a significant but minimal practical association<br />
with customer orientation (adjusted R² = .01). The individual trainee characteristic <strong>of</strong><br />
organizational commitment was added to the model in step 2. The change in R² was<br />
significant (p ≤ .01), and R² increased to .05 (adjusted R² = .04), indicating that<br />
organizational commitment was a significant predictor <strong>of</strong> customer orientation. Cynicism<br />
about organizational change was added in step 3, but did not add significantly to the model.<br />
The four transfer climate variables -- support, extinction, control and outcomes -- were<br />
entered together in step 4. Entering these variables resulted in a significant increase in R²,<br />
raising the explained variance by 10 percent (adjusted R² = .15). Only the transfer climate<br />
dimensions <strong>of</strong> support (β=0.23, p ≤ .01), and control (β=-0.30, p ≤ .01) were statistically<br />
significant. The β weight for control is negative, since high scores on this variable indicate<br />
low levels <strong>of</strong> control were statistically significant. Control has a negative regression weight<br />
because high scores on this subscale indicate low levels <strong>of</strong> control. Step 5 <strong>of</strong> the hierarchical<br />
regression analysis consisted <strong>of</strong> entering the interaction terms <strong>of</strong> training with the two trainee<br />
characteristics (organizational commitment and cynicism about organizational change). The<br />
change in R² was not significant, indicating that the interaction terms do not predict customer<br />
orientation at this step.<br />
To determine the appropriate variables to enter in the next step an effect size comparison<br />
procedure was conducted (Lubinski & Humphreys, 1990). This procedure identifies whether<br />
the underlying structural model is best approximated by a curvilinear relationship or by a<br />
moderator relationship. The effect size comparison procedure involves testing whether a<br />
quadratic (squared) term for transfer climate produces a greater increase in R² than does the<br />
interaction term for transfer climate and training attendance. The results <strong>of</strong> this procedure<br />
suggest that, while a significant curvilinear effect is present, a stronger more meaningful<br />
moderator effect is also present, and the moderator effect better describes the underlying<br />
structure <strong>of</strong> the relationship between transfer climate and customer orientation.<br />
Based on the effect size comparison procedure, step 6 consisted <strong>of</strong> entering the four<br />
interaction terms <strong>of</strong> training attendance with the four training transfer climate variables. The<br />
change in R² was significant and R² increased substantially to .30 (adjusted R² = .26),<br />
indicating that the interaction between attendance and transfer climate was a significant<br />
predictor <strong>of</strong> customer orientation. In this final step <strong>of</strong> the multiple regression analysis, four<br />
first order variables were significantly associated with customer orientation: training, control,<br />
organizational commitment, and support. In addition, three interaction variables were<br />
statistically significant: training with control, support, and cynicism about organizational<br />
change.<br />
Discussion and Conclusions<br />
This study set out to improve our knowledge <strong>of</strong> transfer <strong>of</strong> training in three ways. First,<br />
we hoped to contribute to the construct validation <strong>of</strong> the training transfer climate scale that<br />
has received recent attention. Second, our goal was to further illuminate the effect <strong>of</strong> training<br />
transfer climate, particularly to examine the moderator rather than just the direct effects <strong>of</strong><br />
this variable’s dimensions. Finally, this study investigated the effects on training transfer <strong>of</strong><br />
63
two trainee characteristics that previously received limited attention. The results <strong>of</strong> each <strong>of</strong><br />
these research objectives are discussed below.<br />
Training Transfer Climate Direct Effects<br />
This study reported that two <strong>of</strong> the four training transfer climate variables – control and<br />
support– had strong direct associations in the predicted direction with customer orientation.<br />
The strong effect size for control reflects the idea that employees have less <strong>of</strong> a customer<br />
orientation when faced with heavy workloads. Customer service may require additional time,<br />
which is not available when workloads are high. With less time to provide good customer<br />
service, employees eventually feel less customer orientation.<br />
Support is a broadly-defined construct in the training transfer literature, but consistently<br />
includes actions by the supervisor as well as workplace conditions that support application <strong>of</strong><br />
the learned skills or knowledge back on the job. Several measures <strong>of</strong> support, including the<br />
one in this study, also include setting objectives that encourage use <strong>of</strong> trained material on the<br />
job (McSherry & Taylor, 1994). The significance <strong>of</strong> support as a direct predictor <strong>of</strong> customer<br />
orientation is consistent with many previous studies on transfer <strong>of</strong> training (Brinkerh<strong>of</strong>f &<br />
Montesino, 1995; Facteau et al., 1995; Lim & Johnson, 2002; Nijman, Nijh<strong>of</strong>, Wognum, &<br />
Veldkamp, 2006). Supervisor support, for example, might directly motivate employees to<br />
provide better customer service, whether or not the company <strong>of</strong>fered a training program to<br />
also strengthen this orientation.<br />
Training Transfer Climate Moderator Effects<br />
One <strong>of</strong> the most important findings in this study is that training transfer climate (or<br />
specific dimensions <strong>of</strong> this variable) moderates the relationship between the customer service<br />
training intervention and the employee’s customer orientation. These moderator variables<br />
alone added 11 percent explained variance to the equation, a substantially higher result than<br />
is found in most studies involving moderator effects (Champoux & Peters, 1987; Chaplin,<br />
1991). Equally important, the variance in the dependent variable attributable to the<br />
moderator effect <strong>of</strong> transfer climate was as great as the variance attributable its main effect.<br />
The moderator effect for control indicates that employees with heavy workloads are<br />
much less likely to apply what they learned from training back on the job. Employees might<br />
cope with work overload by applying familiar work procedures rather than new practices<br />
learned through the training program. The moderator effect for support indicates that<br />
employees who receive supervisor support, work resources, and work goals coaching are<br />
more likely to apply what they learned from training back on the job compared to those who<br />
receive less support. This finding makes sense for two reasons. First, employees have an<br />
easier time applying their training back on the job where the work environment has resources<br />
needed to improve customer service. Second, training transfer requires the employee to<br />
generalize what they have learned in the training program to the job (Baldwin & Ford, 1988).<br />
Even the best training programs require on-the-job support from supervisors and co-workers<br />
to guide trainees through this generalization process; in other words, to show them how and<br />
where to apply their newfound skills, knowledge, and attitudes in the workplace.<br />
The strong moderator effect <strong>of</strong> training transfer climate has important implications for<br />
future research on this topic. Most training transfer literature suggests that the effect <strong>of</strong><br />
64
training on workplace behaviour depends on (i.e. is contingent upon) the level <strong>of</strong> training<br />
transfer climate (or its specific dimensions) as well as trainee characteristics and related<br />
variables. In other words, they typically propose that these variables moderate the<br />
relationship between training and workplace behaviour. Yet, the majority <strong>of</strong> training transfer<br />
studies test only for the direct effect <strong>of</strong> each variable, not for the moderator effect. The few<br />
studies that do conduct moderator analysis typically overlook the procedure <strong>of</strong> initially<br />
testing for quadratic rather than moderator relationships. This study’s findings suggest that<br />
training transfer climate not only moderates the effect <strong>of</strong> training on employee behaviour and<br />
attitudes; it is one <strong>of</strong> the most important predictors <strong>of</strong> training outcomes. Future research<br />
doesn’t need to change the underlying model; rather, it needs to conduct a moderator effect<br />
as current models specify.<br />
Individual Trainee Characteristics<br />
This study hypothesized that organizational commitment would be positively related to<br />
employee customer orientation, and that cynicism about organizational change would be<br />
negatively related to employee customer orientation. The significant direct effect <strong>of</strong><br />
organizational commitment on customer orientation reported in this study is consistent with<br />
findings in some earlier studies, but not others (Bartlett, 2001; Facteau et al., 1995; Orpen,<br />
1999; Tesluk, Farr, Mathieu, & Vance, 1995; Tracey et al., 2001). The findings <strong>of</strong> the present<br />
study suggest that employees with a stronger affective commitment to the organization are<br />
more dedicated to the employer’s goals and initiatives, which would include serving<br />
customers where (as in this organization) customer service is one <strong>of</strong> the organization’s core<br />
values. Committed employees simply accept that customer oriented behaviour is appropriate<br />
because they share the values and ideals <strong>of</strong> the organization. In addition, employees with<br />
higher affective commitment may have higher customer orientation because their<br />
commitment motivates them to be more mindful during the learning process and to apply that<br />
knowledge or skill on the job. This is consistent with the observation by Tesluk et al (1995)<br />
that organizational commitment is likely to have a greater effect when a training intervention<br />
is strongly endorsed by the organization.<br />
Cynicism about organizational change is a relatively new construct that has received<br />
limited attention from researchers. Tesluk et al (1995) examined the impact <strong>of</strong> the broad<br />
construct <strong>of</strong> cynicism on generalization <strong>of</strong> training, but we could find no other research that<br />
has investigated the more specific construct <strong>of</strong> cynicism about organizational change in the<br />
context <strong>of</strong> training transfer. Training is an important strategy for managing the change<br />
process, so it seems logical that cynicism about organizational change would have a<br />
significant direct effect on training outcomes. This variable was nonsignificant in the present<br />
study, however. One possible explanation is that, due to the high negative correlation<br />
between organizational commitment and cynicism about organizational change, the latter<br />
<strong>of</strong>fered little additional explanatory value beyond the variance explained by organizational<br />
commitment. Another explanation may be that this variable is a broadly-based attitude that<br />
would not interfere with an ongoing specific training initiative. Perhaps other factors in the<br />
workplace combined to encourage customer oriented behaviour, and this outweighed the<br />
impact <strong>of</strong> individual cynicism about customer orientation.<br />
65
Although not specifically hypothesized, this study also examined whether the trainee<br />
characteristics moderated the relationship between participation in the customer service<br />
training program and employee customer orientation. The interaction term <strong>of</strong> training<br />
participation with organizational commitment was not significant. This suggests that the<br />
strength <strong>of</strong> the relationship between participation in the workshop and customer orientation<br />
was unaffected by the level <strong>of</strong> employee organizational commitment or cynicism about<br />
organizational change.<br />
The interaction term regarding training participation and cynicism about organizational<br />
change was statistically significant in the final step <strong>of</strong> the regression analysis, suggesting that<br />
the strength <strong>of</strong> the relationship between participation in the workshop and customer<br />
orientation varied with the level <strong>of</strong> employee cynicism about organizational change. This<br />
finding is intuitively appealing, since employees with low levels <strong>of</strong> cynicism about change<br />
would logically be expected to apply new learning to a greater extent than employees who<br />
were cynical about the value <strong>of</strong> trying to implement change in their jobs. The finding that an<br />
interaction between attendance at a training intervention and cynicism about organizational<br />
change is a significant predictor <strong>of</strong> transfer is a new finding that has not been previously<br />
reported in the literature. However, this finding seems to run contrary to the earlier finding<br />
that there is no statistically significant main effect on customer orientation for this variable.<br />
While the absence <strong>of</strong> a main effect does not conceptually preclude the existence <strong>of</strong> a<br />
moderator effect (Baron & Kenny, 1986), it makes the moderator effect more difficult to<br />
explain. The most likely explanation is that while employees who are cynical about change<br />
can still have high levels <strong>of</strong> customer orientation, employees who are cynical about change<br />
are less likely to transfer training content, in this case customer orientation, to their jobs.<br />
Influence <strong>of</strong> Participation in the Training Program<br />
In order to identify the influence <strong>of</strong> participation in the training intervention, it was also<br />
hypothesized that attendance at the workshop would be positively related to employee<br />
customer orientation. The direct effect <strong>of</strong> participation in the training intervention was<br />
significant when first entered, became non-significant when the transfer climate variables<br />
were entered, and did not become a significant predictor <strong>of</strong> customer orientation until all<br />
other variables, including the interaction terms, had been entered into the regression model.<br />
This suggests that when individual trainee characteristics and transfer climate variables are<br />
accounted for, participation in the training program does have a positive impact on transfer.<br />
Factor Structure <strong>of</strong> Training Transfer Climate Instrument<br />
We were surprised to discover that the data in this study did not support the six-factor<br />
structure <strong>of</strong> training transfer climate that had been developed (with minor variations) and<br />
applied in two previous widely-cited studies as well as in a more recent study. Instead, the<br />
study found that training transfer climate consists <strong>of</strong> four factors, relating to supervisory and<br />
co-worker support, including both social and material support (support), behaviour from<br />
supervisors that directly discourages transfer (extinction), personal control over workload<br />
and workflows (control), and the degree to which transfer leads to desired personal outcomes<br />
(outcomes).<br />
66
While the hypothesized six-factor structure was not supported, the four-factor structure<br />
found in this study (support, extinction, control, and outcomes) identifies aspects <strong>of</strong> the work<br />
environment that intuitively would be likely to affect employee attitudes to the transfer <strong>of</strong><br />
new skills and knowledge to the job. Furthermore, this analysis provides further insight<br />
regarding the current debate about whether training transfer climate is a multidimensional or<br />
unidimensional construct (Burke & Hutchins, 2007; Tracey, 2005; Tracey et al., 2001). The<br />
workplace cues and consequences dimensions introduced by Rouiller and Goldstein (1993)<br />
loaded on one factor in this study. However, the resulting four-factor structure had minimal<br />
cross-loadings and, equally important, seems to have differential association with other<br />
variables in this study.<br />
These findings complement and extend the results <strong>of</strong> earlier studies on training transfer<br />
climate in one important way. Most previous studies have not attempted to understand the<br />
makeup <strong>of</strong> the training transfer climate construct itself, but have instead sought to examine<br />
the influence <strong>of</strong> particular variables upon transfer. In many cases, they have simply<br />
combined a number <strong>of</strong> items, and claimed that the amalgam constituted transfer climate. In<br />
other cases, researchers adopted scales used in earlier studies, but even these more frequently<br />
used scales were rarely tested for factor structure and other elements <strong>of</strong> construct validity.<br />
While the results <strong>of</strong> this study do not point to the existence <strong>of</strong> transfer climate variables that<br />
have not been previously identified, they do suggest that a different combination <strong>of</strong><br />
dimensions might comprise the transfer climate construct.<br />
Each <strong>of</strong> these four factors has been identified as a transfer climate variable in previous<br />
studies (Bates, Holton, Seyler, & Carvalho, 2000; Elwood F. III Holton, Bates, & Ruona,<br />
2000; E. F. III Holton et al., 1997). In those studies alone, however, a total <strong>of</strong> from nine to<br />
sixteen factors were found to constitute the transfer climate construct. The more<br />
parsimonious structure identified in this study has the practical utility that Noe (2000) has<br />
urged, as well as theoretical rigour, and is consequently a useful contribution to<br />
understanding transfer climate.<br />
One recent study (Clarke, 2002) identified a transfer climate structure almost identical to<br />
the structure found in this study. Clarke found that heavy workloads, time pressures (control),<br />
lack <strong>of</strong> reinforcement (support), an absence <strong>of</strong> feedback (extinction), and the degree to which<br />
employees saw in-service training as personal development rather than job related training<br />
(outcomes) were work environment factors that impede transfer <strong>of</strong> training. The factor<br />
structure identified by Kirby and colleagues (2003) also has similarities to the structure<br />
identified in this study. They identified three factors, which they labeled good supervision,<br />
workload, and choice independence. Their first factor related to a supportive and receptive<br />
work environment, with good supervision and clear expectations, broadly equivalent to the<br />
support factor in the present study. Their second factor related to workload levels, and their<br />
third factor was related to independence about the way employees could approach their work.<br />
These two factors equate to the control factor that emerged in the present study.<br />
A particularly interesting finding <strong>of</strong> this study is the emergence <strong>of</strong> the factor that relates<br />
to control over work and pressure <strong>of</strong> time. Oddly enough, this seems to support one piece <strong>of</strong><br />
Rouiller and Goldstein’s (1993) original model, namely “self-control” cues. That dimension<br />
was removed from subsequent studies by other researchers on the basis that the dimension<br />
67
elated to personality rather than to training transfer climate. However, a careful reading <strong>of</strong><br />
Rouiller and Goldstein’s (1993) paper reveals that they were clearly referring to the extent to<br />
which a trainee has control over his or her own work, and not to some aspect <strong>of</strong> individual<br />
personality.<br />
Implications and Limitations<br />
The findings <strong>of</strong> this study suggest a number <strong>of</strong> ways in which managers can improve the<br />
degree <strong>of</strong> training transfer. The importance <strong>of</strong> supervisor and co-worker support behaviour<br />
made apparent by this study confirms the findings <strong>of</strong> many earlier studies. Managers need to<br />
show that they are prepared to set goals, reinforce learning and provide appropriate tools and<br />
equipment. Failure to do so will substantially inhibit transfer. Supportive co-worker<br />
behaviour is likely to encourage the transfer <strong>of</strong> training to the job through peer pressure. In<br />
addition, managers need to convince employees that they really care about the training being<br />
transferred to the job. They need to ensure that employees have the time and freedom to<br />
apply their newly acquired skills, without the pressure <strong>of</strong> high workload levels or constant<br />
interruptions. This study has shown that employees are more likely to transfer learning to the<br />
job if they can reasonably expect to receive some valued personal outcomes by doing so.<br />
Managers also need to address employee concerns about the sincerity <strong>of</strong> their statements<br />
to employees. The interaction <strong>of</strong> attendance and cynicism about organisational change was<br />
found to a significant predictor <strong>of</strong> customer orientation. Cynicism about organisational<br />
change is a two-dimensional construct consisting <strong>of</strong> pessimism (programs that are supposed<br />
to solve problems around here won’t do much good) and dispositional attribution (the people<br />
responsible for solving problems around here don’t care enough about their jobs). High<br />
levels <strong>of</strong> cynicism about organisational change, and high levels <strong>of</strong> employee cynicism in<br />
general, are likely to be the outcome <strong>of</strong> a series <strong>of</strong> negative organisational events or<br />
experiences in which employees feel that they have in some way been let down or misled by<br />
the organisation or its management.<br />
This raises a number <strong>of</strong> important issues for managers. Firstly, employees are not likely<br />
to transfer learning if they perceive that managers are not serious about the value <strong>of</strong> the<br />
training or the need to transfer. This is particularly so where the training intervention is not a<br />
traditional skill based one, and might be considered by some employees as simply the latest<br />
management fad. Managers who direct employees to attend such a program, even though the<br />
manager may see little value in the program, are likely to be perceived as lacking integrity,<br />
and are unlikely to gain employee commitment to transferring any new skills. Managers who<br />
do not demonstrate that their decisions are based on soundly developed strategic thinking are<br />
also unlikely to inspire confidence and trust in their fitness for management.<br />
Finally, this study has several limitations that need to be acknowledged. One design<br />
limitation is the fact that the study relied on employee self-reports at one point in time. It is<br />
possible that a longitudinal study would have provided better insight regarding the influence<br />
<strong>of</strong> the transfer climate on transfer relapse. Also, data on third-party observations <strong>of</strong> employee<br />
customer service would have greatly added to the quality <strong>of</strong> this research. Unfortunately,<br />
constraints imposed upon the study necessitated complete respondent anonymity. The lower<br />
than desired response rate is another limitation, which raises the possibility that the<br />
68
espondent group was not representative <strong>of</strong> the population and that relationships among the<br />
variables could vary from the population parameters. Finally, this study measured attendance<br />
at the training program, which is a relatively simplistic dichotomous indicator representing<br />
whether the individual learned from the training program (i.e. those who didn’t attend would<br />
not have learned, except possibly indirectly through association with attendees). A more<br />
precise indicator <strong>of</strong> training would be to measure how much learning actually occurred by<br />
each person within the program.<br />
69
References<br />
Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and interpreting<br />
interactions. Thousand Oaks: Sage Publications Inc.<br />
Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. L. (2000). Manufacturing advantage:<br />
Why high-performance work systems pay <strong>of</strong>f. Ithaca, N. Y.: Cornell University Press.<br />
Australian National Audit Office. (2005). Centrelink's Value Creation Program. Canberra:<br />
Government <strong>of</strong> Australia.<br />
Baldwin, T. T., & Ford, J. K. (1988). Transfer <strong>of</strong> training: A review and directions for future<br />
research. Personnel Psychology, 41(1), 63-105.<br />
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social<br />
psychological research: Conceptual, strategic, and statistical considerations. Journal<br />
<strong>of</strong> Personality and Social Psychology, 51(6), 1173-1182.<br />
Bartlett, K. R. (2001). The relationship between training and organizational commitment: A<br />
study in the health care field. Human Resource Development Quarterly, 12(4), 335-<br />
352.<br />
Bates, R. A., Holton, E. F. I., Seyler, D. L., & Carvalho, M. A. (2000). The role <strong>of</strong><br />
interpersonal factors in the application <strong>of</strong> computer-based training in an industrial<br />
setting. Human Resource Development International, 3(1), 19-42.<br />
Becker, B. E., & Huselid, M. A. (2006). Strategic human resources management: Where do<br />
we go from here? Journal <strong>of</strong> Management, 32(6), 898-925.<br />
Bennington, L. (1999). Tools for transforming organisations. In F. S<strong>of</strong>o (Ed.), Human<br />
resource development: Perspectives, roles, and practice choices (pp. 301-320).<br />
Warriewood, NSW: Business and Pr<strong>of</strong>essional Publishing Pty Ltd.<br />
Bennington, L., & Cummane, J. (1998). Measuring service quality: A hybrid methodology.<br />
Total Quality Management, 9(6), 395-405.<br />
Benson, G. S., Young, S. M., & Lawler III, E. E. (2006). High-involvement work practices<br />
and analysts' forecasts <strong>of</strong> corporate earnings. Human Resource Management, 45(4),<br />
519-537.<br />
Brinkerh<strong>of</strong>f, R. O., & Montesino, M. U. (1995). Partnership for training transfer: Lessons<br />
from a corporate study. Human Resource Development Quarterly, 6(3), 263-274.<br />
Brown, T. J., Mowen, J. C., Donavan, D. T., & Licata, J. W. (2002). The customer<br />
orientation <strong>of</strong> service workers: Personality trait effects on self and supervisor<br />
performance ratings. Journal <strong>of</strong> Marketing Research, 39(1), 110-119.<br />
Burke, L., & Baldwin, T. T. (1999). Workforce training transfer: A study <strong>of</strong> the effect <strong>of</strong><br />
relapse prevention training and transfer climate. Human Resource Management, 38(3),<br />
227-241.<br />
Burke, L., & Hutchins, H. M. (2007). Training transfer: An integrative literature review.<br />
Human Resource Development Review, 6(3), 263-<strong>29</strong>6.<br />
Centrelink. (2007). Annual report, 2006-07. Canberra: Commonwealth <strong>of</strong> Australia.<br />
Champoux, J. E., & Peters, W. S. (1987). Form, effect size, and power in moderated<br />
regression analysis. Journal <strong>of</strong> Occupational Psychology, 60, 243-255.<br />
Chaplin, W. F. (1991). The next generation <strong>of</strong> moderator research in personality psychology.<br />
Journal <strong>of</strong> Personality, 59, 143-178.<br />
70
Clarke, N. (2002). Job/work environment factors influencing training transfer within a human<br />
service agency: Some indicative support for Baldwin and Ford's transfer climate<br />
construct. International Journal <strong>of</strong> Training and Development, 6(3), 146-162.<br />
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale,<br />
NJ.: Lawrence Erlbaum,.<br />
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the<br />
behavioral sciences (2nd ed.). Hillsdale, New Jersey: Lawrence Erlbaum Associates,<br />
Inc.<br />
Combs, J., Liu, Y., Hall, A., & Ketchen, D. (2006). How Much Do High-Performance Work<br />
Practices Matter? A Meta-Analysis Of Their Effects On Organizational Performance.<br />
Personnel Psychology, 59(3), 501-528.<br />
Cromwell, S. E., & Kolb, J. A. (2004). An examination <strong>of</strong> work-environment support factors<br />
affecting transfer <strong>of</strong> supervisory skills training to the workplace. Human Resource<br />
Development Quarterly, 15(4), 449-471.<br />
Facteau, J. D., Dobbins, G. H., Russell, J. E. A., Ladd, R. T., & Kudisch, J. D. (1995). The<br />
influence <strong>of</strong> general perceptions <strong>of</strong> the training environment on pretraining<br />
motivation and perceived training transfer. Journal <strong>of</strong> Management, 21(1), 1-25.<br />
Grant, M., & Hughes, P. D. (2007). Learning and development outlook 2007: Are we<br />
learning enough? Ottawa: Conference Board <strong>of</strong> Canada.<br />
Guadagnoli, E., & Velicer, W. F. (1988). Relation <strong>of</strong> sample size to the stability <strong>of</strong><br />
component patterns. Psychological Bulletin, 103(2), 265-275.<br />
Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate Data<br />
Analysis (5th ed.). Upper Saddle River, NJ: Prentice-Hall Inc.<br />
Halligan, J. (2007). Advocacy and Innovation in Interagency Management: The Case <strong>of</strong><br />
Centrelink. Governance, 20(3), 445-467.<br />
Holton, E. F. I., Bates, R. A., & Ruona, W. E. A. (2000). Development <strong>of</strong> a generalized<br />
learning transfer system inventory. Human Resource Development Quarterly, 11(4),<br />
333-360.<br />
Holton, E. F. I., Bates, R. A., Seyler, D. L., & Carvalho, M. B. (1997). Towards construct<br />
validation <strong>of</strong> a transfer climate instrument. Human Resource Development Quarterly,<br />
8(2), 95-113.<br />
Kelley, S. (1992). Developing customer orientation among service employees. Journal <strong>of</strong> the<br />
Academy <strong>of</strong> marketing Science, 20(1), 27-36.<br />
Kirby, J. R., Knapper, C. K., Evans, C. J., Carty, A. E., & Gadula, C. (2003). Approaches to<br />
learning at work and workplace climate. International Journal <strong>of</strong> Training and<br />
Development, 7(1), 31-52.<br />
Kontoghiorghes, C. (2001). Factors affecting training effectiveness in the context <strong>of</strong> the<br />
introduction <strong>of</strong> new technology - a US case study. International Journal <strong>of</strong> Training<br />
and Development, 5(5), 248-260.<br />
Kozlowski, S. W. J., Brown, K., Weissbein, D. A., Cannon-Bowers, J. A., & Salas, E. (2000).<br />
A multilevel approach to training effectiveness. In K. Klein & S. W. J. Kozlowski<br />
(Eds.), Multi-level theory, research, and methods in organizations (pp. 157-210). San<br />
Francisco, CA: Jossey-Bass.<br />
Lawler III, E. E., Mohrman, S. A., & Ledford Jr., G. E. (1998). Strategies for high<br />
performance organizations. San Francisco: Jossey-Bass.<br />
Lim, D. H., & Johnson, S. D. (2002). Trainee perceptions <strong>of</strong> factors that influence learning<br />
transfer. International Journal <strong>of</strong> Training and Development, 6(1), 36-48.<br />
71
Lim, D. H., & Morris, M. L. (2006). Influence <strong>of</strong> trainee characteristics, instructional<br />
satisfaction, and organizational climate on perceived learning and training transfer.<br />
Human Resource Development Quarterly, 17(1), 85-115.<br />
Lubinski, D., & Humphreys, L. G. (1990). Assessing spurious moderator effects: Illustrated<br />
substantively with the hypothesised ("synergistic") relation between spatial and<br />
mathematical ability. Psychological Bulletin, 107(3), 385-393.<br />
MacCallum, G. H., & Mar, C. M. (1995). Distinguishing between moderator and quadratic<br />
effects in multiple regression. Psychological Bulletin, 118(3), 405-421.<br />
Machin, M. A., & Fogarty, G. J. (2004). Assessing the antecedents <strong>of</strong> transfer intentions in a<br />
training context. International Journal <strong>of</strong> Training and Development, 8(3), 222-236.<br />
Mason, C. A., Tu, S., & Cauce, A. M. (1996). Assessing moderator variables: Two computer<br />
simulation studies. Educational and Psychological Measurement, 56(1), 45-62.<br />
Mathieu, J. E., Tannenbaum, S. I., & Salas, E. (1992). Influences <strong>of</strong> individual and<br />
situational characteristics on measures <strong>of</strong> training effectiveness. Academy <strong>of</strong><br />
Management Journal, 35(4), 828-847.<br />
McDougall, M., & Beattie, R. S. (1998). The missing link? Understanding the relationship<br />
between individual and organisational learning. International Journal <strong>of</strong> Training and<br />
Development, 2(4), 288-<strong>29</strong>9.<br />
McSherry, M., & Taylor, P. (1994). Supervisory support for the transfer <strong>of</strong> team-building<br />
training. The International Journal <strong>of</strong> Human Resource Management, 5(1), 107-119.<br />
Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace: Theory, research, and<br />
application. Thousand Oaks, CA: Sage.<br />
Mowday, R. T. (1998). Reflections on the study and relevance <strong>of</strong> organizational commitment.<br />
Human Resource Management Review, 8(4), 387-401.<br />
Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The measurement <strong>of</strong> organizational<br />
commitment. Journal <strong>of</strong> Vocational Behavior, 14, 224-247.<br />
Naquin, S. S., & Holton III, E. F. (2002). The effects <strong>of</strong> personality, affectivity, and work<br />
commitment on motivation to improve work through learning. Human Resource<br />
Development Quarterly, 13(4), 357-376.<br />
Newstrom, J. W. (1986). Leveraging management development through the management <strong>of</strong><br />
transfer. Journal <strong>of</strong> Management Development, 5(5), 33-45.<br />
Nijman, D.-J. J. M., Nijh<strong>of</strong>, W. J., Wognum, A. A. M., & Veldkamp, B. P. (2006). Exploring<br />
differential effects <strong>of</strong> supervisor support on transfer <strong>of</strong> training. Journal <strong>of</strong> European<br />
Industrial Training, 30(7), 5<strong>29</strong>-549.<br />
Noe, R. A. (2000). Invited reaction: Development <strong>of</strong> a generalized learning transfer system<br />
inventory. Human Resource Development Quarterly, 11(4), 361-365.<br />
Orpen, C. (1999). The influence <strong>of</strong> the training environment on trainee motivation and<br />
perceived training quality. International Journal <strong>of</strong> Training and Development, 3(1),<br />
34-43.<br />
Paxson, M. C., Dillman, D. A., & Tarnai, J. (Eds.). (1995). Improving response to business<br />
mail surveys. New York: John Wiley & Sons.<br />
Pfeffer, J. (1998). The human equation: Building pr<strong>of</strong>its by putting people first. Boston:<br />
Harvard University Press.<br />
Reichers, A. E., Wanous, J. P., & Austin, J. T. (1997). Understanding and managing<br />
cynicism about organizational change. Academy <strong>of</strong> Management Executive, 11(1), 48-<br />
59.<br />
72
Richman-Hirsch, W. L. (2001). Posttraining interventions to enhance transfer: The<br />
moderating effects <strong>of</strong> work environments. Human Resource Development Quarterly,<br />
12(2), 105-120.<br />
Rouiller, J. Z., & Goldstein, I. L. (1993). The relationship between organizational transfer<br />
climate and positive transfer <strong>of</strong> training. Human resource Development Quarterly,<br />
4(4), 377-390.<br />
Saks, A. M. A., & Belcourt, M. M. (2006). An investigation <strong>of</strong> training activities and transfer<br />
<strong>of</strong> training in organizations. Human resource management, 45(4), 6<strong>29</strong>.<br />
Sharma, S., Durand, R. M., & Gur-Arie, O. (1981). Identification and analysis <strong>of</strong> moderator<br />
variables. Journal <strong>of</strong> Marketing Research, 18(3), <strong>29</strong>1-300.<br />
Smith-Jentsch, K. A., Salas, E., & Brannick, M. T. (2001). To transfer or not to transfer?<br />
Investigating the combined effects <strong>of</strong> trainee characteristics, team leader support, and<br />
team climate. Journal <strong>of</strong> Applied Psychology, 86(2), 279-<strong>29</strong>2.<br />
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York:<br />
HarperCollins College Publishers.<br />
Tesluk, P. E., Farr, J. L., Mathieu, J. E., & Vance, R. J. (1995). Generalization <strong>of</strong> employee<br />
involvement training to the job setting: Individual and situational effects. Personnel<br />
Psychology, 48(3), 607-632.<br />
Thayer, P. W., & Teachout, M. S. (1995). A climate for transfer model (AL/HR-TP-1995-<br />
0035): Brooks Air Force Base TX: Technical Training Research <strong>Division</strong>, Armstrong<br />
Laboratory.<br />
Tracey, J. B. (2005). Construct validity <strong>of</strong> a general training climate scale. Organizational<br />
research methods, 8(4), 353-374.<br />
Tracey, J. B., Hinkin, T. R., Tannenbaum, S., & Mathieu, J. E. (2001). The influence <strong>of</strong><br />
individual characteristics and the work environment on varying levels <strong>of</strong> training<br />
outcomes. Human Resource Development Quarterly, 12(1), 5-23.<br />
Tracey, J. B., Tannenbaum, S. I., & Kavanagh, M. J. (1995). Applying trained skills on the<br />
job: The importance <strong>of</strong> the work environment. Journal <strong>of</strong> Applied Psychology, 80(2),<br />
239-252.<br />
Tziner, A., Fisher, M., Senior, T., & Weisberg, J. (2007). Effects <strong>of</strong> trainee characteristics on<br />
training effectiveness. International Journal <strong>of</strong> Selection and Assessment, 15(2), 167-<br />
174.<br />
Vardon, S. (1998). Restructuring in the Australian government: Creating a responsive service<br />
delivery agency. Speech to senior Canadian government <strong>of</strong>ficers.<br />
Wanous, J. P., Reichers, A. E., & Austin, J. T. (2000). Cynicism about organizational change.<br />
Group and Organization Management, 25(2), 132-153.<br />
Washington, C. L. (2002). The relationships among learning transfer climate, transfer selfefficacy,<br />
goal commitment, and sales performance in an organization undergoing<br />
planned change. Unpublished PhD, The Ohio State University, Ohio.<br />
73
ASAC <strong>2008</strong><br />
Halifax, Ontario<br />
Nita Chhinzer<br />
College <strong>of</strong> Management and Economics<br />
University <strong>of</strong> Guelph<br />
Chhinzer@uoguelph.ca<br />
416.857.3570<br />
DISAGGREGATING UNEMPLOYMENT: JOB LEAVERS, LOSERS AND LAYOFFS<br />
This research suggests the need to disaggregate unemployed persons<br />
into job leaver, job loser and job lay<strong>of</strong>f categorizations.<br />
Multinominal logit regression on Labour Force Survey data<br />
(n=38,546) suggests that demographic, human capital and work<br />
related variables account for almost a third <strong>of</strong> the variance in<br />
likelihood to fall into the disaggregated unemployment categories.<br />
Khaldoun Ababneh<br />
Faculty <strong>of</strong> Business<br />
McMaster University<br />
ababneki@mcmaster.ca<br />
905-525-9140 x. 26356<br />
There has been much written about pr<strong>of</strong>iling the labour force in North America (i.e. Aaronson et<br />
al., 2006; Cappelli, 2005; Juhn & Potter, 2006; i.e. Panagiotidis & Pelloni, 2007; Van Horn, 2006).<br />
Significantly less has been written about the characteristics <strong>of</strong> the employed. Of this, the majority <strong>of</strong><br />
research treats unemployed persons as a homogenous group (most recently: Aaronson et al., 2006;<br />
Autor et al., 2006; most recently: Riley & Young, 2007). Traditionally, researchers have modelled job<br />
movers into one undifferentiated category when modelling the job mobility decision (Borjas, 1981;<br />
Kidd, 1991; Simpson, 1990). The research that attempts to understand micro level differences in<br />
unemployed persons focuses almost exclusively on economic data and is significantly dated (Borjas,<br />
1981; Henry S Farber, 1997; Henry S. Farber & Hall, 1993; Hamermesh, 1989; Hanisch, 1999; Kidd,<br />
1994; Picot et al., 1998). Additionally, these studies assess antecedents and outcomes <strong>of</strong><br />
unemployment using job leaver (quits) or job lay<strong>of</strong>f categories, with no consideration for the job loser<br />
(dismissals) category. This paper updates and modifies existing research on individual level<br />
differences based on categorizations <strong>of</strong> unemployment in three significant ways.<br />
First, this research treats the unemployed group as non-homogenous. Rather than comparing<br />
employed with unemployed persons, we identify individual level factors correlated with the<br />
likelihood to belong to the job leaver (quit), job loser (dismissal or permanent lay<strong>of</strong>f) or job lay<strong>of</strong>f<br />
category (temporary lay<strong>of</strong>f) by assessing Labour Force Survey (LFS) data collected over a five year<br />
period. We use multinomial logit regression analysis on a sample size <strong>of</strong> 38,546 individuals to begin<br />
to analyse unemployed persons based on causes <strong>of</strong> unemployment.<br />
Second, we examine the Canadian labour force. To date, empirical evidence from Canada on the<br />
distinction <strong>of</strong> types <strong>of</strong> unemployment has been significantly hindered by the lack <strong>of</strong> data. Thus, only<br />
a limited number <strong>of</strong> studies are available exploring distinct categories <strong>of</strong> unemployment in Canada<br />
(Kidd, 1994; McLaughlin, 1991; Picot et al., 1998). The Canadian labour force is significantly<br />
different than the United States labour force. Levels <strong>of</strong> unemployment, employment legislation and<br />
social services to support the unemployed are also significantly different. We extend beyond existing<br />
research to focus on disaggregating unemployment within the Canadian realm.<br />
74
Third, we use demographic, human capital and work related variables to support the notion that<br />
unemployment via job leaver, job lay<strong>of</strong>f and job losers categories results in different pr<strong>of</strong>iles <strong>of</strong><br />
unemployed persons, when compared with a control group <strong>of</strong> job stayers (individuals who did not<br />
change jobs in the survey period). This extends existing research which is focused almost exclusively<br />
on the economic perspectives <strong>of</strong> unemployment.<br />
The primary contribution <strong>of</strong> this paper is to suggest that we must disaggregate the unemployed<br />
group to help us understand the pattern <strong>of</strong> unemployment in a more comprehensive and pragmatic<br />
way. Through pr<strong>of</strong>iling unemployment based on the job leavers, job losers and job lay<strong>of</strong>f categories,<br />
we can understand patterns <strong>of</strong> unemployment, as well as the differences in the composition and<br />
correlations associated with each form <strong>of</strong> job loss.<br />
Next, we present core definitions and provide a brief review <strong>of</strong> existing literature. Following that,<br />
we present a discussion <strong>of</strong> variables used in this study. Due to the exploratory nature <strong>of</strong> this research,<br />
a limited number <strong>of</strong> variables were used in our analysis. Although additional variables may be<br />
influential in characterisizing job leavers, losers and lay<strong>of</strong>fs, these specific variables were<br />
theoretically applicable and available in the LFS. The results section includes evaluation <strong>of</strong> labour<br />
market transition rates and multinomial logit estimates <strong>of</strong> the likelihood to belong to one <strong>of</strong> the three<br />
categories <strong>of</strong> unemployment on two levels; comparisons against the control group <strong>of</strong> job stayers and<br />
intergroup pair-wise comparisons. We end with discussions and limitation <strong>of</strong> the study.<br />
Disaggregating the Unemployed Group: Definitions<br />
Some individuals chose to voluntarily enter unemployment, through turnover in the form <strong>of</strong> a quit<br />
or a resignation. For the purpose <strong>of</strong> this study, the job leavers group is defined as individuals who<br />
were employed within the last twelve months, but were not employed at the time <strong>of</strong> the survey due to<br />
voluntary reasons (e.g. quit) (Statistics Canada, 2007). Comparatively, the job losers group is defined<br />
as individuals who were employed within the last twelve months, but employment was permanently<br />
terminated at the time <strong>of</strong> the survey due to involuntary reasons (e.g. dismissal or permanent lay<strong>of</strong>f)<br />
(Statistics Canada, 2007). The last group is the job lay<strong>of</strong>f group representing individuals who were<br />
temporarily laid <strong>of</strong>f and have a potential for recall (Statistics Canada, 2007). It is important to note<br />
that the job lay<strong>of</strong>f group experience temporary unemployment and may be recalled by the firm,<br />
whereas job loser’s employment is permanently terminated.<br />
As an independent group <strong>of</strong> unemployed persons, the job leavers group has been studied<br />
comprehensively in the past. A number <strong>of</strong> models propose theoretical antecedents <strong>of</strong> an employee’s<br />
intention to leave, including factors such as work and non work related variables, perceived<br />
alternatives and shock (Carnicer et al., 2004; P. W. Hom & Griffeth, 1995; Lee et al., 1996; Mobley,<br />
1982; Porter & Steers, 1973; Steers & Mowday, 1981). Hundreds <strong>of</strong> studies attempt to identify<br />
antecedents <strong>of</strong> voluntary turnover, resulting in many meta analyses on this topic (Cohen & Cohen,<br />
1983; Cotton & Tuttle, 1986; R. W. Griffeth et al., 2000; P. Hom et al., 1992; P. W. Hom & Griffeth,<br />
1995; Steel & Ovaille, 1984; Tett & Meyer, 1993).<br />
The most recent meta-analysis identified a significant number <strong>of</strong> variables as direct and indirect<br />
predictors <strong>of</strong> voluntary turnover in the form <strong>of</strong> quits (R. W. Griffeth et al., 2000). Eliminating<br />
variables about cognitions and behaviours about the withdrawal process (e.g. intention to quit, job<br />
search efforts) individual level variables found to have the highest correlation with voluntary job loss<br />
were organizational commitment (r = -0.27), role clarity (r = -0.24), tenure (r = -0.23), role conflict (r<br />
= 0.22), overall job sat (r = 0.22) and absenteeism (r = -0.21). Additionally, age and marital status are<br />
negatively correlated with voluntary turnover (r = -0.23 and -0.05 respectively) while education is<br />
75
positively correlated with voluntary turnover (r = 0.06) (for all correlations p-value < 0.05). The<br />
research ascertains that gender is not significantly correlated with voluntary turnover (r = 0.03).<br />
Existing research in turnover generally limits the analysis <strong>of</strong> the employed versus unemployed<br />
groups using the job leavers (quit) group to homogeneous represent unemployed person for three<br />
main reasons. First, in the past most turnover was voluntary (Byrt, 1957; Marsh & Mannari, 1977),<br />
therefore the study <strong>of</strong> involuntary turnover may have been perceived as non-imperative. Second,<br />
theory was easier to form when turnover could be treated as homogenous. It would be hard to explain<br />
quits, dismissals and lay<strong>of</strong>fs in the same theory because they may have different determinants.<br />
Focusing on the voluntary turnover category provided a more homogeneous group to research. Third,<br />
involuntary turnover is organizationally initiated, so it was assumed that the selection procedure was<br />
solely performance based. This lead to the belief that “involuntary exits are desirable because<br />
employers would not want to keep poor performers/excess manpower” (R. Griffeth & Hom, 2001: pg<br />
4).<br />
Flows into unemployment, specifically job leavers, job losers and job lay<strong>of</strong>fs receive limited<br />
attention in U.S. literature, and remain focused on economic indicators <strong>of</strong> each category (Borjas, 1981;<br />
Henry S Farber, 1997; Henry S. Farber, 1999; Henry S. Farber & Hall, 1993; Hamermesh, 1989).<br />
With a common sample <strong>of</strong> a series from the Displaced Workers Survey in the 1980’s and 1990’s, the<br />
focus <strong>of</strong> these studies is the antecedents and outcomes <strong>of</strong> unemployment via the job loser and job<br />
lay<strong>of</strong>f categories, without consideration <strong>of</strong> the job leaver category.<br />
Due to a lack <strong>of</strong> data sources on unemployment in Canada, only Kidd (1994) and Picot et al.<br />
(1998) explore the issue <strong>of</strong> categorizing unemployed persons. Kidd (1994) examines whether a<br />
meaningful difference exists between quits and lay<strong>of</strong>fs. The results provide support that job leavers<br />
and job lay<strong>of</strong>fs are a non-homogenous group. The study collected Canadian labour market data,<br />
including reasons for unemployment in 1986 and 1987. Kidd’s econometric model assumed<br />
membership into one <strong>of</strong> four mutually exclusive categories: quit (n=617), lay<strong>of</strong>f (n=261), job stayers<br />
(n=397), and a residual (other mover category). The distinction between a quit and a lay<strong>of</strong>f was<br />
dependent on whether unemployment was voluntary; assuming that a quit was voluntary and a lay<strong>of</strong>f<br />
was involuntary. Based on Kidd’s economic perspective, wage equation estimates were developed<br />
(controlling for marital status, age, education, industry, occupation and province <strong>of</strong> residence) and<br />
dependent on if the job separation was voluntary versus involuntary (quit vs. lay<strong>of</strong>f).<br />
The findings <strong>of</strong> Kidd’s empirical research support the fact that when the voluntary versus<br />
involuntary distinction is made, the party who wishes to initiate the job separation gains monetary<br />
benefits. After job separation, those who quit (voluntarily separated) earned an average salary 18%<br />
higher than job stayers. Likewise, those who were laid <strong>of</strong>f (involuntarily separated) earned an average<br />
salary 30% less than job stayers. This suggests that the economic benefit during an involuntary<br />
separation is positive for the employer, since the employee’s pay level was not representative <strong>of</strong><br />
productivity in a fixed market. However, an employee whose market value was higher than the actual<br />
wage set by the employer gains the economic benefit <strong>of</strong> voluntary separation. Therefore, there are<br />
opposing antecedents to quits versus lay<strong>of</strong>fs.<br />
There are a number <strong>of</strong> limitations <strong>of</strong> this study. Kidd’s study utilized data from the Labour<br />
Market Activity Survey (LMAS) <strong>of</strong> 1986-1987, therefore the data and findings are significantly dated.<br />
The sample accessed was full time, paid male workers between the ages <strong>of</strong> 16-64. Therefore, the<br />
study provides only a partial view <strong>of</strong> unemployment due to limitations in the data collected. Also, this<br />
research compared only quits versus lay<strong>of</strong>fs. There is no awareness <strong>of</strong> those who entered<br />
unemployment through dismissals and no differentiation <strong>of</strong> permanent versus temporary lay<strong>of</strong>fs.<br />
76
In a second Canadian study, Picot, Lin and Pyper (1998) used a random sample <strong>of</strong> all Canadian<br />
workers to study lay<strong>of</strong>f trends. The information was extracted from the Longitudinal Worker File,<br />
which is one component <strong>of</strong> the Labour Force Survey. The purpose <strong>of</strong> Picot and her colleagues study<br />
was three fold; 1) to examine the underlying causes <strong>of</strong> lay<strong>of</strong>fs, 2) to empirically prove an increase in<br />
lay<strong>of</strong>fs in the 1990’s and 3) to examine if lay<strong>of</strong>fs are continuous events. Picot et al. (1998) provide<br />
evidence <strong>of</strong> a number <strong>of</strong> correlations between various factors (such as age, gender, and skill level)<br />
and likelihood to experience. Lay<strong>of</strong>f victims are older (over 55 years <strong>of</strong> age at time <strong>of</strong> lay<strong>of</strong>f) when<br />
the lay<strong>of</strong>f was a single event. In continuous lay<strong>of</strong>f situations (where the individual is laid <strong>of</strong>f from 5<br />
or more companies within a 10 year span), younger individuals are most likely to be the subjects<br />
(between 25 and 34 years <strong>of</strong> age). Annual earnings, education and skill level are found to be the most<br />
influential determinants <strong>of</strong> the likelihood <strong>of</strong> experiencing a lay <strong>of</strong>f. However, this study does not<br />
include comparisons <strong>of</strong> lay<strong>of</strong>fs, quits and dismissals.<br />
A crucial finding in the Picot et al. study was that from 1978-1993 a 1% change in unemployment<br />
was associated with a 0.89% decrease in quit rates, a 0.61% increase in use <strong>of</strong> temporary lay<strong>of</strong>fs and<br />
a 0.38% decrease in hiring rates. There was also a small effect on use <strong>of</strong> permanent lay<strong>of</strong>fs with a<br />
0.34% increase in permanent lay<strong>of</strong>fs for every 1% increase in unemployment. One finding that was<br />
evident in the data, but not explicitly stated by the authors is that quit rates and lay<strong>of</strong>f rates<br />
consistently reacted in opposite directions. This, further support for creating a distinction between job<br />
leavers, job losers and job lay<strong>of</strong>fs categories is secured.<br />
Given the exploratory perspective adopted in the research, only a limited number <strong>of</strong> variables can<br />
be used to demonstrate that disaggregating unemployment will enhance our understand <strong>of</strong> the<br />
unemployment phenomena. To do so, we select and rationalize our choice <strong>of</strong> seven variables in the<br />
next section. Through analysis <strong>of</strong> these variables, we hope to secure support for examining<br />
unemployment within the three categories above.<br />
Theoretical Perspectives in Variable Selection<br />
The variables selected for the data analysis met two minimal standards. One, the variables must<br />
have a theoretical link to studies <strong>of</strong> unemployment. Two, the variables must be measured in the LFS<br />
for unemployed individuals.<br />
Human Capital Theory<br />
The first two variables are embedded in human capital theory; education and tenure. Human<br />
capital refers to the employee’s tacit knowledge <strong>of</strong> firm specific information, such as the knowledge,<br />
skills, abilities and other attributes <strong>of</strong> individual employees (Becker, 1963). Firm specific skills such<br />
as familiarity with products, procedures and technical characteristics <strong>of</strong> a firm cannot be transferred<br />
among firms (Hitt & Ireland, 2002; Schultze, 1999).<br />
Education. Education provides a general training program to help introduce individuals to broad<br />
concepts, industries and ideologies. Generally, the higher an individual’s education, the less firm<br />
specific their skill set, therefore their knowledge and skills are more transferable. This also translates<br />
into a lower level <strong>of</strong> firm specific human capital, which can affect organizational decisions regarding<br />
employee exits. In Griffeth, Hom and Gaertner’s (2000) meta-analysis on antecedents <strong>of</strong> turnover, a<br />
statistically significant positive relationship between education and voluntary turnover (quits) was<br />
noted. Although a correlation between education and turnover is established in existing research, no<br />
study explicitly examines the relationship between education and different flows into unemployment,<br />
as per this study.<br />
77
Individuals with lower levels <strong>of</strong> education may be less mobile in the workforce due to their<br />
lack <strong>of</strong> generalizable skills. These individuals may perceive their chances for re-employment to be<br />
low, due to high requirements for firm specific training upon reemployment. Likewise, individuals<br />
with low education may have experienced high levels <strong>of</strong> firm specific knowledge and training,<br />
therefore from a company perspective may not be candidates for lay<strong>of</strong>fs during a downsizing event.<br />
Education is traditionally segmented in terms <strong>of</strong> four classifications: ‘high school or less’, ‘some<br />
post secondary education’, ‘completed post secondary certificate or diploma’, and ‘university and<br />
above’ (e.g. Industry Canada, Canada’s Office <strong>of</strong> Consumer Affairs). According to the Labour Force<br />
Survey in 2004, 41% <strong>of</strong> the labour force had completed high school or less as their highest level <strong>of</strong><br />
educational attainment. A small percent had some post secondary education as their highest level <strong>of</strong><br />
educational attainment (7.4%). An additional 36.9% <strong>of</strong> the labour force completed a post secondary<br />
certificate or diploma, while 14.7% <strong>of</strong> the labour force possessed a university degree or higher<br />
education.<br />
Tenure. Tenure is defined as the total time an individual is employed by the company. Generally,<br />
the higher an individual’s experience with the organization, the more firm specific their skill set.<br />
Traditionally, job tenure provided some protection from displacements, but the benefits <strong>of</strong> high tenure<br />
are shrinking (Fallick, 1996; Henry S Farber, 1997; Seitchik, 1991). Kidd (1995) and Farber (1993)<br />
provide evidence that tenure is influential in predicting individual unemployment. Individuals who<br />
remain employed have twice as much tenure than those who quit, and individual who quit generally<br />
have slightly longer tenure than lay<strong>of</strong>fs. Tenure groups used in this research align with Kidd’s<br />
analysis (1994).<br />
In addition to Human Capital factors affecting this variable, longer tenure with a company also<br />
suggests that the individual employee may have more invested with the company such as investing in<br />
a house near work, work friendships extending to personal life and an attachment to the job. These<br />
investments are jeopardized when an employee with more experience loses employment, thus it is<br />
possible that high tenure is desirable for employment. Therefore, we would expect the unemployed<br />
groups to have lower tenure regardless <strong>of</strong> means <strong>of</strong> unemployment.<br />
Work Related Variables<br />
The next two variables selected are work related variables; industry and occupation. Individual<br />
work related variables include individual job or organizational factors that may be directly related to<br />
turnover behaviour, such as industry and occupation (Mobley, 1982).<br />
Industry. There has been an ongoing debate regarding the influence <strong>of</strong> industry on turnover. In<br />
the United States, chances <strong>of</strong> displacement increases significantly if the industry an individual is<br />
employed in is doing poorly overall (Fallick, 1996). Aligned with this perspective, industry<br />
differences are proven to have a significant and negative correlation with the likelihood for job loss in<br />
Canada (Kidd, 1994). In contrast, Picot et al (1998) provide evidence that the predictive lay<strong>of</strong>fs are<br />
more <strong>of</strong> an individual company decision than an industry specific decision in Canada. They suggest<br />
that companies in the same industry, facing the same economic concerns can choose to lay<strong>of</strong>f, freeze<br />
hiring, use technology to decrease costs, launch new products, or cut back services as possible<br />
competing responses to industry downturn. Therefore, turnover cannot be explained by cyclical<br />
variations in demand or industry level factors such as growth or decline. This debate remains<br />
unresolved; therefore the inclusion <strong>of</strong> industry as a variable in this study is justifiable.<br />
Industry trends can be explored using the goods versus services industry difference, as per<br />
previous unemployment research. Farber (1993) found that from 1982-1985, job loss was<br />
78
concentrated in the goods (manufacturing) industry, but in 1986-1991, the job loss was concentrated<br />
in the services industry. As manufacturing jobs become increasingly scarce in the Canadian<br />
employment outlook, employees may be less likely to quit due to a perceived lack <strong>of</strong> alternative<br />
employment opportunities in the manufacturing industry. Additionally, lay<strong>of</strong>fs may be more likely to<br />
occur in industries on the decline, such as the manufacturing industry. In contrast, service sector jobs<br />
are increasing in the Canadian economy, therefore, employers may be less likely to dismiss<br />
employees, due to a perceived labour shortage in the service industries.<br />
Occupation. The occupation variable may be directly related to flows into unemployment. Over<br />
time, the number and types <strong>of</strong> occupational available to the labour force change due to technological<br />
advancements, globalization, economic growth, demographics and consumer behaviour. While there<br />
are thousands <strong>of</strong> possible occupations, we categorized jobs according to the North American Industry<br />
Classification System (NAICS). Most users <strong>of</strong> this type <strong>of</strong> information create clusters <strong>of</strong> similar<br />
occupations for the use and interpretation <strong>of</strong> the results (Fallick, 1996; Henry S Farber, 1997; Henry S.<br />
Farber & Hall, 1993; Kidd, 1994).<br />
Similarly, we use five main categories. First, ‘management, business, finance and administrative’<br />
occupations create a bundle <strong>of</strong> jobs that are directly involved in providing services and guidance to<br />
employees at multiple levels in the organization. This group <strong>of</strong> occupations does not directly<br />
contribute to revenue generation or production, but is part <strong>of</strong> the core organizational structure.<br />
Second, occupations ‘unique to primary industries, processing and manufacturing’ create a cluster <strong>of</strong><br />
jobs that directly impact an organizations production and are traditionally considered blue collar jobs.<br />
Next, ‘trades, transportation, equipment operators and related’ occupations focus on supply chain<br />
and logistical jobs in Canada. These account for almost a quarter <strong>of</strong> all occupations in the sample.<br />
Fourth, ‘sales and service’ occupations include jobs which require brokering <strong>of</strong> products/services or<br />
actual delivery <strong>of</strong> services (e.g. call centres). Lastly, a catch all group <strong>of</strong> ‘others’ includes jobs that<br />
the labour force survey could not clearly define under one occupational title.<br />
Projections for 2009 suggest that the services industry will continue to grow, while growth will<br />
stagnate, then decline in primary industries in Canada (Government <strong>of</strong> Canada and the Ontario<br />
Ministry <strong>of</strong> Training, 2007). In Ontario, 24% <strong>of</strong> job growth between 2004 and 2009 is projected to<br />
come from pr<strong>of</strong>essional jobs while manufacturing will be responsible for an additional 10% <strong>of</strong> job<br />
growth (Government <strong>of</strong> Canada and the Ontario Ministry <strong>of</strong> Training, 2007). Given that the labour<br />
market and perceived alternative employment opportunities differ by occupation, it is perceivable that<br />
individuals in occupations with expected growth in Canada may be more likely to be in the job leaver<br />
category. These individuals may perceive that their occupations are in high demand, therefore may<br />
choose to exit the firm to gain more desireable employment terms or compensation. Individuals in<br />
occupations with minimal growth forecasts may be more likely to fall into the job loser and job lay<strong>of</strong>f<br />
category, since some <strong>of</strong> these jobs may become obsolete. As well, these individuals would be less<br />
likely to turnover via quits (job leaver) since the labour market may be flooded with qualified<br />
candidates, therefore reducing perceived chances for reemployment. Unemployment trends may vary<br />
by occupation, therefore this was included in the analysis.<br />
Demographics<br />
The remaining three variables selected are demographic variables: age, marital status and gender.<br />
Demographic variables refer to individual attributes that employees bring to work. These traits<br />
remain stable across jobs, employers and contexts and are associated with the individual employee in<br />
all settings. Personal characteristic may influence the decision to quit since they help the employee<br />
predict their internal versus external advantages and disadvantages. These demographic conditions<br />
79
can dictate the employee’s perception <strong>of</strong> opportunities they have outside <strong>of</strong> the company, as well as<br />
likelihood to voluntarily leave, since most <strong>of</strong> these attributes are non-changeable by the employee.<br />
In contrast, a combination <strong>of</strong> federal, state and provincial laws has limited management’s ability<br />
to decide whom to separate from their jobs. Although management should be aware that using<br />
demographics as a determinant in practice (dismissals or lay<strong>of</strong>fs) is legal only under exceptional<br />
circumstances (e.g. Bona Fida Occupational Requirement), cases <strong>of</strong> lay<strong>of</strong>fs and dismissals are being<br />
met with increased legal resistant (Balkin, 1992). As a result, it is possible that the group <strong>of</strong> job<br />
leavers may be demographically different than the job stayers group, but there should be no<br />
demographic differences between the job stayers, job losers and job lay<strong>of</strong>fs groups.<br />
Methods<br />
Data used in this analysis was collected from the Labour Force Survey (LFS) . Started in 1945<br />
and maintained by a body <strong>of</strong> the Canadian Federal Government (Statistics Canada), the LFS provides<br />
the only source <strong>of</strong> monthly labour force data in Canada. Responses are self-reported, therefore the<br />
analysis is one <strong>of</strong> self reported measures <strong>of</strong> flows into unemployment.<br />
Sampling 2% <strong>of</strong> the Canadian population a month, the LFS clusters individuals into three main<br />
categories: employed, unemployed and not in the labour force. This study uses the group <strong>of</strong> employed<br />
persons (or job stayers) as the base or control group. When the complete sample is assessed, the<br />
control group <strong>of</strong> employed persons overwhelms the unemployed persons group by a ratio <strong>of</strong> more<br />
than 30:1. A correction for this is required, since when the proportion <strong>of</strong> employed to unemployed<br />
(or vice versa) diverges from 50%, variance and correlations are attenuated (Pedhazur, 1982). A<br />
randomly selected group <strong>of</strong> employed persons equal in count to the group <strong>of</strong> unemployed persons<br />
allows for true unbiased assessment <strong>of</strong> correlates and variance <strong>of</strong> the flows into unemployment.<br />
Additionally, individuals who are not in the labour force (people unwilling and unable to work)<br />
represent a group that is neither employed or unemployed. Therefore, individuals not in the labour<br />
force were not included in the analysis. A mix <strong>of</strong> questions about the unemployment decision and<br />
activity prior to unemployment aid Statistics Canada in their classifying individuals into one <strong>of</strong> three<br />
unemployed groups: job leavers (quits/resignations), job losers (dismissals/permanent lay<strong>of</strong>f) and job<br />
lay<strong>of</strong>fs (temporary lay<strong>of</strong>fs).<br />
Participants in the survey are sampled monthly and form a six month panel. The response rate<br />
averages 95% a month. In the rare case that a participant drops out <strong>of</strong> the panel, a weight adjustment<br />
is applied to the account and the individual is not replaced. The LFS also utilizes a seasonal<br />
adjustment for institutional events like vacation, holidays and climate events. Seasonal variations<br />
from almost 1,300 participants are adjusted to prevent seasonal factors from effecting<br />
employment/unemployment analysis.<br />
Since significant revisions were made to LFS questionnaire in 1976 and 1997, the period <strong>of</strong> study<br />
is limited to 2000-2004. This ensures reliability <strong>of</strong> the results, by guaranteeing survey questions were<br />
consistent. We pooled monthly data to develop an annual measure <strong>of</strong> flows into unemployment. More<br />
specifically, given that survey participants were part <strong>of</strong> a 6 month panel, we merged information from<br />
the March and September data files to create our database. The risk <strong>of</strong> sampling the same individual<br />
in the labour force twice is completely eliminated, through forcing six months <strong>of</strong> difference in the<br />
sample, increasing the validity <strong>of</strong> the results.<br />
A total <strong>of</strong> 19,273 unique individuals experienced unemployment during the survey period. An<br />
equal randomly selected number <strong>of</strong> employed persons from as the control group. Of the unemployed<br />
80
persons group, 3,542 participants experienced temporary laid <strong>of</strong>fs, falling into the job lay<strong>of</strong>f category.<br />
An additional 12,397 persons experienced some form <strong>of</strong> permanent involuntary flow into<br />
unemployment (e.g. dismissal or permanent lay<strong>of</strong>f) falling into the job loser category. The remaining<br />
3,334 individuals initiated the flow into unemployment by resigning or quitting, thereby fall into the<br />
job leaver category.<br />
Multinomial Logit Regression Analysis<br />
Results<br />
Multinomial logit regression (MNL) analysis is conducted here to estimate the probability (or odd<br />
ratios) <strong>of</strong> an individual flowing into the job leaver, job loser, or job lay<strong>of</strong>f category relative to remain<br />
in the job stayer category by using seven predictors (i.e. education, tenure, industry, occupation, age,<br />
gender, and martial status). To conduct the MNL analysis, we set the coefficients <strong>of</strong> the job stayer<br />
category to zero so that the other estimated parameters could be interpreted relative to this reference<br />
group (i.e. job stayer category).<br />
Using a chi-squared test to evaluate fit (χ² ( 54 , N = 38,546 ) = 12119.754 ), the model including<br />
seven predictors against the null model (constant-only model) is statistically significant. This means<br />
that the model with the seven predictors as a set is outperforming the null model in predicting<br />
individual unemployment status. Thus, we reject the null hypothesis that all coefficients in our model<br />
are jointly zero. The likelihood ratio-test for each <strong>of</strong> the seven predictors shows that each predictor <strong>of</strong><br />
our model is statistically significant (p-value < .001) in influencing the likelihood that an individual<br />
will flow into a specific employment status.<br />
Table 1 provides an assessment <strong>of</strong> the over or under representation <strong>of</strong> each variable based on the<br />
corresponding flow into unemployment, compared to the control group (job stayers). For example,<br />
there are 18.7% more individuals with education levels <strong>of</strong> `high school or less` in the job lay<strong>of</strong>f group<br />
than in the job stayer group. In contrast, there are 15.1% less individuals with `high school or less`<br />
education in the job leaver group than in the job stayer group.<br />
Table 1.<br />
Labour Market Transition Rates (percentage difference in population as compared to job<br />
stayers)<br />
Job<br />
Lay<strong>of</strong>f<br />
Job<br />
Loser Job Leaver<br />
Education<br />
High school or less 18.7% -1.3% -15.1%<br />
Some post secondary -12.3% -4.1% 30.1%<br />
Post secondary certificate or diploma -5.3% 1.7% -1.7%<br />
university and above -60.0% 2.0% 54.0%<br />
Tenure<br />
1 year -27.8% 11.2% -12.2%<br />
2 years -14.8% -2.6% 26.1%<br />
3-4 years 8.9% -8.9% 25.0%<br />
5-9 years 38.7% -13.5% 12.6%<br />
10 years + 75.7% -19.3% -8.6%<br />
Occupation<br />
Management, Business, Finance, and Administrative<br />
Occupations -50.0% 3.1% 43.8%<br />
Sales and Service Occupations -33.2% -4.5% 51.5%<br />
Trades, Transportation and Equipment Operators and<br />
Related Occupations 36.3% 0.3% -39.7%<br />
Occupations Unique to Primary Industries, Processing and 44.7% 0.5% -22.8%<br />
81
Manufacturing<br />
Other -46.7% 2.5% 38.5%<br />
Industry<br />
Goods 38.5% 0.7% -48.4%<br />
Services -32.1% -0.6% 40.4%<br />
Sex<br />
Male 7.8% 2.3% -17.1%<br />
Female -12.6% -3.7% 27.8%<br />
Age<br />
25-34 -15.0% -1.8% 22.9%<br />
35-44 2.1% 0.0% -2.1%<br />
45-54 6.4% 0.8% -10.5%<br />
55-64 15.3% 2.5% -24.6%<br />
Marital status<br />
Married 44.8% 28.9% 24.8%<br />
Single -56.7% -39.5% -37.6%<br />
Divorced/Widowed/Separated -14.3% 6.7% 14.3%<br />
The results in Table 1 suggest that job losers (dismissals) and job stayers are similar in proportion<br />
<strong>of</strong> educational experience, suggesting that education level is not a main variable in the decision<br />
function to dismiss an employee. In fact, we consistently find that the job stayer and job leavers<br />
groups are similar in representation by the majority <strong>of</strong> the variables analysed in this study, with the<br />
exception <strong>of</strong> tenure and marital status.<br />
There is a negative relationship between education level and likelihood to experience a temporary<br />
lay<strong>of</strong>f. Specifically, employees with high school or less education represent 18.7% more <strong>of</strong> the job<br />
lay<strong>of</strong>f group than the job stayer group. Comparatively, employees with university and above<br />
education represent almost 60% less <strong>of</strong> those experiencing a job lay<strong>of</strong>f that their proportion in the<br />
remaining labour force. This provides empirical support that education levels provide protection<br />
from lay<strong>of</strong>f, as theorized in human capital theory.<br />
Flow into unemployment is more likely to be employee initiated when an employee possessed<br />
higher levels <strong>of</strong> education. Specifically, employees with high level <strong>of</strong> education represent 54% more<br />
<strong>of</strong> the job leaver category than they represent in the labour force. Aligned with this, individuals with<br />
lower levels <strong>of</strong> education were 15% less likely to initiate job loss. Following human capital theory,<br />
those with transferable skill sets (gained from general training and education) are more prone to<br />
initiate job loss through participation in the job leaver category than the job stayers and those with<br />
lower levels <strong>of</strong> education.<br />
At first glance, the results <strong>of</strong> tenure do not clearly identify a relationship between experience with<br />
the firm and flows into unemployment. However, if year 1 is eliminated, the relationships become<br />
more evident. Employees with high tenure make up more than their fair share <strong>of</strong> the employees laid<br />
<strong>of</strong>f; almost doubling their representation in the 10 years plus group, as compared to the control group<br />
<strong>of</strong> job stayers. As tenure increases, so does the representation in the group <strong>of</strong> laid <strong>of</strong>f persons. The<br />
opposite situation hold true for the job leavers group. As tenure increases, representation in the job<br />
leavers group decreases. In contrast to patterns between job losers and job stayers outlined in the<br />
education variable, tenure appears to negatively influence likelihood to be dismissed. Surprisingly,<br />
employees in their first year <strong>of</strong> employment are less likely to quit or be laid <strong>of</strong>f, but slightly more<br />
likely to be dismissed, which is the opposite <strong>of</strong> the results for the remaining years.<br />
Tenure can also be viewed as a unique variable in that the group <strong>of</strong> job losers is somewhat<br />
affected by tenure when compared to job stayers. With the exception <strong>of</strong> tenure and marital status, the<br />
group <strong>of</strong> job losers was not proportionately different than the control group. The results suggest that<br />
dismissals and permanent lay<strong>of</strong>f likelihood decreases with tenure. In the first year, employees are<br />
most likely to fall into the job losers category (represented 11.2% more in the job losers group than<br />
82
the job stayers group), and tenure does provide protection from dismissals and permanent lay<strong>of</strong>fs in<br />
later years, when compared to the job stayers group (represented 19.3% less in the job losers group<br />
than the job stayers group).<br />
Management and business related occupations showed the largest range <strong>of</strong> change based on<br />
unemployment category. Individuals with these occupations has 50% less representation in the job<br />
lay<strong>of</strong>f group, but 43.8% more representation in the job leaver group than their portion <strong>of</strong> the job<br />
stayers group. A similar pattern is revealed with sales and service jobs. The opposite situation holds<br />
true for trades, primary industries and manufacturing in Canada. Therefore, individuals in more white<br />
collar jobs are significantly more likely to initiate job loss via quitting or resigning from their job,<br />
while those in traditional blue collar jobs are much more likely to experience a lay<strong>of</strong>f in Canada.<br />
A similar pattern is formed when we contrast the goods industry with the services industry.<br />
Individuals in the goods industry are significantly more likely to experience a job lay<strong>of</strong>f, while 48.4%<br />
less likely to voluntarily leave their job. The opposite relationship is found in the services industries.<br />
As hypothesized, perhaps the growth <strong>of</strong> the services industries and the perceived tightness in the<br />
services labour market is responsible for these differences. The results suggest that when the services<br />
versus manufacturing differentiation is made, trends <strong>of</strong> flows into unemployment significant change.<br />
Thus, this variable is valuable in differentiating unemployment in Canada.<br />
Females are disproportionately overrepresented in the job leaver category by almost 28% and<br />
underrepresented in the job lay<strong>of</strong>f category. There may be a number <strong>of</strong> factors associated with this<br />
such as age, non work related responsibilities, occupation and industry differences that can be<br />
attributed to gender. Future studies may be valuable in explaining the differences highlighted in the<br />
results for unemployment trends based on gender in Canada, as per Table 1.<br />
The age variable identifies an interesting trend. While the job losers group is almost identical in<br />
terms <strong>of</strong> representation to the job stayers group, as age increases, the chances <strong>of</strong> unemployment<br />
through job lay<strong>of</strong>fs increases and unemployment through voluntary job loss (job leavers) decreases.<br />
This suggests a positive relationship between age and likelihood to be laid <strong>of</strong>f, and a negative<br />
relationship between age and likelihood to quit or resign. This variable is important in that if the<br />
results were aggregated (comparing employed versus unemployed persons) no effect <strong>of</strong> age would be<br />
evident. However, when the age category is disaggregated, it is highly influential in determining<br />
types <strong>of</strong> unemployment in Canada.<br />
Married individuals are overrepresented (24.8% - 44.8%), while singles are underrepresented<br />
significantly (37.6% - 56.7%) in all categories <strong>of</strong> unemployment. This is one <strong>of</strong> two variables in<br />
which the job loser category is significantly different than the job stayers category. As well,<br />
individuals who were in a partnership that no longer exists due to death, divorce or separation are<br />
14.3% overrepresented in the job leaver category, and 14.3% underrepresented in the job lay<strong>of</strong>f<br />
category. This is another ideal example supporting the need to disaggregate unemployment data in<br />
Canada. The relationship between marital status and unemployment remains largely unexplored in<br />
research, but the results suggest a direct or indirect influence <strong>of</strong> marital status on quits, dismissals,<br />
permanent lay<strong>of</strong>fs and temporary lay<strong>of</strong>fs in the Canadian labour force.<br />
Table 2.<br />
Multinomial Logit Estimates <strong>of</strong> the Likelihood <strong>of</strong> Flows into Unemployment<br />
Job Lay<strong>of</strong>f<br />
Vs.<br />
Job Loser<br />
Vs. Stayers<br />
Job<br />
Leaver<br />
All movers<br />
Vs.<br />
83
Stayers Vs. Stayers leavers<br />
Education<br />
high school or less 1.145*** 0.494*** 0.246*** 0.512**<br />
(3.142) (1.639) (1.280) (1.668)<br />
some post secondary 0.782*** 0.259*** 0.285** 0.309**<br />
(2.186) (1.<strong>29</strong>6) (1.3<strong>29</strong>) (1.362)<br />
post secondary certificate or diploma 0.802*** 0..<strong>29</strong>8*** 0.057 .284**<br />
Tenure<br />
(2.230) (1.347) (1.059) (1.328)<br />
1 year 1.376*** 2.626*** 2.078*** 2.279**<br />
(3.956) (13.820) (7.992) (9.768)<br />
2 years 0.834** 1.664*** 1.508*** 1.454**<br />
(2.303) (5.280) (4.516) (4.280)<br />
3-4 years 0.771*** 1.<strong>29</strong>4*** 1.235*** 1.153**<br />
(2.162) (3.647) (3.438) (3.167)<br />
5-9 years 0.391*** 0.664*** 0.624*** 0.577**<br />
Occupation<br />
(1.478) (1.942) (1.865) (1.781)<br />
Occupations Unique to Primary Industry,<br />
1.440*** 0.449*** 0.083 0.581**<br />
Processing, Manufacturing (4.222) (1.568) (1.086) (1.789)<br />
Sales and Service Occupations<br />
0.680*** 0.128** 0.261*** 0.243**<br />
(1.975) (1.136) (1.<strong>29</strong>9) (1.275)<br />
Trades, Transportation and Equipment Operators 1.696*** 0.526*** 0.148* 0.682**<br />
and Related Occupations (5.454) (1.691) (1.160) (1.977)<br />
Others .244** -0.069 -0.193** -.052<br />
Industry<br />
(1.276) (.933) (.824) (.949)<br />
Goods 0.804*** 0.478*** -0.215** 0.436**<br />
Sex<br />
(2.234) (1.613) (0.807) (1.547)<br />
Male -0.270*** 0.098** -0.013 0.007<br />
Age<br />
(0.764) (1.103) (0.987) (1.007)<br />
25-34 -0.403*** -0.549*** 0.017 -0.407**<br />
(0.668) (0.577) (1.017) (.666)<br />
35-44 -0.213** -0.306*** 0.010 -0.232**<br />
(0.809) (0.736) (1.010) (.793)<br />
45-54 -0.137* -0.165*** 0.024 -0.127*<br />
Marital status<br />
(0.872) (0.848) (1.024) (.881)<br />
Single 0.109* 0.420*** 0.478** 0.375**<br />
(1.115) 1.522 (1.613) (1.455)<br />
Others 0.203*** 0.369*** 0.448 0.349<br />
(1.225) (1.447) (1.565) (1.418)<br />
The reference variables are: job stayer (employment status), university and above (education), 10+ years (tenure), Management, Business,<br />
Finance, and Administrative (occupation), services (industry), female (sex), 55-64 (age), and married (marital status). The coefficients for<br />
the reference groups are all zero.<br />
Table 2 presents more related information concerning the multinomial logit analysis (MNL). The<br />
first, second, and third columns <strong>of</strong> Table 2 report the coefficient estimates and the odds ratios<br />
(reported in parentheses) <strong>of</strong> our multinomial logit analysis. The first column presents the coefficient<br />
estimates and the odd ratios comparing job lay<strong>of</strong>fs with job stayers. The second column presents the<br />
coefficient estimates and the odd ratios comparing job losers with job stayers. The third column<br />
presents the coefficient estimates and the odd ratios comparing job leavers with job stayers. A<br />
positive coefficient indicates that a specific category <strong>of</strong> an independent variable increases the<br />
likelihood <strong>of</strong> being in a certain work status in comparison to the job stayers group (the reference<br />
group), while a negative coefficient indicates that a specific category (corresponding category or<br />
variable) <strong>of</strong> an independent variable decreases the likelihood <strong>of</strong> being in a certain work status in<br />
comparison to the job stayers group (the reference group). Asterisks identify the categories <strong>of</strong> the<br />
independent variables that have significant effects on the flows into unemployment based on our three<br />
categories. The probability (odds) ratios indicate the magnitude <strong>of</strong> the likelihood <strong>of</strong> belonging to a<br />
certain flow into unemployment relative to the job stayers group. A variable that increases the<br />
likelihood <strong>of</strong> being in a specific unemployed group relative to the job stayers group has probability<br />
ratio greater than one, while a variable that decreases the likelihood <strong>of</strong> being in a specific work status<br />
84
has probability ratio lower than one. Following is a presentation for the effects <strong>of</strong> each <strong>of</strong> the seven<br />
variables on the flows into unemployment.<br />
Education. Comparing job lay<strong>of</strong>f with job stayer groups, column 1 <strong>of</strong> Table 2 shows that the<br />
coefficient for the categories ‘high school or less’, ‘some post secondary’, and ‘post secondary<br />
certificate or diploma’ are positive and significant. This means that individuals who hold ‘high school<br />
or less’, ‘some post secondary’, and ‘post secondary certificate or diploma’ level <strong>of</strong> education are<br />
more likely than individuals who hold a level <strong>of</strong> education <strong>of</strong> ‘university and above’ (reference group)<br />
to experience temporary lay<strong>of</strong>fs. Specifically, Table 2 shows that the probability <strong>of</strong> being in a job<br />
lay<strong>of</strong>f status relative to the probability <strong>of</strong> being in a job stayer status is higher for individuals who<br />
hold ‘high school or less’ (3.142 times higher), ‘some post secondary’ (2.186 times higher), and ‘post<br />
secondary certificate or diploma’ (2.230 times higher) than for individuals who hold ‘university and<br />
above’ degree. Comparing the job loser group with job stayers, column 2 <strong>of</strong> Table 2 reports similar<br />
effects for the education categories as those reported above when comparing job lay<strong>of</strong>f with job<br />
stayers, but with different odds ratio. Comparing job leavers with job stayer, column 3 <strong>of</strong> Table 2 also<br />
reports similar result as those discussed above except that individuals who hold ‘post secondary<br />
certificate or diploma’ were not significant.<br />
Tenure. With regard to tenure, the results demonstrate that the coefficient for individuals who<br />
have been with the organization for ‘1 year’, ‘2 years’, ‘3 years’ and ‘5-9 years’ are more likely to be<br />
in job lay<strong>of</strong>f, job loser, or job leaver category than to be in the job stayer category, as compared to<br />
those who have been with the organization for ten years and above (i.e., the reference group for<br />
tenure). For example, individuals who have been with the organization for two years are 13.820 times<br />
more likely to flow into job losers’ status than individuals who have been with the organization for<br />
ten years or more. The results also indicate that the likelihood <strong>of</strong> flowing into job loser, lay<strong>of</strong>f, or<br />
leaver status diminish as individuals gain more tenure with an organization. For example, the odds<br />
ratios <strong>of</strong> flow into job lay<strong>of</strong>f status are highest, in descending order, for: (a) individuals who have one<br />
year tenure (3.956); (b) individuals who have two years (2.303), (c) individuals who have 3-4 years<br />
(2.162) and individuals how have 5-9 years (1.478).<br />
Occupation. The result from Table 2 show that the probabilities <strong>of</strong> a job lay<strong>of</strong>f relative to the<br />
probability <strong>of</strong> a job stayer status is higher for individuals who worked for ‘trades, transportation and<br />
related occupations’, ‘primary industry and manufacturing’, ‘sales and service occupations’, and the<br />
other occupations category (in that order) than for individuals in the reference group (i.e. the category<br />
‘management and administrative related’). This means that individuals who hold occupations in<br />
management and administrative related were less likely to be laid <strong>of</strong>f than individuals who are<br />
holding all the other type <strong>of</strong> occupations. Similar results are observed when comparing the probability<br />
for those <strong>of</strong> job losers to those <strong>of</strong> job stayers, however with different odd ratios.<br />
Industry. Individuals who worked in goods industry were significantly more likely to flow into<br />
job lay<strong>of</strong>f or job loser category than to remain in job stayer category, as compared to those in the<br />
service industry. However, individuals who worked in good industry are significantly less likely to<br />
flow into job leaver category than to remain in job stayer category, as compared to those in the<br />
service industry.<br />
Gender. Comparing flows into job lay<strong>of</strong>fs with flows into job stayers, the coefficient for “Male”<br />
is negative and significantly different from zero. This indicates that males are less likely than females<br />
to flow into the job lay<strong>of</strong>f category. Comparing flows into job losers with flows into job stayers, the<br />
85
esult also shows that males are less likely than females to flow into job lay<strong>of</strong>fs category. However,<br />
although the above findings concerning gender are significant, their magnitudes are relatively small.<br />
Age. According to the results in Table 2, age is also an important factor in determining the flows<br />
into unemployment. All the coefficients that comparing job lay<strong>of</strong>fs and job losers with job stayers for<br />
the different levels <strong>of</strong> the age categories (i.e. 25-34,35-44, and 45-54) are negative and significantly<br />
different from the reference group (i.e. 55-64 age group). The interpretation <strong>of</strong> this is that younger<br />
workers are less likely to flow into the job lay<strong>of</strong>f or job loser categories relative to job stayers. When<br />
compared the flows into job leavers relative to job stayers, we observed no significant effect for age.<br />
Marital status. Although the results in Table 2 show significant relationships between the subcategories<br />
<strong>of</strong> the marital status variable and the flows into unemployment, the magnitudes <strong>of</strong> flowing<br />
into a specific work status are very minimal.<br />
Univariate Logit Analysis<br />
Previous researchers (Borjas, 1981; Kidd, 1994) examining job separation did not differentiate<br />
among the different types unemployment. Specifically, past research grouped all the job movers into<br />
one category, which they labelled as job movers. However in our research, we argue that there are<br />
different categories <strong>of</strong> unemployment and these categories are differently influenced by the seven<br />
predictors suggested in this study. To support our argument, we conducted a univariate logit analysis<br />
and compare its results with the multinomial analysis results discussed above.<br />
Column 4 <strong>of</strong> Table 2 provides the binary logit analysis results. Although the result <strong>of</strong> the binary<br />
analysis demonstrates that the seven factors significantly influenced the probability (odd ratios) <strong>of</strong> an<br />
individual being classified into job stayer or job mover work status, a multinomial logit analysis that<br />
disaggregate job movers into three categories (job lay<strong>of</strong>fs, job losers, and job leavers) demonstrates<br />
that there are significant differences in the effect <strong>of</strong> the factors associated with these categories. For<br />
example, the results clearly demonstrate that the odds ratios <strong>of</strong> flowing into job lay<strong>of</strong>f, job loser, or<br />
job leaver category, compared with the flows into a job stayer category, are respectively, 3.14, 1.63,<br />
and 1.28 times higher for individuals who have high school or less level <strong>of</strong> education. In addition,<br />
table 2 also shows that the odds <strong>of</strong> flowing into job lay<strong>of</strong>f, job loser, or job leaver category, compared<br />
with the flows into a job stayer category, are respectively, 3.95, 13.82, and 7.99 times higher for<br />
individuals who worked with an organization for a year or less.<br />
Table 3.<br />
Multinomial Logit Estimates <strong>of</strong> the Likelihood <strong>of</strong> Flows into Unemployment<br />
Job losers<br />
Vs.<br />
leavers<br />
Job lay<strong>of</strong>fs<br />
Vs.<br />
leavers<br />
Job lay<strong>of</strong>fs<br />
Vs.<br />
losers<br />
Education<br />
high school or less 0.248** 0.898** 0.651**<br />
(1.281) (2.456) (1.917)<br />
some post secondary -0.025 0.498** 0.523**<br />
(.975) (1.645) (1.687)<br />
post secondary certificate or diploma 0.241** 0.744** 0.504**<br />
(1.272) (2.105) (1.655)<br />
Tenure<br />
1 year .548** -.702** -1.250**<br />
(1.7<strong>29</strong>) (.495) (.286)<br />
2 years .156* -0.673** -0.830**<br />
(1.169) (.510) (.436)<br />
3-4 years .059 -.464** -.523**<br />
86
(1.061) (.6<strong>29</strong>) (.593)<br />
5-9 years 0.040 -0.233@ -0.273**<br />
Occupation<br />
(1.041) (.792) (.761)<br />
-.124 -0.437** -0.313*<br />
Management, Business, Finance, and Administrative Occupations (.884) (.646) (.731)<br />
Sales and Service Occupations -0.257** -0.018 0.240*<br />
(.773) (.983) (1.271)<br />
Trades, Transportation and Equipment Operators and Related 0.254* 1.111** 0.858**<br />
Occupations (1.289) (3.038) (2.358)<br />
Occupations Unique to Primary Industries, Processing and<br />
0.243* 0.921** 0.678**<br />
Manufacturing (1.275) (2.512) (1.969)<br />
Industry<br />
Goods 0.693** 1.019** 0.326**<br />
(1.999) (2.769) (1.385)<br />
Sex<br />
Male 0.111* -0.257** -0.368**<br />
(1.117) (.744) (.692)<br />
Age<br />
25-34 -0.566** -0.420** 0.146*<br />
(.568) (.657) (1.157)<br />
35-44 -0.316** -0.222@ 0.094<br />
(.7<strong>29</strong>) (.801) (1.098)<br />
45-54 -0.188* -0.160 0.028<br />
(.828) (.852) (1.028)<br />
Marital status<br />
Married -0.058** 0.369** 0.311**<br />
(1.059) (1.446) (1365)<br />
Single -0.021 0.124 0.145@<br />
(0.979) (1.446) (.1156)<br />
Reference variables are: university and above (education), 10+ years (tenure), other (occupation), services (industry), female (sex), 55-64<br />
(age), divorced, widowed or separated (marital status)<br />
To examine if these ratios are significantly different from each other, we conducted two more<br />
new separate multinomial logit analyses by using the job leavers as reference group for the first<br />
analysis and using job losers as the reference group for the second analysis. Table 3 shows the result<br />
<strong>of</strong> these multinomial logit analyses. For example, table 3 shows that individuals who have high<br />
school or less are more likely to experience dismissal or permanent lay<strong>of</strong>f (odd ratio = 1.281) and<br />
temporary lay<strong>of</strong>f (odd ratio = 2.456) than voluntarily enter unemployment through resigning. Table 3<br />
also provides several significant results that demonstrate that there are different types <strong>of</strong> movers (or<br />
flows into unemployment) and these types are differently influenced by the seven predictors<br />
suggested in this study. In sum, the above discussed analysis provide strong support for our<br />
arguments regarding the need <strong>of</strong> disaggregating job movers into three categories (job lay<strong>of</strong>fs, job<br />
losers, and job leavers) instead <strong>of</strong> pooling them into one group.<br />
Discussion<br />
There are a number <strong>of</strong> noteworthy contributions <strong>of</strong> this paper. Overall, this research is just the<br />
beginning <strong>of</strong> our understanding patterns <strong>of</strong> flows into unemployment at both practical and theoretical<br />
levels. The unemployed group can no longer be assumed to be homogeneous. There are significant<br />
differences in the composition and correlations associated with the job lay<strong>of</strong>f, job leaver and job loser<br />
groups. Table 2 clearly identifies that disaggregating the unemployed group helps us understand the<br />
patterns <strong>of</strong> flows into unemployment in a more comprehensive and pragmatic way.<br />
As outlined in Table 1, the job loser group is not significantly different than the job stayer group.<br />
These two groups are different only in the marital status and tenure <strong>of</strong> their members. This suggests<br />
that when dismissing an employee, age, gender, education, occupation and industry do not change the<br />
87
composition <strong>of</strong> the active labour force. Instead, the composition <strong>of</strong> the active labour force is<br />
significantly affected by job lay<strong>of</strong>fs and job leavers. It is critical to note that on average, if the job<br />
lay<strong>of</strong>f group was over represented in a specific category, then the opposite relationship is found for<br />
the job leaver group (e.g. there were 48.4% less people in the job leaver group and 38.5% more<br />
people in the job lay<strong>of</strong>f group than in the job stayer group for the goods industry) and vice versa.<br />
Future research can evaluate if there are consistently opposite antecedents to job lay<strong>of</strong>f and job leaver<br />
groups.<br />
Patterns <strong>of</strong> unemployment have been outlined in this paper. For example, the higher the education,<br />
the lower the likelihood to be laid <strong>of</strong>f. Additionally, individuals employed in the services industry are<br />
predominantly more likely to quit (job leaver), whereas individuals employed in the manufacturing<br />
industry are more likely to be laid <strong>of</strong>f (job lay<strong>of</strong>f). These patterns can be further solidified in future<br />
research to address why these norms are occurring and what the practical implications <strong>of</strong> these<br />
differences are.<br />
The multinomial data analysis can be arranged into an algorithm to help predict patterns <strong>of</strong> labour<br />
force movement as the mix <strong>of</strong> the labour force changes. This data can be used for multiple purposes<br />
such as policy development for unemployment insurance, human resources planning within the firm,<br />
and even career selection. The sample provides statistical strength and the data spans across a five<br />
year period, securing our confidence in the generalizability <strong>of</strong> the results.<br />
Limitations <strong>of</strong> this study include the limited number <strong>of</strong> variables used in this research. Doe to the<br />
exploratory nature <strong>of</strong> the research, only seven variables were analysed to suggest differences in<br />
categories <strong>of</strong> unemployed persons. Future studies can include behaviour or cognitive variables<br />
(organizational commitment, job satisfaction, work relates stress), economic variables (wages,<br />
inflation) and additional demographic variables (race, visible minority status, number <strong>of</strong> dependants).<br />
As well, the data is limited to the Canadian labour force and cross sectional in nature. The diversity <strong>of</strong><br />
the labour force and the complexity <strong>of</strong> understanding employment patterns can be evaluated using<br />
similar studies across countries, or in a longitudinal analysis.<br />
References<br />
Aaronson, S., Fallick, B., Figura, A., & Pingle, J. (2006). The recent decline in the labor<br />
force participation rate and its implications for potential labor supply/comments and<br />
discussion. Brookings Papers on Economic Activity, 69-145.<br />
Autor, D. H., Katz, L. F., & Kearney, M. S. (2006). The polarization <strong>of</strong> the u.S. Labor market.<br />
The American Economic Review, 96(2), 189.<br />
Balkin, D. B. (1992). Managing employee separations with the reward system. The Executive,<br />
6(4), 64.<br />
Becker, G. (1963). Human capital. New York: Columbia Press.<br />
Borjas, G. J. (1981). Job mobility and earnings over the life cycle. Industrial & Labor<br />
Relations Review, 34(3), 365.<br />
Byrt, W. (1957). Methods <strong>of</strong> measuring labour turnover. Personnel Practices Bulletin, 13,<br />
413-417.<br />
Canada, S. (2007). Guide to the labour force survey. In L. S. <strong>Division</strong> (Ed.): Minister <strong>of</strong><br />
Industry.<br />
Cappelli, P. (2005). Will there really be a labor shortage? Human Resource Management,<br />
44(2), 143-150.<br />
88
Carnicer, M., Sanchez, A., Perez, M., & Jimenez, M. (2004). Analysis <strong>of</strong> internal and<br />
external labour mobility: A model <strong>of</strong> job-related and non-related factors. Personnel<br />
Review, 33(2), 222.<br />
Cohen, J., & Cohen, P. (1983). Applied multiple regression correlation analysis for the<br />
behaviour sciences (2nd ed). Hillsdale, NJ: Erlbaum.<br />
Cotton, J., & Tuttle, J. (1986). Employee turnover: A meta- analysis and review with<br />
implications for research. Academy <strong>of</strong> Management Review, 11, 55-70.<br />
Fallick, B. C. (1996). A review <strong>of</strong> the recent empirical literature on displaced workers.<br />
Industrial & Labor Relations Review, 50(1), 5.<br />
Farber, H. S. (1997). The changing face <strong>of</strong> job loss in the united states, 1981-1995.<br />
Brookings Papers on Economic Activity.<br />
Farber, H. S. (1999). Mobility and stability: The dynamics <strong>of</strong> job change in labor markets.<br />
The Handbook <strong>of</strong> Labor Economics, 3.<br />
Farber, H. S., & Hall, R. (1993). The incidence and costs <strong>of</strong> job loss: 1982-91; comments and<br />
discussion. Brookings Papers on Economic Activity(1), 73.<br />
Government <strong>of</strong> Canada and the Ontario Ministry <strong>of</strong> Training, C. a. U. (2007). Ontario job<br />
futures. In C. a. U. a. t. E. A. a. I. D. Labour Market Information and Research group<br />
<strong>of</strong> the Ministry <strong>of</strong> Training, Ontario Regional Office (Ed.) (Vol.): Ontario Ministry <strong>of</strong><br />
Training, Colleges and Universities and the Government <strong>of</strong> Canada.<br />
Griffeth, R., & Hom, P. (2001). Retaining valued employees. Thousand Oaks: Sage<br />
Publications.<br />
Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis <strong>of</strong> antecedents and<br />
correlates <strong>of</strong> employee turnover: Update, moderator tests, and research implications<br />
for the next millennium. Journal <strong>of</strong> Management, 26(3), 463.<br />
Hamermesh, D. S. (1989). What do we know about worker displacement in the u.S.?<br />
Industrial Relations, 28(1), 51.<br />
Hanisch, K. (1999). Job loss and unemployment research from 1994-1998: A review and<br />
recommendations for research and intervention. Journal <strong>of</strong> Vocational Behaviour, 55,<br />
188-220.<br />
Hitt, M., & Ireland, D. (2002). The essence <strong>of</strong> strategic leadership: Managing human and<br />
social capital. Journal <strong>of</strong> Leadership & Organizational Studies, 9(1), 3.<br />
Hom, P., Caranikas-Walker, F., Prussia, G. E., & Griffeth, R. (1992). A meta-analytical<br />
structural equation model <strong>of</strong> employee turnover. Journal <strong>of</strong> Applied Psychology, 77,<br />
890-909.<br />
Hom, P. W., & Griffeth, R. (1995). Employee turnover. Cincinnati, OH: South-Western.<br />
Juhn, C., & Potter, S. (2006). Changes in labor force participation in the united states. The<br />
Journal <strong>of</strong> Economic Perspectives, 20(3), 1.<br />
Kidd, M. P. (1991). An econometrc model <strong>of</strong> interfirm labour mobility. Econometrica, 24,<br />
517-535.<br />
Kidd, M. P. (1994). Some canadian evidence on the quit/lay-<strong>of</strong>f distinction. The Canadian<br />
Journal <strong>of</strong> Economics, 27(3), 709.<br />
Lee, T. L., Mitchell, T. R., Wise, L., & Fireman, S. (1996). An unfolding model <strong>of</strong> voluntary<br />
employee turnover. Academy <strong>of</strong> Management Journal, 39(1), 5.<br />
Marsh, R. M., & Mannari, H. (1977). Organizational commitment and turnover - a prediction<br />
study. Administrative Science Quarterly, 22(1), 57.<br />
McLaughlin, K. (1991). A theory on quits and lay<strong>of</strong>fs with efficient turnover. Journal <strong>of</strong><br />
Political Economy, 99(1-<strong>29</strong>).<br />
89
Mobley, W. (1982). Employee turnover: Causes, consequences and control. Reading:<br />
Addison-Wesley.<br />
Panagiotidis, T., & Pelloni, G. (2007). Nonlinearity in the canadian and u.S. Labor markets:<br />
Univariate and multivariate evidence from a battery <strong>of</strong> tests. Macroeconomic<br />
Dynamics, 11(5), 613.<br />
Pedhazur, E. J. (1982). Multiple regression in behavioral research: Explanation and<br />
prediction (2nd ed.). New York: Holt, Rinehart and Winston.<br />
Picot, G., Lin, Z., & Pyper, W. (1998). Permanent lay<strong>of</strong>fs in canada: Overview and<br />
longitudinal analysis. The Canadian Journal <strong>of</strong> Economics, 31(5), 1154.<br />
Porter, L. W., & Steers, R. M. (1973). Organizational, work and personal factors in employee<br />
turnover and absenteeism. Psychological Bulletin, 80(2), 151.<br />
Riley, R., & Young, G. (2007). Skill heterogeneity and equilibrium unemployment. Oxford<br />
Economic Papers, 59(4), 702-726.<br />
Schultze, C. (1999). Downsized and out? The Brookings Review, 17, 9-14.<br />
Seitchik, A. (1991). Who are displaced workers? In John T Addison ed. “Job Displacement:<br />
Consequences and Implications for Policy” Detroit: Wayne University Press.<br />
Simpson, W. (1990). Starting even? Job mobility and the wage gap between young single<br />
males and females. Applied Economics, 22, 723-737.<br />
Steel, R. P., & Ovaille, N. K. (1984). A review and meta-analysis <strong>of</strong> research on the<br />
relationship between behavioral intentions and employee turnover. Journal <strong>of</strong> Applied<br />
Psychology, 69, 673-686.<br />
Steers, R., & Mowday, R. (1981). Employee turnover and post-decision accommodation<br />
process: in L. Cummings and B. Staw Research in Organizational Behaviour, JAI<br />
Press, Connecticut.<br />
Tett, R. P., & Meyer, J. P. (1993). Job satisfaction, organizational commitment, turnover<br />
intention, and turnover: Path analyses based on meta-analytical findings. Personnel<br />
Psychology, 46(2), 259.<br />
Van Horn, C. E. (2006). Mega-trends in the american workforce. Human Resource<br />
Development Quarterly, 17(4), 475.<br />
90
ASAC <strong>2008</strong> Richard J. Long<br />
Halifax, Nova Scotia University <strong>of</strong> Saskatchewan<br />
Tony Fang<br />
York University<br />
DO EMPLOYEES PROFIT FROM PROFIT SHARING?<br />
A LONGITUDINAL ANALYSIS OF CANADIAN ESTABLISHMENTS<br />
Using panel data from a large sample <strong>of</strong> Canadian establishments, this<br />
paper examines whether there is any link between adoption <strong>of</strong> an<br />
employee pr<strong>of</strong>it sharing plan and subsequent employee earnings. Overall,<br />
growth in employee earnings during the three-year period subsequent to<br />
adoption <strong>of</strong> pr<strong>of</strong>it sharing did not differ significantly between<br />
establishments that had or had not adopted pr<strong>of</strong>it sharing. However,<br />
growth in employee earnings was significantly higher among pr<strong>of</strong>it<br />
sharing adopters that paid above-market wages prior to adoption <strong>of</strong> pr<strong>of</strong>it<br />
sharing. This suggests that pr<strong>of</strong>it sharing may be financially beneficial to<br />
employees in establishments making high investments in human capital.<br />
Although employee pr<strong>of</strong>it sharing is a pay practice that has a long history (Coates, 1991), and one<br />
that many firms continue to adopt (Lawler, Mohrman, & Ledford, 1998; Long & Shields, 2005;<br />
Parent, 2002), there is little evidence on whether and under what conditions employees benefit<br />
financially from pr<strong>of</strong>it sharing. While proponents argue that pr<strong>of</strong>it sharing increases total employee<br />
earnings (Bell & Hanson, 1987; Tyson, 1996), others contend that the effect <strong>of</strong> pr<strong>of</strong>it sharing on<br />
employee earnings will be neutral (Weitzman, 1984), and still others argue that pr<strong>of</strong>it sharing can<br />
actually serve to reduce employee earnings (Katz & Meltz, 1991).<br />
This is an important issue, from both a conceptual and practical perspective. Understanding the<br />
effects <strong>of</strong> pr<strong>of</strong>it sharing on employee earnings can help improve our theoretical understanding <strong>of</strong> this<br />
concept, and will also provide guidance to those (such as employees, their unions, and public policy<br />
makers) who have had a long-standing interest in knowing whether pr<strong>of</strong>it sharing is financially<br />
beneficial for employees (Florkowski, 1991). Using panel data from a sample <strong>of</strong> 1,717 Canadian<br />
establishments, this paper examines whether the adoption <strong>of</strong> employee pr<strong>of</strong>it sharing increases,<br />
decreases, or does not change employee earnings subsequent to adoption, and whether several factors<br />
(employee participation, union density, company size, and compensation level) condition its effects<br />
on employee earnings. Our research contributes to knowledge by applying a longitudinal design to a<br />
carefully constructed sample <strong>of</strong> Canadian establishments (the first pr<strong>of</strong>it sharing study to do so), by<br />
assessing both cash earnings and total earnings, and by testing for interactions with important<br />
establishment-level characteristics.<br />
Theoretical and Empirical Background<br />
Why should employee pr<strong>of</strong>it sharing affect employee earnings? The answer seems obvious—if<br />
employees start receiving pr<strong>of</strong>it sharing payments in addition to their regular compensation, then their<br />
total earnings should increase. But note that this outcome is contingent on two key circumstances.<br />
Total employee earnings will increase only when (a) the employer is pr<strong>of</strong>itable and actually pays a<br />
pr<strong>of</strong>it sharing bonus to its employees, and (b) this bonus exceeds any downward adjustments that may<br />
91
e made to other pay components subsequent to adoption <strong>of</strong> pr<strong>of</strong>it sharing.<br />
Regarding the first circumstance, if employee pr<strong>of</strong>it sharing is itself a practice that serves to<br />
increase company pr<strong>of</strong>its, as proponents would argue (Bell & Wallace, 1987; Tyson, 1996), this<br />
would provide an additional avenue for increased employee earnings. 2 Increased pr<strong>of</strong>its would<br />
produce a larger pr<strong>of</strong>it sharing bonus, and would also provide the employer with a greater financial<br />
capacity on which to base increases to components <strong>of</strong> pay other than pr<strong>of</strong>it sharing. But note that a<br />
causal connection between employee pr<strong>of</strong>it sharing and employer pr<strong>of</strong>itability is not a necessary<br />
condition for pr<strong>of</strong>it sharing to result in an increase in employee earnings, providing the employer is<br />
pr<strong>of</strong>itable subsequent to adoption <strong>of</strong> pr<strong>of</strong>it sharing. Absent a pr<strong>of</strong>itability-enhancing effect <strong>of</strong> pr<strong>of</strong>it<br />
sharing, pr<strong>of</strong>it sharing could increase employee earnings through a redistribution <strong>of</strong> pr<strong>of</strong>it from<br />
capital to labor. Of course, this would occur only if employers refrain from making downward<br />
adjustments (in excess <strong>of</strong> the pr<strong>of</strong>it sharing bonus) to other pay components.<br />
Whether employers would make such downward adjustments depends on their motives for<br />
adopting employee pr<strong>of</strong>it sharing. Theory suggests three main sets <strong>of</strong> motives for adopting pr<strong>of</strong>it<br />
sharing, all aimed at enhancing firm performance, but operating through different processes. The first<br />
set <strong>of</strong> motives is based on the substitution argument (Weitzman, 1984; Kruse, 1993). Under this<br />
argument, firms use pr<strong>of</strong>it sharing to substitute for fixed pay components (i.e. wages and benefits) to<br />
better align the firm’s labor costs with fluctuations in its ability to pay. When the firm’s financial<br />
capacity is high (i.e. in times <strong>of</strong> high pr<strong>of</strong>itability), employees receive a higher level <strong>of</strong> earnings, but<br />
when the firm’s financial capacity declines, so do employee earnings, thus reducing labor costs.<br />
Absent a variable pay component (as provided by employee pr<strong>of</strong>it sharing), the main alternative for<br />
reducing labor costs is employee lay<strong>of</strong>fs, which risks loss <strong>of</strong> valuable human capital. In his<br />
longitudinal study <strong>of</strong> US firms, Kruse (1996) found that firms with higher financial variability were<br />
indeed somewhat more likely to adopt pr<strong>of</strong>it sharing, although the magnitude <strong>of</strong> this effect was small.<br />
Firms pursuing the substitution motive may effect this substitution gradually (through<br />
constraining future increases in fixed pay components) or more immediately. For example, a common<br />
use <strong>of</strong> employee pr<strong>of</strong>it sharing in both the United States and Canada is as a vehicle to accumulate<br />
retirement savings for employees (Kruse, 1993; Long, 2006). One possibility, then, is to discontinue<br />
an existing retirement savings plan that requires fixed commitments from the employer in favor <strong>of</strong> a<br />
retirement plan based on pr<strong>of</strong>it sharing. Another possibility is simply to cut wages in conjunction with<br />
the introduction <strong>of</strong> pr<strong>of</strong>it sharing, as was done by financially-troubled North American automakers<br />
during the 1980s (Katz & Meltz, 1991). In this latter circumstance, however, the motive was not so<br />
much to introduce pr<strong>of</strong>it sharing as to make employers and their unions more amenable to the<br />
substantial pay cuts that the automakers argued were necessary for firm survival.<br />
In contrast, the second set <strong>of</strong> motives centers around using pr<strong>of</strong>it sharing as a vehicle explicitly<br />
intended to increase total employee earnings, with the object <strong>of</strong> enhancing employee attraction and<br />
retention. In this case, pr<strong>of</strong>it sharing may be regarded by employers as a less risky way to move to<br />
above-market “efficiency wages” than by increasing fixed wages and benefits. Alternatively, lowwage<br />
employers may use pr<strong>of</strong>it sharing to move closer to market pay rates. Either way, the employer<br />
has no intention <strong>of</strong> reducing other pay components, which would defeat the purpose <strong>of</strong> the pr<strong>of</strong>it<br />
sharing plan. To effect this increase in total employee earnings, some employers may simply add<br />
pr<strong>of</strong>it sharing to their current compensation practices, while others may use pr<strong>of</strong>it sharing as a vehicle<br />
2 Although the empirical evidence that employee pr<strong>of</strong>it sharing increases company productivity is quite clear, the<br />
empirical evidence on whether or not employee pr<strong>of</strong>it sharing actually increases firm pr<strong>of</strong>its is inconclusive<br />
(Magnan & St-Onge, 2005). For example, in his US study, Kim (1998) found that while pr<strong>of</strong>it sharing appeared to<br />
lead to increased company productivity, it did not lead to increased company pr<strong>of</strong>itability. He speculated that the<br />
productivity gain was fully captured by employees, in the form <strong>of</strong> pr<strong>of</strong>it sharing bonuses.<br />
92
to deliver a new benefit to employees. For example, employers that do not currently have employee<br />
pension plans may introduce employee pr<strong>of</strong>it sharing as a vehicle for generating retirement savings<br />
(Kruse, 1993; Tyson, 1996).<br />
The third set <strong>of</strong> motives centers around the explicit use <strong>of</strong> pr<strong>of</strong>it sharing as a productivityenhancing<br />
vehicle, by serving to increase employee motivation and cooperation on the job (Long,<br />
2000). In contrast to the efficiency-wages motive described above, the motive is not to increase firm<br />
performance by increasing employee earnings (which does so by attracting and retaining more highly<br />
qualified and productive employees), but to create a work context in which employees are motivated<br />
to work more diligently and effectively towards organizational goals. Employee pr<strong>of</strong>it sharing<br />
provides both the incentive and the reward for so doing. The implicit assumption is that an effective<br />
employee pr<strong>of</strong>it sharing plan should increase employee earnings, but that is not the direct motive for<br />
adoption <strong>of</strong> pr<strong>of</strong>it sharing.<br />
Given this, to understand the impact <strong>of</strong> employee pr<strong>of</strong>it sharing on employee earnings, it would<br />
be useful to understand the motives <strong>of</strong> top management for introducing employee pr<strong>of</strong>it sharing. One<br />
study that attempted to directly tap the motives <strong>of</strong> top management for adopting employee pr<strong>of</strong>it<br />
sharing could be found (Long, 1997). In his study, Long (1997) conducted telephone interviews with<br />
Chief Executive Officers <strong>of</strong> Canadian firms that had recently implemented employee pr<strong>of</strong>it sharing.<br />
Using an open-ended question, he found that the most frequently cited motives for adopting pr<strong>of</strong>it<br />
sharing could be clustered into two main groups. The first group <strong>of</strong> motives centered around<br />
improving company performance, through “improving employee motivation,” “promoting<br />
teamwork,” or “helping employees better understand the business.” The second set <strong>of</strong> motives<br />
centered around providing better rewards to employees, with pr<strong>of</strong>it sharing seen as “improving the<br />
compensation package,” “rewarding loyal employees,” “retaining employees,” and “building<br />
employee commitment.”<br />
Overall, no Canadian CEO mentioned any motive that might imply making pay more variable by<br />
reducing the fixed portion <strong>of</strong> the compensation package. However, in a survey <strong>of</strong> managers employed<br />
at US firms that had employee pr<strong>of</strong>it sharing, Mitchell and Broderick (1990) queried managers (not<br />
necessarily chief executive <strong>of</strong>ficers) on whether pr<strong>of</strong>it sharing was best at “raising productivity,”<br />
“increasing loyalty”, or “linking labor costs to firm’s economic conditions.” The answer most<br />
frequently selected was “linking labor costs,” followed by “increasing loyalty.” However, these<br />
responses are likely less indicative <strong>of</strong> the motives for introducing pr<strong>of</strong>it sharing than the perceived<br />
results <strong>of</strong> pr<strong>of</strong>it sharing.<br />
What is the overall evidence on the impact <strong>of</strong> employee pr<strong>of</strong>it sharing on employee earnings?<br />
Unfortunately, most evidence on this question is not recent, tends to be cross-sectional in nature, and<br />
shows mixed results. Mitchell, Lewin and Lawler (1990) used 1974 survey data from the United<br />
States Bureau <strong>of</strong> Labor Statistics, and found that both employee hourly wages and total compensation<br />
were higher in firms with pr<strong>of</strong>it sharing. However, in further research, they examined union contracts<br />
during the 1981-1988 period, and found that 36% <strong>of</strong> union contracts containing pr<strong>of</strong>it sharing<br />
provisions contained first year wage reductions, compared to only 14% <strong>of</strong> contracts that did not<br />
contain pr<strong>of</strong>it sharing. Consistent with this, data from the 1979-1988 period, Bell and Neumark (1993)<br />
examined US data from the 1979-1988 period and found that, among unionized manufacturing firms<br />
in the United States, those with pr<strong>of</strong>it sharing showed a lower growth in labor costs than firms<br />
without pr<strong>of</strong>it sharing (implying that pr<strong>of</strong>it sharing served to constrain employee earnings).<br />
A survey by the US Chamber <strong>of</strong> Commerce (1989) conducted in 1989 found that, in the<br />
manufacturing sector, fixed compensation was lower in pr<strong>of</strong>it sharing firms than those without pr<strong>of</strong>it<br />
sharing, but the opposite was true among nonmanufacturing firms.) Kim (1998) used a US data base<br />
93
collected in 1986 to conclude that pr<strong>of</strong>it sharing had increased labor costs in US firms, and, by<br />
implication, employee earnings. Finally, a study by Handel and Gittleman (2004), based on 1995 data<br />
from US business establishments, found pr<strong>of</strong>it sharing to be significantly positively related to<br />
employee earnings.<br />
In European research, Estrin and Wilson (1989) found that, among British engineering and<br />
metalworking firms during 1978-1982, fixed pay was higher in firms with pr<strong>of</strong>it sharing than those<br />
firms without pr<strong>of</strong>it sharing. Wadhwani and Wall (1990) also found in their sample British<br />
manufacturing firms that pr<strong>of</strong>it sharing tended to increase total employee compensation. Hart and<br />
Hubler (1991) examined a survey <strong>of</strong> German workers conducted in 1984-1985, and found that pr<strong>of</strong>it<br />
sharing was associated with higher individual wages, and found no support for the substitution<br />
argument. However, a more recent British study (Forth & Millward, 2004) based on 1998 data found<br />
no relationship between “financial participation” (which co-mingled pr<strong>of</strong>it sharing and employee<br />
stock plans) and employee hourly earnings, and a recently-published paper (McNabb & Whitfield,<br />
2007) based on the same data set found no relationship between presence <strong>of</strong> employee pr<strong>of</strong>it sharing<br />
and employee earnings.<br />
However, to adequately address the question <strong>of</strong> whether pr<strong>of</strong>it sharing affects employee earnings,<br />
longitudinal research is needed, and just two such studies could be found. Parent (2002) used data<br />
from the US National Longitudinal Survey <strong>of</strong> Youth collected during the 1990s to examine individual<br />
employee earnings before and after employees started receiving pr<strong>of</strong>it sharing payments and found<br />
interesting results—pr<strong>of</strong>it sharing apparently increased the earnings <strong>of</strong> male employees, but had no<br />
impact on the earnings <strong>of</strong> female employees. Kruse (1993:117) examined data from US public<br />
corporations during the 1975-1990 period and concluded that there was “very little difference in<br />
average compensation growth” between those firms that had or had not adopted pr<strong>of</strong>it sharing. He<br />
also concluded that there was some evidence for a substitution effect, as a slight increase in total<br />
compensation among pr<strong>of</strong>it sharing firms (relative to non-pr<strong>of</strong>it sharing firms) was less than the<br />
amount <strong>of</strong> the pr<strong>of</strong>it sharing bonuses. But as he notes, the sample on which he bases this conclusion is<br />
small, numbering just 30 companies.<br />
Conditions Affecting Pr<strong>of</strong>it Sharing and Earnings<br />
While the question <strong>of</strong> whether pr<strong>of</strong>it sharing affects employee earnings is our central focus, we<br />
also examine several conditions that may influence this relationship. Specifically, does employee<br />
participation in decision-making, union density, firm size, or the pre-existing compensation level <strong>of</strong><br />
employees affect the relationship between the introduction <strong>of</strong> employee pr<strong>of</strong>it sharing and subsequent<br />
employee earnings?<br />
If pr<strong>of</strong>it sharing is to increase total employee earnings, it is more likely to do so when pr<strong>of</strong>it<br />
sharing serves to increase the total financial resources available to the firm. Although the precise<br />
conditions under which pr<strong>of</strong>it sharing is most likely to enhance the financial performance <strong>of</strong> the firm<br />
are not well understood, one <strong>of</strong> these possible conditions is scope for employee participation in<br />
decision making. Numerous commentators (Bell and Hanson, 1987; Strauss, 1990; Kandel and Lazear,<br />
1992) argue that pr<strong>of</strong>it sharing will be more effective in improving organizational performance when<br />
accompanied by participatory practices. Proponents argue that the participatory mechanism serves as<br />
a key channel through which employees can operationalize the interest in workplace performance that<br />
has been generated by the financial incentive (Levine & Tyson, 1990). This argument is also<br />
consistent with the “high-involvement,” “high-commitment,” or “high-performance” models <strong>of</strong><br />
strategic human resources management, which posit that it is the interaction between a variety <strong>of</strong><br />
complementary human resources practices which produces significant increases in company<br />
94
performance (Allen & Wright, 2007; Delery & Doty, 1996; Lawler, 1986, 1992; Pil & MacDuffie,<br />
1996). If so, this higher company performance should translate into higher employee earnings, by<br />
increasing the size <strong>of</strong> the pr<strong>of</strong>it sharing bonuses, by increasing the firm’s capacity for higher fixed pay,<br />
or both.<br />
There is in fact some empirical evidence that supports this argument. Kim (1998) found in his US<br />
sample that pr<strong>of</strong>it sharing increased firm pr<strong>of</strong>itability when combined with employee involvement<br />
programs, but not otherwise. McNabb & Whitfield (2007) found a significant positive interaction<br />
between pr<strong>of</strong>it sharing and employee participation (in the form <strong>of</strong> joint consultative schemes) on<br />
employee earnings in Britain.<br />
Another condition under which pr<strong>of</strong>it sharing may be more likely to lead to increased employee<br />
earnings is when employees are members <strong>of</strong> a trade union. The argument here is that, if they accept<br />
pr<strong>of</strong>it sharing, employees represented by a union will be better able to protect and maintain their other<br />
(fixed) pay components than employees who are not so represented. Moreover, through their<br />
bargaining power, unionized workers may also be in a better position to negotiate pr<strong>of</strong>it sharing plans<br />
that are beneficial to employees, by ensuring, for example, that “threshold” levels for the payment <strong>of</strong><br />
pr<strong>of</strong>it sharing bonuses are not unrealistically high.<br />
However, rather than working to create more lucrative employee pr<strong>of</strong>it sharing plans, unions have<br />
generally focused their attention on simply opposing pr<strong>of</strong>it sharing plans, because <strong>of</strong> a preference for<br />
negotiated fixed wages, because <strong>of</strong> a concern that pr<strong>of</strong>its are subject to management manipulation,<br />
and because <strong>of</strong> concerns that pr<strong>of</strong>it sharing will be used as a tool to weaken loyalty to the union (Katz<br />
& Meltz, 1991). Indeed, a large body <strong>of</strong> evidence can be found that points to a negative association<br />
between union presence and presence <strong>of</strong> pr<strong>of</strong>it sharing, including studies in Canada (Long, 1989; Ng<br />
& Maki, 1994; Wagar & Long, 1995; Jones & Pliskin, 1997; Long & Shields, 2005), the United<br />
States (Cheadle, 1989; Kim, 1993; Cooke, 1994; Kruse, 1993, 1996), Britain (Poole, 1989), Germany<br />
(Heywood et al, 1998), and Australia (Long & Shields, 2005). But other studies for Canada (Long,<br />
1992), Britain (Estrin & Wilson, 1989), Germany (Carstensen et al, 1992), and Australia (Drago &<br />
Heywood, 1995) show no association, while two studies show positive relationships, one in Britain<br />
(Gregg & Machin, 1988) and the other in Brazil (Zylberstajn, 2002).<br />
However, all <strong>of</strong> the above results are from cross-sectional studies. In the two longitudinal studies<br />
that have addressed this issue, Kruse (1993; 1996) found some very interesting results. When he<br />
examined cross-sectional data, he found that union presence was negatively related to presence <strong>of</strong><br />
pr<strong>of</strong>it sharing in his two U.S. samples. However, when he examined adoptions <strong>of</strong> pr<strong>of</strong>it sharing, he<br />
found no relation between union presence and subsequent adoption <strong>of</strong> pr<strong>of</strong>it sharing in either sample.<br />
And when he examined cash-based pr<strong>of</strong>it sharing plans only, he found a significant positive<br />
relationship between union presence and adoption <strong>of</strong> pr<strong>of</strong>it sharing. This may indicate that, in the<br />
United States at least, unions are attempting to shape the nature <strong>of</strong> pr<strong>of</strong>it sharing plans, preferring<br />
cash-based plans over deferred pr<strong>of</strong>it sharing plans.<br />
A third factor that may influence the relationship between employee pr<strong>of</strong>it sharing and employee<br />
earnings is company size. If company size affects the success <strong>of</strong> pr<strong>of</strong>it sharing, then this may also<br />
affect the employee earnings produced by pr<strong>of</strong>it sharing. However, arguments can be made that firm<br />
size will be either negatively or positively related to pr<strong>of</strong>it sharing effectiveness. The negative<br />
argument is based on what economists refer to as the “1/N” or “free-rider” problem (Olson, 1971;<br />
Jensen & Meckling, 1976). If an individual employee increases effort and productivity, that<br />
individual receives only a small portion <strong>of</strong> the productivity gain, having to share it with all others<br />
included in the pr<strong>of</strong>it sharing system (“N”). Even if an individual does not change his or her effort, he<br />
or she can still gain from the increased effort <strong>of</strong> others, and thus become a “free rider.” The<br />
95
magnitude <strong>of</strong> this problem is thought to increase as “N” increases, suggesting that larger firms will<br />
benefit less from pr<strong>of</strong>it sharing than smaller firms.<br />
However, there are several counterarguments to this line <strong>of</strong> thinking. Proponents <strong>of</strong> pr<strong>of</strong>it sharing<br />
contend that pr<strong>of</strong>it-sharing systems can put into motion a set <strong>of</strong> dynamics that will discourage “free<br />
riding” and “loafing.” For example, Fitzroy and Kraft (1986) argue that if employees as a group<br />
understand that their economic well-being is maximized under conditions where free riding and<br />
loafing are minimized, then group norms may develop that value high productivity, and workers<br />
themselves will monitor adherence to these norms—“mutual monitoring” in agency terms. Similarly,<br />
Cooper, Dyck, and Frohlich (1992) argue that collective pay systems can overcome these problems if<br />
they include worker participation and a fair distribution rule. To the extent that these processes negate<br />
the “1/N” problem, then smaller firms may not benefit more from pr<strong>of</strong>it sharing than larger firms, and<br />
would therefore be no more likely to adopt pr<strong>of</strong>it sharing.<br />
Indeed, there may be reasons why pr<strong>of</strong>it sharing might be more beneficial to larger firms than to<br />
smaller firms. Larger firms will inherently have higher monitoring costs, because <strong>of</strong> their greater<br />
number <strong>of</strong> employees, and if pr<strong>of</strong>it sharing can play a role in reducing these monitoring costs, as<br />
suggested by Long’s (1994) study, pr<strong>of</strong>it sharing may be more beneficial to larger firms. Finally, if<br />
pr<strong>of</strong>it sharing helps to make employees more knowledgeable about and interested in the performance<br />
<strong>of</strong> the company as Long (1997) found, employees may be more motivated to share information that<br />
will contribute to higher quality organizational decision making (Kruse, 1993). Since it might be<br />
easier for employees to withhold this information in larger firms, where top management is more<br />
remote from the shop floor, pr<strong>of</strong>it sharing may benefit larger firms more than smaller firms.<br />
Overall, the weight <strong>of</strong> empirical evidence is against a negative relationship between firm size and<br />
presence <strong>of</strong> employee pr<strong>of</strong>it sharing, as most studies show either no relationship between firm size<br />
and pr<strong>of</strong>it sharing, or a positive relationship. No relationship between firm size and presence <strong>of</strong><br />
employee pr<strong>of</strong>it sharing was found by studies conducted in Canada (Long, 1989; Wagar & Long,<br />
1995; Long, 2002), the United States (Cheadle, 1989; Kruse, 1993, 1996) and Germany (Heywood et<br />
al, 1998), while longitudinal studies <strong>of</strong> pr<strong>of</strong>it sharing adoptions in the U.S. (Kruse, 1993, 1996) also<br />
showed no relationship with firm size. Positive relationships between firm size and presence <strong>of</strong> pr<strong>of</strong>it<br />
sharing were found in studies in Canada (Jones & Pliskin, 1997; Long, 1997; Long & Shields, 2005),<br />
in Britain (Poole, 1989; Poole & Jenkins, 1998), in Germany (Carstensen et al, 1992), and in<br />
Australia (Long and Shields, 2005). Just two studies show a negative relationship, one in Britain<br />
(Estrin & Wilson, 1989) and one in Japan (Kato & Moroshima, 2003).<br />
Finally, one other factor that may condition the relationship between pr<strong>of</strong>it sharing and employee<br />
earnings is the relative value <strong>of</strong> the human capital employed within the firm, as proxied by whether<br />
the firm compensates its employees above the market average for its industry. Presumably, firms pay<br />
above-market wages in order to attract and retain a higher quality <strong>of</strong> human capital within the firm.<br />
These firms may see pr<strong>of</strong>it sharing as a mechanism for leveraging their high investment in human<br />
capital, as employees at these firms may have more capacity to contribute to firm success than<br />
employees at firms with a lower quality <strong>of</strong> human capital. Hart and Hubler (1991) also point out that<br />
under rent-sharing theory, workers with relatively high levels <strong>of</strong> wage compensation are more likely<br />
to participate in pr<strong>of</strong>it sharing, as Long and Fang (2007) have found. Because <strong>of</strong> the value added by<br />
these employees, and their relative scarcity, it seems unlikely that firms with a high investment in<br />
human capital would use pr<strong>of</strong>it sharing for any purpose other than sharing rents with these employees,<br />
thus raising the total earnings <strong>of</strong> these employees.<br />
Methodology<br />
96
Sample and Analytical Procedure<br />
In conducting this research, we utilize a longitudinal panel <strong>of</strong> data, based on the Workplace and<br />
Employee Surveys (WES) conducted by Statistics Canada in 1999, 2001, and 2004. These surveys are<br />
designed to be representative <strong>of</strong> the total population <strong>of</strong> “workplaces” in Canada, but exclude business<br />
locations in the sparsely populated Yukon, Nunavut and Northwest Territories, as well as those in<br />
agriculture, fishing, road, bridge and highway maintenance, government services and religious<br />
organizations. The WES then follows the same workplaces over time, although replacement is made<br />
in every third year for workplaces that drop out <strong>of</strong> the survey. As utilized by Statistics Canada, a<br />
“workplace” is a business unit located at a single geographic location, and is analogous to the term<br />
“establishment” as frequently used in survey research. In this paper, we will generally use the more<br />
commonly understood term “establishment” to denote the unit <strong>of</strong> analysis. Our sample is limited to<br />
for-pr<strong>of</strong>it organizations only.<br />
The sample frame for the WES was generated from the Statistics Canada Business Register,<br />
which is a list <strong>of</strong> all businesses in Canada, updated monthly. Prior to sample selection, the business<br />
locations on the frame were stratified by industry, region, and size (based on estimated employment),<br />
and the sample was then selected using a Neyman allocation (Statistics Canada, 2004). The response<br />
rates for the 1999, 2001, and 2004 workplace surveys are stated by Statistics Canada as 95.2%, 85.9%,<br />
and 81.7% respectively, with most <strong>of</strong> the “non-responders” comprising owner-operators with no paid<br />
employees (Statistics Canada, 2006). Given the breadth and sensitivity <strong>of</strong> the information collected,<br />
these are rather remarkable response rates, no doubt facilitated by the facts that cooperation with<br />
Statistics Canada is obligatory and that extensive legal protections ensure confidentiality <strong>of</strong> responses.<br />
Data were collected through computer-aided telephone interviews with senior management<br />
<strong>of</strong>ficials at each workplace, conducted by trained interviewers based in Statistics Canada’s regional<br />
<strong>of</strong>fices. Each workplace was first sent a copy <strong>of</strong> the survey, with instructions to regard the survey as<br />
“as a working tool to inform you ahead <strong>of</strong> time <strong>of</strong> the questions being asked and to help you in<br />
preparing your answers.” As the survey is lengthy, and includes many questions requiring reference<br />
to company records, the intent was to allow respondents time to locate this information before being<br />
interviewed. The instructions emphasized that the survey forms are not to be returned by mail, but<br />
that the information is to be provided directly to the interviewer. The intent here is to provide the<br />
opportunity for the interviewer to clarify questions and answers, and then to follow up if necessary.<br />
After each survey, before any data were made available for research purposes, Statistics Canada<br />
spent more than two years conducting various procedures to ensure a clean data set. During data<br />
collection, the computer aided interview format provided various checks to reduce the possibility <strong>of</strong><br />
input errors or incorrectly recorded values. Following data collection, extended input editing was<br />
applied, followed by extensive data analysis and ratio editing to determine outlying observations<br />
based on robust outlier detection programs (Statistics Canada, 2004).<br />
The longitudinal panel <strong>of</strong> data for our study was constructed by first taking the 1999 WES and<br />
eliminating all workplaces with less than ten employees, those that are not for-pr<strong>of</strong>it enterprises, those<br />
that reported having pr<strong>of</strong>it sharing in 1999, and those that were not also included in the 2001 and<br />
2004 surveys. We also eliminated workplaces that adopted pr<strong>of</strong>it sharing during the 2001-2003 period.<br />
This resulted in 1,717 workplaces in our panel. A new variable was created in the 2001 data set based<br />
on whether the workplace reported having employee pr<strong>of</strong>it sharing in the 2001 survey, with “1”<br />
indicating that the workplace had adopted pr<strong>of</strong>it sharing, and “0” indicating the workplace continued<br />
not to have employee pr<strong>of</strong>it sharing.<br />
The rationale for this approach is that we wished to identify recent adopters <strong>of</strong> pr<strong>of</strong>it sharing (i.e.<br />
97
those that adopted between the 1999 and 2001 surveys, and then follow the growth <strong>of</strong> employee<br />
earnings during the three-year period 2001-2004. In this way, we have pre-existing data for both the<br />
workplaces that did and those that did not adopt pr<strong>of</strong>it sharing, and can compare earnings growth in<br />
the two groups, while incorporating a large array <strong>of</strong> control variables (as measured in 2001).<br />
The average workplace size (number <strong>of</strong> employees) is 51, and the average union density is 22.3<br />
percent. The distribution <strong>of</strong> workplaces by industry is: resources (1.4%), labor-intensive tertiary<br />
manufacturing (5.8%), primary product manufacturing (3.1%), secondary product manufacturing<br />
(5.8%), capital intensive tertiary manufacturing (6.3%), construction (6.1%),<br />
transportation/wholesaling (14.3%), communication/utilities (3.3%), retailing/consumer services<br />
(21.9%), finance/insurance (7.5%), real estate (1.6%), business services (15.2%), education and<br />
health services (5.3%), and information/cultural services (2.4%).<br />
Variable Measures<br />
An establishment was deemed to have adopted employee pr<strong>of</strong>it sharing if respondents to the 2001<br />
WES responded “yes” to the following question:<br />
“Does your compensation system include … [a] pr<strong>of</strong>it sharing plan? Pr<strong>of</strong>it-sharing plan<br />
is any plan in which employees receive a share <strong>of</strong> the pr<strong>of</strong>its from the workplace.”<br />
Any plans that applied only to managers were not deemed to be “employee pr<strong>of</strong>it sharing plans,”<br />
and these cases were eliminated from the panel. All remaining cases were designated “0” (no pr<strong>of</strong>it<br />
sharing adoption) or “1” (pr<strong>of</strong>it sharing adoption). Of the 1,717 establishments that did not have<br />
employee pr<strong>of</strong>it sharing in 1999, 247 (14.4%) had adopted it by 2001.<br />
Two measures <strong>of</strong> employee earnings growth are utilized. In so doing, we use an approach similar<br />
to that used by Kruse (1993) to control for industry differences. Growth in direct employee earnings<br />
was calculated by first taking the total gross payroll (including regular wages, commissions, overtime<br />
pay, piecework payments, and special payments) during the most recent fiscal year prior to data<br />
collection in 2001, and dividing this sum by the number <strong>of</strong> employees at the establishment. This<br />
quotient was then divided by the average employee earnings within the industrial sector in which the<br />
establishment operates, thus producing a ratio indicating whether the establishment paid above (a<br />
score <strong>of</strong> greater than 1) or below (a score <strong>of</strong> less than 1) the market in 2001. We label this variable<br />
“Cash Earnings 2001.” The cash earnings growth variable was derived based on the difference in real<br />
earnings (CPI-adjusted total payroll) between 2004 and 2001, divided by the 2001 cash earnings<br />
variable.<br />
However, merely using “cash real earnings growth” as the dependent variable may not portray the<br />
entire earnings picture. For example, some employers may consider pr<strong>of</strong>it sharing payments as<br />
“benefits” rather than direct earnings. This may be particularly true in cases where pr<strong>of</strong>it sharing<br />
payments are deferred, and are used as part <strong>of</strong> a retirement pension plan. Therefore, we also used a<br />
second measure <strong>of</strong> employee earnings growth—“Total Real Earnings Growth.” This measure was<br />
calculated in the same way as “cash real earnings growth,” except that it also included the cash value<br />
<strong>of</strong> non-wage benefits, such as the employer’s contribution to pension plans and other employee<br />
benefits. Another advantage <strong>of</strong> this second measure <strong>of</strong> earnings growth is that it will incorporate any<br />
decreases in the value <strong>of</strong> non-wage benefits subsequent to adoption <strong>of</strong> pr<strong>of</strong>it sharing. It is conceivable,<br />
for instance, that an employer may reduce non-wage benefits in concert with or subsequent to the<br />
adoption <strong>of</strong> pr<strong>of</strong>it sharing.<br />
To assess employee participation, we utilize a “participatory practices” index.<br />
98
Respondents were asked which (if any) <strong>of</strong> the following practices are currently in place, on a formal<br />
basis, for nonmanagerial employees: (a) suggestion systems, (b) problem solving teams, (c) joint<br />
labor-management committees, (d) information sharing programs, (e) flexible job design, and (f) selfdirected<br />
work groups. The participatory practices score for each establishment is the total number <strong>of</strong><br />
these practices in place, and thus varies from “0” to “6”. This type <strong>of</strong> method has been commonly<br />
used in attempting to ascertain the extent to which firms practice employee participation (Pil &<br />
MacDuffie, 1996; Walsworth & Verma, 2007; Zatzick & Iverson, 2006). Because this variable<br />
measure is an index, and not a scale, reporting <strong>of</strong> a value for Cronbach’s alpha is not appropriate<br />
(Delery, 1998). Union density is the proportion <strong>of</strong> total employees at a given establishment covered<br />
by a collective bargaining agreement. Establishment size is taken as the total number <strong>of</strong> employees at<br />
a given establishment.<br />
To control for industry sector, thirteen dummy variables are created, representing all <strong>of</strong> the<br />
sectors discussed earlier in this section, with the exception <strong>of</strong> retailing, which serves as the omitted<br />
(comparison) variable for analytical purposes. A further set <strong>of</strong> controls is used to control for the<br />
possible effect <strong>of</strong> performance pay other than pr<strong>of</strong>it sharing. For example, individual incentives have<br />
long been positively associated with employee earnings (Mitchell, Lewin, & Lawler, 1990; Lazear,<br />
2000; Parent, 2002). Therefore, through the use <strong>of</strong> dummy variables, we control for the presence <strong>of</strong><br />
individual incentives, merit pay, gain sharing, and employee stock plans. These controls are used in<br />
all multivariate analysis.<br />
Data analysis was carried out using OLS multiple regression, with each workplace weighted to<br />
represent its proportion in the general population. All continuous variables included in interaction<br />
terms were mean-centered before calculation <strong>of</strong> each interaction term.<br />
Results and Discussion<br />
Table 1 shows the means, standard deviations, and bivariate correlations for the sample. As can<br />
be seen, numerous variables are significantly related to pr<strong>of</strong>it sharing adoption. Pr<strong>of</strong>it sharing<br />
adopters are more likely to have all four types <strong>of</strong> performance pay plans—especially individual<br />
incentives and gain sharing—than establishments that did not adopt pr<strong>of</strong>it sharing. Establishments<br />
with more participatory practices are significantly more likely to adopt pr<strong>of</strong>it sharing, while<br />
establishments with more unionization are significantly less likely to adopt pr<strong>of</strong>it sharing. Pr<strong>of</strong>it<br />
sharing adopters also show significantly higher employee earnings in 2001—both in terms <strong>of</strong> cash<br />
earnings and total earnings—than establishments that did not adopt pr<strong>of</strong>it sharing. Indeed, the only<br />
variable not significantly related to pr<strong>of</strong>it sharing adoption is establishment size. This confirms the<br />
importance <strong>of</strong> controlling for these variables in our multiple regression analysis.<br />
Table 2 shows the multiple regression results, for both cash real employee earnings growth and<br />
total real employee earnings growth. As can be seen, pr<strong>of</strong>it sharing adoption is not significantly<br />
related to real employee earnings growth over the three-year period subsequent to adoption <strong>of</strong> pr<strong>of</strong>it<br />
sharing, regardless <strong>of</strong> whether the measure <strong>of</strong> real earnings growth is cash compensation or total<br />
compensation. The unstandardized regression coefficients are positive, but do not reach statistical<br />
significance. Thus, adoption <strong>of</strong> pr<strong>of</strong>it sharing appears to neither decrease employee earnings, as<br />
critics would fear, nor increase employee earnings, as advocates would expect. This result is virtually<br />
identical to that <strong>of</strong> Kruse’s (1993: 117) longitudinal study in the United States, where he found “very<br />
little difference in average compensation growth” between firms that had or had not adopted<br />
employee pr<strong>of</strong>it sharing.<br />
One possible interpretation <strong>of</strong> this result is support for the substitution argument, as Kruse (1993)<br />
99
has also noted. Assuming that the majority <strong>of</strong> establishments that adopted pr<strong>of</strong>it sharing were<br />
pr<strong>of</strong>itable in the period subsequent to adoption <strong>of</strong> pr<strong>of</strong>it sharing (our data in fact confirm this), it is<br />
not unreasonable to assume that pr<strong>of</strong>it sharing bonuses must have been paid by most establishments.<br />
Yet, payment <strong>of</strong> these bonuses has apparently not increased employee earnings. If so, then other<br />
earnings components must have been reduced in magnitude. What this implies is that in many<br />
establishments the adoption <strong>of</strong> pr<strong>of</strong>it sharing has served to maintain net employee earnings while<br />
reducing the fixed component <strong>of</strong> their compensation structures.<br />
100
Table 1<br />
Means, Standard Deviations, and Correlations a<br />
Variable Mean s.d. 1 2 3 4 5 6 7 8 9 10 11<br />
1. Pr<strong>of</strong>it Sharing Adoption .14 .35 -<br />
2. Participation <strong>Index</strong> 1.24 1.42 .19*** -<br />
3. Union Density .22 .35 -.10*** .09*** -<br />
4. Establishment Size 51.0 142.4 .04 .12*** .13*** -<br />
5. Cash Employee Earnings 2001 1.00 .61 .11*** -.08*** -.05* .01 -<br />
6. Total Employee Earnings 2001 1.00 .60 .10*** -.07*** -.04 .02 .99** -<br />
7. Individual Incentives .44 .50 .20*** .15*** -.06** .05* .09*** .09*** -<br />
8. Merit Pay .31 .46 .08*** .04* -.02 .07* .02 .02 .36*** -<br />
9. Gain Sharing .20 .40 .17*** .15*** .03 .03 -.05 -.04 .33*** .10** -<br />
10. Employee Stock Plan .11 .32 .09*** .18*** .11*** .09** -.01 .002 .<strong>29</strong>*** .31** .21*** -<br />
11. Cash Real Earnings Growth .08 .47 .01 -.07** .00 -.01 -.25*** -.25*** .06* .02 .00 .08*** -<br />
12. Total Real Earnings Growth .09 .48 .02 -.06** .01 -.01 -.25*** -.25*** .06* .01 .00 .07*** .99**<br />
a n = 1,717. *p
Table 2<br />
Multiple Regressions Predicting Employee Earnings Growth a<br />
Variable<br />
Growth in Cash Real Earnings Growth in Total Real Earnings<br />
Coefficient Standard Error Coefficient Standard Error<br />
Constant .271*** .094 .266*** .090<br />
1. Industry Controls<br />
Resources .004 .114 .037 .127<br />
Labor-Intensive Mfg. .082 .084 .086 .086<br />
Primary Product Mfg. .111 .093 .111 .095<br />
Secondary Product Mfg. .002 .068 .013 .068<br />
Capital-Intensive Mfg. .154* .083 .181** .088<br />
Construction .207* .119 .223* .120<br />
Transport/Wholesaling .103 .080 .110 .080<br />
Communications/Utilities .010 .065 .017 .066<br />
Finance/Insurance .143 .124 .137 .123<br />
Real Estate .203 .156 .213 .154<br />
Business Services .089 .134 .100 .141<br />
Education/ Health Services -.087 .096 -.109 .096<br />
Info/Cultural Services -.013 .088 -.011 .089<br />
2. Performance Pay Controls<br />
Individual Incentives .127*** .047 .127** .047<br />
Merit Pay -.031 .053 -.039 .054<br />
Gain Sharing -.082 .057 -.079 .056<br />
Employee Stock Plan .100 .079 .094 .085<br />
3. Establishment Characteristics<br />
Participation <strong>Index</strong> -.038* .022 -.035 .022<br />
Union Density .023 .077 .042 .080<br />
Establishment Size .00001 .00008 .00002 .00009<br />
Employee Earnings 2001 b -.273*** .048 -.268*** .046<br />
4. Pr<strong>of</strong>it Sharing Adoption .040 .059 .053 .061<br />
5. Interaction Terms<br />
PS X Participation .012 .023 .011 .023<br />
PS X Union Density -.055 .159 .015 .157<br />
PS X Size .00004 .00012 .00002 .00013<br />
PS X Earnings 2001 b .127** .060 .114 † .059<br />
Cases 1717 1717<br />
R 2 .116*** .118***<br />
a OLS specification (unstandardized regression coefficients) with standard errors for both specifications.<br />
b Cash Earnings 2001 used for regressions predicting Growth in Cash Real Earnings, and Total Earnings 2001 used<br />
in regressions predicting Growth in Total Real Earnings.<br />
† p
Although pr<strong>of</strong>it sharing may therefore be seen, on average, as earnings neutral to employees, this<br />
implied substitution <strong>of</strong> variable pay for fixed pay increases the variability <strong>of</strong> employee earnings, and<br />
shifts some additional business risk to employees. However, unlike in the well-known case <strong>of</strong> individual<br />
incentives (Mitchell, Lewin, & Lawler, 1990), employees are apparently not receiving any compensation<br />
for undertaking this additional risk, since total employee earnings are unchanged subsequent to adoption<br />
<strong>of</strong> pr<strong>of</strong>it sharing. From an employee perspective, therefore, the implementation <strong>of</strong> employee pr<strong>of</strong>it sharing<br />
may actually result in a less attractive compensation outcome, once the (uncompensated) additional risk is<br />
factored in.<br />
However, are there any circumstances under which pr<strong>of</strong>it sharing adoption may positively affect<br />
employee earnings? Earlier, we speculated that pr<strong>of</strong>it sharing may show an interaction with any one <strong>of</strong><br />
several variables—participatory practices, union density, company size, and compensation level policy. In<br />
fact, Table 2 does show one significant interaction effect—between pr<strong>of</strong>it sharing adoption and a highwage<br />
compensation policy. Employees in establishments that adopted pr<strong>of</strong>it sharing which had a highwage<br />
policy showed cash earnings growth about 12.7 percentage points higher than employees in other<br />
pr<strong>of</strong>it sharing establishments, and total earnings growth about 11.4 percentage points higher. This result<br />
suggests that employees in high-wage establishments will benefit from employee pr<strong>of</strong>it sharing, while<br />
employees in other establishments will not. This is consistent with the notion that establishments with a<br />
high investment in human capital will use pr<strong>of</strong>it sharing as a means to “share economic rent” and thus<br />
enhance the financial rewards to their employees.<br />
None <strong>of</strong> the other three variables show any interaction with pr<strong>of</strong>it sharing. That there is no evidence<br />
<strong>of</strong> an interaction between pr<strong>of</strong>it sharing and employee participation on employee earnings suggests that<br />
accompanying pr<strong>of</strong>it sharing with employee participation will not help employee earnings. That there is<br />
no significant interaction between pr<strong>of</strong>it sharing and union density suggests that unions do not play a<br />
significant role in developing more lucrative pr<strong>of</strong>it sharing plans, or in avoiding earnings substitution.<br />
That establishment size appears to play no significant role in the effects <strong>of</strong> pr<strong>of</strong>it sharing on employee<br />
earnings growth may indicate that size does not necessarily affect the results <strong>of</strong> pr<strong>of</strong>it sharing.<br />
That one prominent effect <strong>of</strong> employee pr<strong>of</strong>it sharing is apparently the substitution <strong>of</strong> variable for<br />
fixed pay is interesting, given that the Canadian CEOs in Long’s (1997) gave no indication <strong>of</strong> such a<br />
motive for their adoption <strong>of</strong> employee pr<strong>of</strong>it sharing. Of course, it is possible that other factors motivate<br />
adoption <strong>of</strong> pr<strong>of</strong>it sharing and that substitution <strong>of</strong> variable for fixed pay is an unintended side-effect.<br />
Another possibility is that motives for adoption <strong>of</strong> pr<strong>of</strong>it sharing have changed since Long’s study, which<br />
was conducted in 1989/90. Of course, it is also possible that many CEOs were not being entirely candid<br />
in identifying their motives for adoption <strong>of</strong> pr<strong>of</strong>it sharing.<br />
Finally, before concluding our discussion, we would like to briefly touch on several findings not<br />
central to our main question, but which are interesting nonetheless. One such finding is the relationship<br />
between employee participation and employee earnings. While there is some evidence that “highinvolvement”<br />
human resource practices can be beneficial to organizational performance (Ichniowski,<br />
Shaw, & Prennushi, 1997; Bailey, 1993; Huselid, 1995), there is little evidence to indicate that these<br />
practices are beneficial to employees, at least in terms <strong>of</strong> their earnings. Indeed, only one positive study<br />
could be found. Forth and Millward (2004) utilized a summative index <strong>of</strong> “high-involvement management<br />
practices” and found that this index was positively associated with employee earnings in their British<br />
sample. However, other studies in the United Kingdom (McNabb and Whitfield, 2007) and in the United<br />
States (Osterman, 2000; Handel & Gittleman, 2004) find no significant relationship between participatory<br />
practices and employee earnings. Indeed, McNabb and Whitfield (2007: 1014) find that “employees in<br />
establishments that have adopted quality circles actually face lower earnings, on average.” They conclude<br />
that even when tied to performance pay, employee participation <strong>of</strong>fers little financial advantage to<br />
employees.<br />
103
Our results also provide no support for the argument that employee participation is beneficial to<br />
employee earnings. Indeed, we found our “participatory practices index” to be significantly negatively<br />
related to cash employee earnings (in 2001) and to cash real earnings growth, whether accompanied by<br />
employee pr<strong>of</strong>it sharing or not (as the insignificant interaction coefficients show).<br />
Turning to other performance pay plans, we found that individual incentives are significantly<br />
positively related to both cash and total employee earnings (in 2001) and to growth in both cash real<br />
compensation and total real compensation. This result is consistent with a long line <strong>of</strong> research suggesting<br />
that employees on individual incentives earn more than those who are not (Parent, 2002). Our research<br />
also extends this finding to growth in employee earnings. As for gain sharing plans, while our results<br />
show that although employees on gain sharing plans did not earn less than other employees in 2001, they<br />
did experience significantly slower growth in cash real earnings. Finally, merit pay showed no significant<br />
relationship to earnings in 2001 nor to earnings growth. Note, however, none <strong>of</strong> the results for the various<br />
types <strong>of</strong> performance pay (except <strong>of</strong> course for pr<strong>of</strong>it sharing) or for participation are based on adoptions<br />
<strong>of</strong> these practices, but simply presence <strong>of</strong> these practices.<br />
As with all empirical studies, our study has both strengths and limitations. Strengths include use <strong>of</strong> a<br />
data set which embodies a large-scale sample, a very high response rate, and is carefully designed to be<br />
representative <strong>of</strong> Canadian for-pr<strong>of</strong>it establishments. Use <strong>of</strong> the establishment level <strong>of</strong> analysis allows for<br />
more precise measurement <strong>of</strong> the study variables than the corporate-wide measures that are <strong>of</strong>ten used in<br />
this kind <strong>of</strong> research. Another strength is that the data base allows for longitudinal analysis, but a<br />
limitation is that we are able to follow employee earnings for only a three-year period subsequent to<br />
adoption <strong>of</strong> pr<strong>of</strong>it sharing. Whether the results would change if a longer time span were used is unknown.<br />
A potential problem for all types <strong>of</strong> survey research is the reliability <strong>of</strong> the information collected.<br />
Gerhart, Wright, McMahon, and Snell (2000) have found reliability to be a major concern for survey data<br />
when it is collected from a single respondent, as is the case for the WES. However, this single respondent<br />
issue may not pose as much <strong>of</strong> a concern for the WES as it does for some surveys. First, the format <strong>of</strong> the<br />
WES is designed to enhance reliability and validity <strong>of</strong> responses, by allowing for preparation by the<br />
respondent, while utilizing interviews for actual data collection, which enables clarification <strong>of</strong> both<br />
questions and answers. In so doing, trained Statistics Canada interviewers are used, who have no vested<br />
interest in the particular outcomes <strong>of</strong> studies based on the survey information.<br />
Second, Gerhart, Wright, McMahon and Snell (2000) note that establishment-level surveys are likely<br />
more reliable than corporate-level surveys, because the units <strong>of</strong> analysis are smaller, managers are more<br />
familiar with HR practices because they are responsible for implementing them, and HR practices are<br />
more homogenous, and Gerhart, Wright, and McMahon (2000) do indeed find higher generalizability<br />
coefficients at the plant than at the company level. Further research by Wright and his colleagues (Wright<br />
et al., 2001) concluded that single-respondent surveys should use a single business or single location as its<br />
unit <strong>of</strong> analysis, as is true for the WES. Compared to the mean size <strong>of</strong> the organizational unit used by<br />
other well-known HR studies, such as Huselid’s (1995) survey (4,413 employees), the WES use <strong>of</strong> a<br />
single location results in a much smaller mean size (about 40 employees).<br />
Conclusions and Implications<br />
The results <strong>of</strong> our research suggest that while, on average, pr<strong>of</strong>it sharing may be earnings-neutral for<br />
employees, pr<strong>of</strong>it sharing may not generally be financially desirable for employees, if it shifts business<br />
risk from the employer to the employee, with no additional compensation to the employee for undertaking<br />
this additional risk. Our results imply that, for most Canadian employers, employee pr<strong>of</strong>it sharing serves<br />
to substitute variable pay for fixed pay, with no significant change in either cash real earnings or total real<br />
104
earnings over time. But note that if this substitution <strong>of</strong> variable pay for fixed pay serves to enhance<br />
employment stability, as economic theory would predict (Weitzman and Kruse, 1990), a trade<strong>of</strong>f <strong>of</strong><br />
earnings stability for employment stability is one that many employees might find acceptable or even<br />
desirable.<br />
However, our results also show that the implications <strong>of</strong> pr<strong>of</strong>it sharing for employee earnings vary,<br />
depending on whether workers are employed in high-wage establishments. In high-wage establishments,<br />
employee pr<strong>of</strong>it sharing does apparently serve to increase earnings, perhaps suggesting that high-wage<br />
employers see pr<strong>of</strong>it sharing as a way <strong>of</strong> leveraging their high investments in human capital and<br />
enhancing employee earnings, while other employers see pr<strong>of</strong>it sharing as a way <strong>of</strong> substituting variable<br />
for fixed employee earnings and/or shifting business risk to employees.<br />
Employees should not count on their unions to prevent this latter outcome, as union density made no<br />
difference to the earnings outcomes flowing from adoption <strong>of</strong> employee pr<strong>of</strong>it sharing. Therefore, unless<br />
they represent employees at high-wage establishments, unions would probably best serve their members’<br />
financial interests by avoiding pr<strong>of</strong>it sharing, something most North American unions already attempt to<br />
do. However, a second caveat to a policy <strong>of</strong> blanket opposition to pr<strong>of</strong>it sharing may be where unions can<br />
succeed in linking any loss in employee earnings stability to a gain in employment stability. Overall, what<br />
this suggests is that there may be circumstances under which unions find that pr<strong>of</strong>it sharing is beneficial<br />
to their members, and that a policy <strong>of</strong> blanket opposition to pr<strong>of</strong>it sharing may not always be in the best<br />
interests <strong>of</strong> their members.<br />
For employers, our results suggest that employee pr<strong>of</strong>it sharing is generally beneficial, if only<br />
because it serves to make employee compensation more responsive to the financial circumstances facing<br />
the firm, and because it serves to shift business risk to employees without adding to compensation costs.<br />
Of course, if employee pr<strong>of</strong>it sharing improves the productivity <strong>of</strong> the firm in some way, then the<br />
consequences are all the more beneficial to employers. Overall, our results imply either that pr<strong>of</strong>it sharing<br />
does not increase the overall financial performance <strong>of</strong> firms (and thus that there are no additional “rents”<br />
to share with workers) or that pr<strong>of</strong>it sharing does improve financial performance, but employees are not<br />
capturing any <strong>of</strong> this gain. This latter explanation is consistent with Kruse’s (1993) longitudinal results,<br />
which indicated that pr<strong>of</strong>it sharing had increased company productivity, but had not increased employee<br />
earnings. However, Kruse (1993) also observed wide variance across firms in the extent to which<br />
employee pr<strong>of</strong>it sharing brought productivity benefits, with about a third <strong>of</strong> pr<strong>of</strong>it sharing adopters<br />
showing no productivity gains at all. So it is possible that both explanations apply, but to different<br />
employers.<br />
The fact that we did find a set <strong>of</strong> employers (high-wage firms) whose employees do appear to benefit<br />
financially from adoption <strong>of</strong> pr<strong>of</strong>it sharing implies that pr<strong>of</strong>it sharing brings performance benefits to<br />
employers who invest relatively more than their competitors in human capital. Arguably, leveraging this<br />
investment through use <strong>of</strong> employee pr<strong>of</strong>it sharing serves to produce additional rents that can then be<br />
shared with employees. To the extent that this occurs, employee pr<strong>of</strong>it sharing will be beneficial to both<br />
employers and employees when adopted by high-wage firms. However, research specifically aimed at<br />
testing this proposition, which incorporates financial performance measures, is necessary before firm<br />
conclusions are drawn.<br />
105
References<br />
Allen, M.R., & Wright, P. (2007). Strategic management and HRM. In P. Boxall, J. Purcell & P Wright<br />
(Eds).The Oxford handbook <strong>of</strong> human resource management (pp. 88-107). Oxford: Oxford<br />
University Press.<br />
Bailey, T. (1993). Organizational innovation in the apparel industry. Industrial Relations, 32, 30-48.<br />
Bell, D.W., & Hanson, C.G. (1987). Pr<strong>of</strong>it Sharing and pr<strong>of</strong>itability: How pr<strong>of</strong>it sharing promotes<br />
business success. London: Kogan Page.<br />
Bell, L., & Neumark, D. (1993). Lump sum payments and pr<strong>of</strong>it sharing plans in the union sector <strong>of</strong> the<br />
United States economy. Economic Journal, 103, 602-619.<br />
Carstensen V., Gerlach, K. & Hubler, O. (1992). Determinants and microeconomic effects <strong>of</strong> pr<strong>of</strong>itsharing:<br />
Hypotheses and first results from German firms. Presented at the second meeting <strong>of</strong> the<br />
Impact <strong>of</strong> Pr<strong>of</strong>it Sharing in Europe (IPSE) Project, Paris.<br />
Cheadle, A. (1989). Explaining patterns <strong>of</strong> pr<strong>of</strong>it-sharing activity. Industrial Relations, 28(3), 387-400.<br />
Coates EM. (1991). Pr<strong>of</strong>it sharing today: Plans and provisions. Monthly Labor Review, 114(4), 19-25.<br />
Cooke, W.N. (1994). Employee participation programs, group-based incentives, and company<br />
performance: A union-nonunion comparison. Industrial and Labor Relations Review, 47(4), 594-<br />
609.<br />
Cooper, C., Dyck, B., & Frohlich, N. (1992). Improving the effectiveness <strong>of</strong> gainsharing: The role <strong>of</strong><br />
fairness and participation. Administrative Science Quarterly, 37(3), 471-490.<br />
Delery, J.E. (1998). Issues <strong>of</strong> fit in strategic human resource management: Implications for research.<br />
Human Resource Management Review, 8(3), 289-309.<br />
Delery, J.E., & Doty, D.H. (1996). Modes <strong>of</strong> theorizing in strategic human resource management: Tests<br />
<strong>of</strong> universalistic, contingency, and configurational performance predictions. Academy <strong>of</strong><br />
Management Journal, 39(4), 802-835.<br />
Drago R., & Heywood J.S. (1995). The choice <strong>of</strong> payment schemes: Australia establishment data.<br />
Industrial Relations, 34 (4), 507-531.<br />
Estrin S., & Wilson, N. (1989). Pr<strong>of</strong>it sharing, the marginal cost <strong>of</strong> labour and employment variability.<br />
University <strong>of</strong> Bradford, Management Centre.<br />
Fitzroy, F., & Kraft, K. (1986). Pr<strong>of</strong>itability and pr<strong>of</strong>it sharing. Journal <strong>of</strong> Industrial Economics, 35(2):<br />
113-130.<br />
Forth, John, and Neil Millward. (2004). High-involvement management and pay in Britain. Industrial<br />
Relations, 43(1), 98-119.<br />
Florkowski, G. (1991). Pr<strong>of</strong>it sharing and public policy: Insights for the United States. Industrial<br />
Relations, 30(1), 96-115.<br />
106
Gerhart, B., Wright, P.M., & McMahan, G.C. (2000). Measurement error in research on the human<br />
resources and firm performance relationship: Further evidence and analysis. Personnel<br />
Psychology, 53(4), 855-872.<br />
Gerhart, B., Wright, P.M., McMahan, G.C., & Snell, S.A. (2000). Measurement error in research on the<br />
human resources and firm performance relationship: How much error is there and how does it<br />
influence effect size estimates? Personnel Psychology, 53, 803-834.<br />
Gregg, P.A., & Machin, S.J. (1988). Unions and incidence <strong>of</strong> performance linked pay schemes in Britain.<br />
International Journal <strong>of</strong> Industrial Organization, 6(1), 91-107.<br />
Handel, Michael J., & Gittleman, Maury. (2004). Is There a Wage Pay<strong>of</strong>f to Innovative Work Practices?<br />
Industrial Relations, 43(1), 67-97.<br />
Hart, R.A., & Hubler, O. (1991). Are pr<strong>of</strong>it shares and wages substitute or complementary forms <strong>of</strong><br />
compensation? KYKLOS, 44(2), 221-231.<br />
Heywood, J.S., Hubler, O., & Jirjahn, U. (1998). Variable payment schemes and industrial relations:<br />
Evidence from Germany. KYKLOS, 51 (2), 237-257.<br />
Huselid, M.A. (1995). The impact <strong>of</strong> human resource management practices on turnover, productivity,<br />
and corporate financial performance. Academy <strong>of</strong> Management Journal, 38(3), 635-672.<br />
Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects <strong>of</strong> human resource management on<br />
productivity. American Economic Review, 87, <strong>29</strong>1-310.<br />
Jensen, M., & Meckling, W. (1976). Theory <strong>of</strong> the firm: Managerial Behavior, agency costs, and<br />
ownership structure. Financial Economics, 3, 305-360.<br />
Jones, D.C., & Pliskin, J. (1997). Determinants <strong>of</strong> the incidence <strong>of</strong> group incentives: Evidence from<br />
Canada. Canadian Journal <strong>of</strong> Economics, 30(4B), 1027-1045.<br />
Kandel, E., & Lazear, E. (1992). Peer pressure and partnerships. Journal <strong>of</strong> Political Economy, 100, 801-<br />
817.<br />
Kato, T., & Morishima, M. (2003). The nature, scope and effects <strong>of</strong> pr<strong>of</strong>it sharing in Japan: Evidence<br />
from new survey data. International Journal <strong>of</strong> Human Resource Management, 14 (6), 942-955.<br />
Katz, H.C., & Meltz, N.M. (1991). Pr<strong>of</strong>it Sharing and auto workers earnings: The United States vs<br />
Canada. Relations industrielles/Industrial Relations, 46 (3), 515-530.<br />
Kim, S. (1998). Does pr<strong>of</strong>it sharing increase firm pr<strong>of</strong>its? Journal <strong>of</strong> Labor Research, 19(2), 351-370.<br />
Kruse, D.L. (1993). Pr<strong>of</strong>it sharing: Does it make a difference? Kalamazoo, MI: W.E. Upjohn Institute.<br />
Kruse, D.L. (1996). Why do firms adopt pr<strong>of</strong>it sharing and employee ownership plans? British Journal <strong>of</strong><br />
Industrial Relations, 34, 515-538.<br />
Lawler, E.E. (1986). High-involvement management: Participative strategies for improving<br />
organizational effectiveness. San Francisco: Jossey-Bass.<br />
107
Lawler, E.E. (1992). The ultimate advantage: Creating the high-involvement organization. San Francisco:<br />
Jossey Bass.<br />
Lawler, E.E., Mohrman, S.A., & Ledford, G.E. (1998). Strategies for High Performance Organizations.<br />
San Francisco: Jossey Bass.<br />
Lazear, E.P. (2000). Performance pay and productivity. American Economic Review, 90, 1346-1361.<br />
Levine, D.I., & Tyson, L.D. (1990). Participation, productivity, and the firm’s environment. In A.S.<br />
Blinder (Ed.), Paying for productivity: A look at the evidence (pp. 15-94). Washington, DC: The<br />
Brookings Institution.<br />
Long, R.J. (1989). Patterns <strong>of</strong> workplace innovation in Canada. Relations industrielles/Industrial<br />
Relations, 44(4), 805-826.<br />
Long, R.J. (1992). The incidence and nature <strong>of</strong> employee pr<strong>of</strong>it sharing and stock ownership in Canada.<br />
Relations industrielles/Industrial Relations, 47(3), 463-488.<br />
Long, R.J. (1994). Gain sharing, hierarchy, and managers: Are they substitutes? <strong>Proceedings</strong> <strong>of</strong> the 1994<br />
Annual Conference, Administrative Sciences Association <strong>of</strong> Canada, 15(12):51-60.<br />
Long, R.J. (1997). Motives for pr<strong>of</strong>it sharing: A study <strong>of</strong> Canadian chief executive <strong>of</strong>ficers. Relations<br />
industrielles/Industrial Relations, 52(4), 712-733.<br />
Long, R.J. (2000). Employee pr<strong>of</strong>it sharing: Consequences and moderators. Relations<br />
industrielles/Industrial Relations, 55(3), 477-504.<br />
Long, R.J. (2002). Performance pay in Canada. In M. Brown & J.S. Heywood (Eds.), Paying for<br />
Performance: An International Comparison (pp. 52-89). Armonk, NY: M.E. Sharpe.<br />
Long, R.J. (2006). Strategic compensation in Canada. Toronto: Thomson Nelson.<br />
Long, R.J., & Shields, J.L. (2005). Performance pay in Canadian and Australian firms: A comparative<br />
study. International Journal <strong>of</strong> Human Resource Management, 16(10), 1783-1811.<br />
Long, R.J., & Fang, T. (2007). How strategic is strategic pay? The case <strong>of</strong> employee pr<strong>of</strong>it sharing.<br />
<strong>Proceedings</strong> <strong>of</strong> the Administrative Sciences Association <strong>of</strong> Canada Annual Conference, Ottawa.<br />
MacDuffie J.P. (1995). Human-resource bundles and manufacturing performance: Organizational logic<br />
and flexible production systems in the world auto industry. Industrial and Labor Relations<br />
Review, 48(2), 197-221.<br />
Magnan, Michel, & St-Onge, Sylvie. (2005). The impact <strong>of</strong> pr<strong>of</strong>it sharing on the performance <strong>of</strong> financial<br />
services firms. Journal <strong>of</strong> Management Studies, 42(4), 761-791.<br />
McNab, Robert, & Whitfield, Keith. (2007). The impact <strong>of</strong> varying types <strong>of</strong> performance-related pay and<br />
employee participation on earnings. International Journal <strong>of</strong> Human Resource Management,<br />
18(6), 1004-1025.<br />
Mitchell, D.J.B., Lewin, D., & Lawler, E.E. (1990). Alternative pay systems, firm performance, and<br />
productivity. In A.S. Blinder (Ed.), Paying for productivity: A look at the evidence (pp. 15-94).<br />
108
Washington, DC: The Brookings Institution.<br />
Mitchell, Daniel J.B., & Broderick, Renae F. (1991). Flexible systems in the American context: History,<br />
policy, research, and implications.” In D. Lewin, D. Lipski, & D. Sockell (Eds.) Advances in<br />
Industrial and Labor Relations, Greenwich, CT: JAI Press.<br />
Ng I., & Maki D. (1994). Trade-union influence on human resource management practices. Industrial<br />
Relations, 33(1), 121-135.<br />
Olson, M. (1971). The logic <strong>of</strong> collective action. Cambridge: Harvard University Press.<br />
Osterman, Paul. (2000). Work reorganization in an era <strong>of</strong> restructuring: Trends in diffusion and effects on<br />
employee welfare. Industrial and Labor Relations Review, 53, 179-196.<br />
Parent, M. (2002). Performance pay in the United States: Its determinants and effects. In M. Brown & J.S.<br />
Heywood (Eds.), Paying for Performance: An International Comparison (pp. 17-51). Armonk,<br />
NY: M.E. Sharpe.<br />
Pil, F.K., & MacDuffie, J.P. (1996). The adoption <strong>of</strong> high-involvement work practices. Industrial<br />
Relations, 35(3), 423-455.<br />
Poole, M. (1989). The origins <strong>of</strong> economic democracy: pr<strong>of</strong>it-sharing and employee-shareholding<br />
Schemes. London: Routledge.<br />
Poole, M. & Jenkins, G. (1990). The impact <strong>of</strong> economic democracy: pr<strong>of</strong>it sharing and employee<br />
shareholding schemes. London: Routledge.<br />
Poole, Michael, & Jenkins, G. (1998). Human resource management and the theory <strong>of</strong> rewards: Evidence<br />
from a national survey. British Journal <strong>of</strong> Industrial Relations, 36(2), 227-247.<br />
Statistics Canada. 2004. Workplace and Employee Survey compendium: 2001. Ottawa: Statistics Canada.<br />
Statistics Canada. 2006. Guide to the analysis <strong>of</strong> the Workplace and Employee Survey: 2003. Ottawa:<br />
Statistics Canada.<br />
Strauss, G. (1990). Participatory and gain sharing systems: History and hope. In M.J. Roomkin (Ed.)<br />
Pr<strong>of</strong>it sharing and gain sharing (pp. 1-45). Metuchen, NJ: Scarecrow Press.<br />
Tyson, D.E. (1996). Pr<strong>of</strong>it sharing in Canada. Toronto: John Wiley and Sons.<br />
US Chamber <strong>of</strong> Commerce. (1989). Employee Benefits. Washington, DC: Economic Policy <strong>Division</strong>,<br />
Chamber <strong>of</strong> Commerce <strong>of</strong> the United States.<br />
Wagar, T.H., & Long, R.J. (1995). Pr<strong>of</strong>it sharing in Canada: Incidence and predictors. <strong>Proceedings</strong> <strong>of</strong> the<br />
Administrative Sciences Association <strong>of</strong> Canada (Human Resources <strong>Division</strong>), 16:97-105.<br />
Walsworth, S., & Verma, A. (2007). Globalization, human resource practices and innovation: Recent<br />
evidence from the Canadian workplace and employee survey. Industrial Relations, 46(2), 222-<br />
240.<br />
Wahdwani, S., & Wall, M. (1990). The effects <strong>of</strong> pr<strong>of</strong>it sharing on employment, wages, stock returns, and<br />
109
productivity: Evidence from UK microdata. Economic Journal, 100, 1-17.<br />
Weitzman, M.L. (1984). The Share Economy. Cambridge, MA: Harvard University Press.<br />
Weitzman, M.L., & Kruse, D.L. 1990. Pr<strong>of</strong>it sharing and productivity. In A.S. Blinder (Ed.), Paying for<br />
productivity: A look at the evidence (pp. 95-142). Washington, D.C.: The Brookings Institution.<br />
Wright, P.M., Gardner, T.M., Moynihan, L.M., Park, H.J., Gerhart, B., & Delery, J.E. (2001).<br />
Measurement error in research on human resources and firm performance: Additional data and<br />
suggestions for future research. Personnel Psychology, 54(4), 875-901.<br />
Zatzick, C.D., & Iverson, R.D. ( 2006). High involvement management and workforce reduction:<br />
Competitive advantage or disadvantage? Academy <strong>of</strong> Management Journal, 49(5), 999-1015.<br />
Zylberstajn, H. (2002). The Brazilian case: Performance pay as workers’ right. In M. Brown and J.S.<br />
Heywood (Eds.), Paying for Performance: An International Comparison (pp. 236-260). Armonk,<br />
NY: M. E. Sharpe.<br />
110
ASAC <strong>2008</strong><br />
Halifax, Nova Scotia<br />
EXPLAINING CHRONOLOGICAL AND SUBJECTIVE AGE<br />
DIFFERENCES IN JOB SATISFACTION<br />
Results <strong>of</strong> a survey <strong>of</strong> 282 middle managers provide a partial explanation<br />
for the age-job satisfaction relationship. Subjective age was a stronger<br />
predictor <strong>of</strong> job satisfaction. Recognizing, or having others recognize<br />
one’s work experience partially mediated the relationship between<br />
chronological age and subjective age, and job satisfaction (global and<br />
facet).<br />
Age and Job satisfaction<br />
Nicole Bérubé (Student)<br />
John Molson School <strong>of</strong> Business<br />
Concordia University<br />
Understanding the relationship between age and work attitudes is widely recognized as crucial for<br />
organizational success (Glass, 2007; Greller & Stroh, 1995; Kanfer & Ackerman, 2004). Our need for this<br />
information is emphasized by recent demographic trends. Workers over 40 years <strong>of</strong> age currently<br />
dominate the North American workforce (Bureau <strong>of</strong> Labor Statistics, 2002; Fullerton, 1999) and labor<br />
force participation among workers over 50 is reported to be up 32% from the previous decade (Conlin,<br />
2003). Although it is widely recognized that aging workers are valuable human capital (Fullerton, 1999),<br />
concerns about how to manage aging workers are widespread (Cloutier, Lefebvre, Ledoux, Chatigny, &<br />
St-Jacques, 2002; Crampton, Hodge, & Mishra, 1996; Critchley, 2004; DeLong, 2004). Starting from the<br />
premise that age provides temporal and social benchmarks for self-assessment (Lawrence, 1987, 1988),<br />
empirical studies have demonstrated that age measures account for significant variance in important work<br />
outcomes such as job satisfaction and organizational commitment (Cleveland & Shore, 1992). Generally,<br />
past studies <strong>of</strong> the relationship between age and job satisfaction report that older workers are more<br />
satisfied with their jobs than are younger workers (e.g., Kalleberg & Loscocco, 1983; Rhodes, 1983;<br />
Wright & Hamilton, 1978). However, work attitudes evolve over time and current workplace<br />
demographics call for more research to update and increase our knowledge about the relationship between<br />
age and important work attitudes.<br />
Job satisfaction is one <strong>of</strong> the most frequently researched concepts in management research because it<br />
predicts many important individual work outcomes such as absenteeism, turnover, and counterproductive<br />
behaviors (Dormann & Zapf, 2001). Within the substantial body <strong>of</strong> research on job satisfaction, studies<br />
have directly investigated explanations for the relationship between age and job satisfaction have found<br />
that salary, tenure, and organizational level (e.g., White & Spector, 1987) are mediators <strong>of</strong> the<br />
relationship with job satisfaction. However, these studies investigated only the effect <strong>of</strong> chronological age<br />
on global job satisfaction, and did not consider perceptual age (the age a person feels), or job satisfaction<br />
facets. More importantly, the bulk <strong>of</strong> studies on the relationship between age and job satisfaction dates<br />
back at least two decades and may no longer reflect the situation in more modern workplaces. For<br />
example, in today’s rapidly changing organizational context, one’s salary, tenure or organizational level<br />
cannot be assumed to be linearly related to age. Younger employees could be better educated than older<br />
ones, and thus have access to higher-level positions and higher salaries in the organization. Organizational<br />
111
changes leading to downsizing and attrition may have affected older employees. By finding new<br />
employment in other organizations, they may have less tenure than younger employees already<br />
established in the organization. It is therefore necessary to update our knowledge about job satisfaction as<br />
it relates to age in modern workplaces. This investigation must consider how age is defined in current<br />
work contexts.<br />
In management research, age is most <strong>of</strong>ten measured chronologically (Rhodes, 1983) and most<br />
organizational commitment studies have followed this norm. However, chronological age alone may not<br />
adequately reflect age-related attitudes. For example, individuals <strong>of</strong> identical chronological age may<br />
attach different subjective meanings to their age (Cleveland, Shore, & Murphy, 1997). Cleveland and<br />
Shore (1992) define two types <strong>of</strong> perceptual age – subjective age is a personal evaluation <strong>of</strong> how old or<br />
young individuals perceive themselves to be on the basis <strong>of</strong> shared characteristics with others, and<br />
perceived relative age is a self-evaluation that individuals have about their own age, based on general<br />
social comparisons with members <strong>of</strong> a referent group, such as co-workers. In this study, we will focus on<br />
subjective age. Cleveland and Shore (1992) found that perceptual age measures could account for<br />
additional variance in individual outcomes not accounted for by chronological age. For example, they<br />
found that among individuals who were relatively older than their coworkers, those who perceived<br />
themselves as subjectively older were more satisfied than individuals who perceived themselves as<br />
subjectively younger. Subsequently, Cleveland et al. (1997) developed perceptual age scales and reported<br />
that perceptual age accounted for variance in perceived organizational support that was not accounted for<br />
by chronological age. Bobko and Barispolets (2002) utilized a similar measure <strong>of</strong> subjective age and<br />
found that the relationship <strong>of</strong> this variable with job performance and job stress was significantly different<br />
from the relationships between chronological age, performance and stress. Although job satisfaction was<br />
not a focal outcome in either <strong>of</strong> these studies, the results <strong>of</strong> these studies suggest that studying subjective<br />
age might add to our understanding <strong>of</strong> the age-job satisfaction link.<br />
Why focus more on subjective age? As the Baby Boom cohort ages, new interpretations arise about<br />
age and aging. For example, the popular press abounds with advertisement slogans and headlines touting<br />
that “50 is the new 40” and “70 is the new 60” (i.e., Lindgren, 2007). These social signals may influence<br />
the way people subjectively perceive their age. Furthermore, the elimination <strong>of</strong> mandatory retirement age,<br />
the increase in the age for pension eligibility, the increase in shorter organizational tenure due to<br />
organizational reorganization and downsizing, as well as the increase in protean career patterns are all<br />
expected to contribute to continued increases in the proportion <strong>of</strong> older workers in the next few decades.<br />
The age norms relevant to work and career are likely to change with the increase in the proportion <strong>of</strong><br />
older workers in the workforce, as it becomes increasingly acceptable to work, and even to establish new<br />
careers at an older age. In light <strong>of</strong> these trends, subjective age is becoming an increasingly important<br />
construct for studies that investigate relationships between age and work outcomes. However, it has more<br />
<strong>of</strong>ten been the object <strong>of</strong> studies in the field <strong>of</strong> gerontology and has received relatively little attention in the<br />
management literature to date. It is time to update and refine what we know about the age-job satisfaction<br />
relationship, and to investigate specifically and more thoroughly, the role <strong>of</strong> subjective age in work<br />
attitudes.<br />
Scholars have observed that older people tend to perceive themselves as subjectively younger than<br />
their chronological age (Rubin & Berntsen, 2006; Uotinen, Rantanen, Suutama, & Ruoppila, 2006), while<br />
younger people may tend to perceive themselves as subjectively older (Barnes-Farrell & Piotrowski,<br />
1989; Montepare & Lachman, 1989). This phenomenon could be the result <strong>of</strong> transitional phases in adult<br />
development that incite questioning and reformulation <strong>of</strong> aspirations (Peterson, 1996). The common<br />
saying that people are only as old as they feel can be informed by investigating subjective age. This<br />
construct <strong>of</strong>fers individuals a way to adjust their age to suit the way they feel, or to subjectively accelerate<br />
or slow down the process <strong>of</strong> aging to suit their needs. For example, older employees may report being<br />
subjectively younger than their chronological age in order to preserve their self-perception <strong>of</strong> their<br />
112
development potential at work. Younger workers who are building their career may be focused on<br />
objectives that more experienced colleagues have achieved and seeing themselves as subjectively older<br />
may be a way to make these goals seem more readily accessible.<br />
Levy (2003) discussed that age self-stereotypes are readily internalized and reinforced through<br />
various cognitive processes. These self-stereotypes would most likely be strengthened by commonly held<br />
age stereotypes in the workplace, and together these would constitute an efficient way to evaluate one’s<br />
subjective age. The literature on age stereotypes at work has documented that older workers are generally<br />
perceived to be more knowledgeable and experienced. For example, Forte and Hansvick (1999) found<br />
evidence <strong>of</strong> job specific stereotyping against younger workers for jobs requiring supervisory abilities.<br />
These qualities are generally relevant for managerial positions and may be associated with appreciation<br />
and respect from others. Therefore, younger managers may subjectively rate themselves as older to<br />
attribute to themselves the positive biases <strong>of</strong> knowledge and experience that are generally attributed to<br />
older managers.<br />
Similarly, older individuals may attempt to capture the advantages <strong>of</strong> performance biases that favour<br />
younger managers, who are <strong>of</strong>ten perceived to be more adaptable and faster learners (Hassell & Perrewe,<br />
1995). By thinking in terms <strong>of</strong> both chronological and subjective age, individuals can enjoy both the<br />
advantages <strong>of</strong> positive biases relative to their chronological age while concurrently benefiting from the<br />
positive biases related to their subjective age. By reducing the negative work-related effects associated<br />
with advancing chronological age, subjective age may thus contribute to increasing job satisfaction, as<br />
employees get chronologically older. Given the advantage <strong>of</strong> greater work experience, older managers are<br />
more likely to be able to draw from knowledge that facilitates their work. Consequently, they are<br />
expected to be more satisfied with their jobs than younger managers who are more likely to be at a lower<br />
point on the job experience learning curve.<br />
An abundant body <strong>of</strong> literature confirms a positive relationship between chronological age and overall<br />
job satisfaction (e.g., Brush, Moch, & Pooyan, 1987; e.g., Dormann & Zapf, 2001; Rhodes, 1983). In<br />
addition, empirical evidence strongly supports the existence <strong>of</strong> a positive association between age and<br />
satisfaction with the work itself, or work on present job (Rhodes, 1983). Scholars have <strong>of</strong>fered various<br />
explanations for the age-job satisfaction relationship. One common explanation for this relationship is<br />
based on the job change hypothesis, which maintains that compared to younger workers, older workers<br />
are more satisfied with their jobs because they have access to more desirable jobs, and because their needs<br />
and values have evolved over time to become more congruent with the jobs they occupy (Wright &<br />
Hamilton, 1978). Presumably, more desirable jobs have more satisfying job content and chronologically<br />
older persons should thus enjoy more satisfying jobs.<br />
We now turn to the relationship between subjective age and job satisfaction. Several studies report<br />
that people who felt subjectively older were more fatigued and felt less capable <strong>of</strong> working (Iskra-Golec,<br />
2002), had greater job and <strong>of</strong>f-job stress (Barnes-Farrell, Rumery, & Swody, 2002), and had lower<br />
general work ability. Such factors are likely to reduce the amount <strong>of</strong> pleasure one derives from<br />
accomplishing one’s work. Based on the foregoing, we propose the following hypotheses.<br />
Hypothesis 1: Chronological age will be positively related to overall job satisfaction and to satisfaction<br />
with work on present job.<br />
Hypothesis 2: Subjective age will be negatively related to overall job satisfaction and to satisfaction with<br />
work on present job<br />
Cleveland and Shore (1997) and Bobko and Barispolets (2002) reported evidence that subjective age<br />
could account for additional variance in individual work outcomes beyond that <strong>of</strong> chronological age.<br />
113
Since subjective age is within the individual’s control, it is more likely to reflect one’s age identity at<br />
work and age-related work values than chronological age. Feeling subjectively older involves more than<br />
acknowledging a number that reflects the passage <strong>of</strong> time. It involves an assessment <strong>of</strong> the way one feels,<br />
looks, one’s interests and one’s ideal age. As such, it is a better reflection <strong>of</strong> a person’s age self-concept.<br />
Therefore, it should be a better predictor <strong>of</strong> job satisfaction than chronological age. Based on these<br />
arguments, we propose the following hypothesis.<br />
Hypothesis 3: Subjective age will be a stronger predictor than chronological age <strong>of</strong> overall job<br />
satisfaction and satisfaction with work on present job.<br />
In line with the job change hypothesis (Wright & Hamilton, 1978), workers should, over time,<br />
develop expertise that facilitates their work in some manner. Even if they occupy different jobs, they<br />
bring to their job previous experience and their knowledge <strong>of</strong> working provides them with tools to cope<br />
with the everyday challenges <strong>of</strong> work. Older workers also tend to be perceived by others as having a great<br />
deal <strong>of</strong> work experience, even if they may actually have less than younger workers (Forte & Hansvick,<br />
1999; Gibson & Klein, 1970). We propose that recognizing one’s work experience and realizing that<br />
others recognize it might explain the link between age and job satisfaction.<br />
Traditional management thinking about age at work is rooted in life and career development models<br />
(Levinson, Darrow, Klein, Levinson, & McKee, 1978; Super, 1957). These models argue fundamentally<br />
that different experiences define life and career stages. Aging increases the probability that one will<br />
encounter situations that provide opportunities to recognize one’s own work experience and have it be<br />
recognized by others. These might include encountering a work situation where past experience comes in<br />
handy, being told by someone at work that one’s experience is valued, or being perceived as having a<br />
certain level <strong>of</strong> expertise in one’s field <strong>of</strong> work. Since various work experience is likely to increase with<br />
age, older employees should be more likely than younger employees to be aware <strong>of</strong> circumstances where<br />
their work experience is appreciated or assessed in some way. Therefore, older employees should be<br />
more likely to entertain expectations about their work experience being valued by others and experiencing<br />
situations where they could appreciate their own work experience first hand.<br />
Evaluating experience might apply in the arena <strong>of</strong> perceptual age as well as chronological age. For<br />
example, individuals who are chronologically 50 years old, but who subjectively assess the way they<br />
look, feel and act as older, should be more likely to value already acquired experience and expect others<br />
to recognize and appreciate it. Therefore, they are likely to be more sensitive to signals in their<br />
environment that emphasize their acquired experience. In contrast, 50 year-olds who are subjectively<br />
younger in appearance, deportment and interests, are more likely to want to maintain their youthful selfimage<br />
and social connections. They would thus be expected to be less sensitive to recognition <strong>of</strong> their<br />
acquired experience, but may instead be more focused on learning new things. This discussion leads to<br />
the following hypothesis:<br />
Hypothesis 4: Chronological and subjective age will be positively related to occurrences <strong>of</strong> recognition <strong>of</strong><br />
one’s work experience.<br />
Job satisfaction denotes a personal evaluation about the content and context <strong>of</strong> one’s job. Over the<br />
span <strong>of</strong> a person’s career, work and performance become important sources <strong>of</strong> relational and ego needs.<br />
As individuals acquire relevant work experience, having this experience recognized by others or<br />
recognizing it themselves, could engage the process <strong>of</strong> job satisfaction. Self-recognitions could be<br />
generated by successfully handling a challenging work situation, or by recognizing that one’s expertise<br />
has facilitated tasks. The level <strong>of</strong> enjoyment in the job should thus increase. Aside from the intrinsic<br />
satisfaction individuals get from accomplishing jobs they enjoy, people identify with their jobs for various<br />
reasons, including satisfying their sense <strong>of</strong> self-worth (Lemme, 1999), accessing socially satisfying<br />
114
elationships (Nuttman-Schwartz, 2004; Williams Walsh, 2001) and maintaining a recognized sense <strong>of</strong><br />
place, usefulness and purpose in society (Shaw & Grubbs, 1981). Receiving validation from others <strong>of</strong><br />
one’s experience is a strong social signal that validates one’s worthiness at work, a source <strong>of</strong> job<br />
satisfaction. Based on the foregoing, we hypothesize that:<br />
Hypothesis 5: Recognitions about work experience will be positively related to job satisfaction.<br />
The foregoing discussion leads us to the final goal <strong>of</strong> the research – to examine the various impacts <strong>of</strong><br />
chronological and subjective age on global job satisfaction and satisfaction with work on present job,<br />
through the mediating influence <strong>of</strong> receiving recognition from others about one’s work experience, and<br />
recognizing one’s own work experience. Based on the discussion leading to the previous two hypotheses,<br />
we propose that:<br />
Hypothesis 6: Assessments about work experience will partially mediate the relationship between age and<br />
job satisfaction.<br />
Participants<br />
Method<br />
Data collection was conducted using an online questionnaire, administered to a sample <strong>of</strong> 588<br />
Canadian middle managers in a large multinational firm in the services industry. A total <strong>of</strong> 282<br />
respondents completed the questionnaire (response rate: 48%). Respondents ranged in age from 21 to 64<br />
years (mean = 41.97). Their average organization tenure was 9.46 years and their average position tenure<br />
was 4.70 years. Both genders were adequately represented in this sample (56.7% women).<br />
Measures<br />
Subjective age was measured using the Subjective Age Scale, a four-item measure developed by<br />
Cleveland and Shore (1997) and tested longitudinally, with good reliability (α = 0.88 and 0.93). Items<br />
asked respondents to describe the age group that corresponded to the following: 1) The way you generally<br />
feel; 2) The way you look or your appearance; 3) The age <strong>of</strong> people whose interests and activities are<br />
most like yours; 4) The age that you would most like to be if you could choose your age right now. In our<br />
study, respondents chose one <strong>of</strong> six options (16-25 years, 26-35 years, 36-45 years, 46- 55 years, 56-65<br />
years, and 66-75 years), coded 1 (16-25 years) to 6 (66-75 years). In the present study the Cronbach’s<br />
alpha internal consistency reliability was .884.<br />
Overall job satisfaction was measured using five items from the Brayfield and Rothe scale (Brayfield<br />
& Rothe, 1951). This shortened version has been used in other research projects exhibiting acceptable<br />
reliabilities between .82 and .86 (e.g., Bono & Judge, 2003; Judge, Bono, & Locke, 2000) . Sample items<br />
are: “Most days I am enthusiastic about my work,” and “I feel fairly satisfied with my present job.”<br />
Responses were made on seven-point disagree-agree scales, and composite scores were computed by<br />
averaging across items. The alpha reliability for this study was .835.<br />
Two facets <strong>of</strong> the Job Descriptive <strong>Index</strong> (Smith, Kendall, & Hulin, 1969) were measured: Satisfaction<br />
with work on present job and Satisfaction with promotion opportunities. Indices for each facet are based<br />
on three types <strong>of</strong> responses (“yes,” “no,” and “uncertain?”). This instrument is widely used, reliable<br />
(internal consistency coefficients between .80 and .88) and well validated (Spector, 1997). In the present<br />
study, Alpha reliabilities were .902 (Satisfaction with work on present job) and .835 (Satisfaction with<br />
promotion opportunities). The present paper utilises, only Satisfaction with work on present job.<br />
115
To explore individuals’ age-relevant experiences, we generated a list <strong>of</strong> 49 age-related events based<br />
on themes that emerged from a review <strong>of</strong> the literature on age and work attitudes. Several <strong>of</strong> these events<br />
were inspired or adapted from the Life Events Inventory (Holmes & Rahe, 1967) because <strong>of</strong> their<br />
relevance to age-related work attitudes (Hughes, Blazer, & George, 1988). The list <strong>of</strong> events was pretested,<br />
and subsequently, two raters independently coded the events according to themes, yielding several<br />
categories <strong>of</strong> events (inter-rater reliability = 97.9%). The questionnaire included six events that pertained<br />
to recognition <strong>of</strong> one’s work experience. Three <strong>of</strong> these signalled recognition <strong>of</strong> one’s experience by<br />
others: “A less experienced colleague came to me for advice,” “Someone appreciated my work<br />
experience,” and “Someone mentioned that I had substantial work experience in my field.” The other<br />
three represented self-recognition <strong>of</strong> one’s work experience: “My work required me to rely on knowledge<br />
that less experienced people do not have,” “I showed a less experienced colleague a “trick <strong>of</strong> the trade”,”<br />
and “I was able to work at my own pace or in my own way because <strong>of</strong> my experience.” Experience <strong>of</strong><br />
age-relevant events was inventoried by asking respondents to report how <strong>of</strong>ten they experienced given<br />
events <strong>of</strong> occurrence during the past year (none = 0, once = 1, more than once = 2). Occurrences were<br />
summed to create an index for each type <strong>of</strong> recognition. Demographic data on chronological age, gender,<br />
years to retirement, and tenure were also measured on the questionnaire.<br />
Results<br />
Means, standard deviations and intercorrelations for the predictor and criterion variables are reported<br />
in Table 1. Chronological age and subjective age were positively correlated, which indicates that<br />
respondents identified subjectively with their chronological age group.<br />
Table 1<br />
Means, standard deviations and intercorrelations among variables<br />
1.Chronological age 282 41.97 8.99 1<br />
N M SD 1 2 3 4 5 6<br />
2. Subjective age 282 2.73 0.73 .761** 1<br />
3. Overall Job Satisfaction 282 5.46 1.17 .170** .230** 1<br />
4. Satisfaction with work<br />
on present job<br />
5. Experience recognized<br />
by self<br />
6. Experience recognized<br />
by others<br />
** p < .01<br />
* p < .05<br />
‡<br />
p = .058<br />
282 42.02 12.44 .194** .212** .615** 1<br />
281 4.11 1.72 .135* .113 ‡ .180** .307** 1<br />
282 3.25 1.53 .191** .165** .228** .367** .313** 1<br />
116
We found significant, positive, zero-order correlations between chronological age and overall job<br />
satisfaction (β = .170; p< .001), as well as between chronological age and satisfaction with work on<br />
present job (β = .194; p < .001). Thus our first hypothesis was fully supported. Contrary to our second<br />
hypothesis, which predicted a negative relationship between subjective age and job satisfaction, we found<br />
significant, positive relationships between subjective age and overall job satisfaction (β = .230; p < .001),<br />
as well as between subjective age and satisfaction with work on present job (β = .212; p < .001).<br />
Hypothesis 3 predicted that subjective age would be a stronger predictor than chronological age <strong>of</strong><br />
overall job satisfaction and satisfaction with work on present job. Table 1 shows that zero-order<br />
correlations for the relationship between subjective age and the two measures <strong>of</strong> job satisfaction were<br />
consistently stronger for subjective age, supporting H3. Additionally, Table 2 shows that when job<br />
satisfaction was regressed onto predictor variables in a multiple regression analysis, three variables<br />
contributed significantly to this regression: subjective age and the two types <strong>of</strong> work experience<br />
assessments. Job tenure and organizational tenure were used as control variables in these analyses.<br />
Table 2<br />
Multiple Regressions for Job Satisfaction and Satisfaction with Work on Present Job on Predictors<br />
Overall Job Satisfaction Satisfaction w/ Work on Present Job<br />
Beta<br />
t Beta<br />
t<br />
Chronological Age -.042 -.440 .043 .487<br />
Subjective age .205 2.324* .132 1.599†<br />
Org tenure -.026 -.340 .047 .647<br />
Job tenure .066 .928 -.135 -2.034*<br />
Experience recognized by self .119 1.963* .200 3.519**<br />
Experience recognized by others .189 3.031* .268 4.583**<br />
† p
Table 3<br />
Hierarchical Regressions for Overall Job Satisfaction on Predictors<br />
Variable entered<br />
R 2<br />
∆ R 2<br />
t to enter<br />
Experience recognized by self .032 .032 1.806†<br />
Experience recognized by others .065 .033 2.728*<br />
Chronological age .079 .014 -0.563<br />
Subjective age<br />
† p < .10<br />
* p < .05<br />
** p < .001<br />
.101 .022 2.596*<br />
The hierarchical regression procedure was repeated using satisfaction with work on present job as the<br />
criterion. When the explanatory variables were analyzed in hierarchical fashion with subjective age<br />
entered last, the model explained 20.7% <strong>of</strong> the variance in satisfaction with work on present job.<br />
Partialling other explanatory variables out <strong>of</strong> the chronological age-job satisfaction relationship reduced<br />
this correlation from .158 (p < .05) to .040 and it was no longer significant. Collinearity tolerance<br />
statistics were all over 0.87 except for the last variables entered in the equation. Collinearity tolerance<br />
statistics were lower for chronological age when it was entered last (Tolerance = .385) than for subjective<br />
age was it was entered last (Tolerance = .420). Overall, these results suggest that in this study, subjective<br />
age was a stronger predictor <strong>of</strong> job satisfaction than chronological age.<br />
Table 4<br />
Hierarchical Regressions for Satisfaction with Work on Present Job on Predictors<br />
Variable entered<br />
R 2<br />
∆ R 2<br />
t to enter<br />
Job tenure .001 .001 -1.971*<br />
Experience recognized by self .096 .095 3.563**<br />
Experience recognized by others .179 .082 4.820**<br />
Subjective age .205 .026 1.599†<br />
Chronological age<br />
† p < .10<br />
* p < .05<br />
** p < .001<br />
.207 .001 .662<br />
As hypothesized, chronological age was positively related to recognitions by self (β = .135, p < .05)<br />
and others (β = .191, p < .001) about one’s work experience. Furthermore, subjective age was positively<br />
related to recognitions by self (β = .113, p = .058) and others (β = .165, p < .05) about one’s work<br />
experience. Hypothesis 4 is thus fully supported. As hypothesized, self-recognition <strong>of</strong> work experience<br />
was positively related to overall job satisfaction (β = .180, p < .001) and to satisfaction with work on<br />
present job (β = .307, p < .001). Furthermore, recognition <strong>of</strong> one’s work experience by others was a<br />
118
significant, positive correlate <strong>of</strong> both overall job satisfaction (β = .228, p < .001) and satisfaction with<br />
work on present job (β = .367, p < .001). Hypothesis 5 is thus fully supported.<br />
In the next stage <strong>of</strong> the analysis, we examined whether the two types <strong>of</strong> assessments <strong>of</strong> one’s work<br />
experience mediated the link between age and job satisfaction. For these analyses, we followed the<br />
procedures described by Baron and Kenny (1986) and Sobel (1982). Because the mediation analysis was<br />
exploratory, we used the p < .10 levels for marginal significance to enable us to identify marginally<br />
significant relationships that we could follow up with a larger sample in future research.<br />
The analysis began by examining the relationship between chronological age and overall job<br />
satisfaction. In a regression analysis, chronological age was regressed on overall job satisfaction and<br />
found to be a significant predictor (β = .170, p < .001). At step 2, the “self-recognition <strong>of</strong> experience”<br />
index was introduced. The beta weight for chronological age associated with job satisfaction decreased<br />
but remained significant (β = .160, p = .007). These results suggest weak, partial mediation, which was<br />
confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z = 1.778, p = .075). We repeated this analysis using the<br />
“recognition <strong>of</strong> one’s experience by” index at step 2. The beta weight for age associated with job<br />
satisfaction decreased but remained significant (β = .132, p = .026). These results suggest weak, partial<br />
mediation, which was confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z = 2.375, p = .017).<br />
Next, subjective age was regressed on overall job satisfaction (β = .230; p< .001) and the “selfrecognition<br />
<strong>of</strong> experience” index was introduced at step 2. The beta weight for subjective age associated<br />
with overall job satisfaction decreased but remained significant (β = .211, p = .058). These results suggest<br />
weak, partial mediation, but the Sobel test <strong>of</strong> mediation was nonsignificant (Sobel’s z = 1.56, p = .12). We<br />
repeated this analysis using the “recognition <strong>of</strong> one’s experience by others” index at step 2. The beta<br />
weight for age associated with job satisfaction decreased but remained significant (β = .198, p = .001).<br />
These results suggest weak, partial mediation. This was confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z<br />
= 2.15, p = .031).<br />
Following this, chronological age was regressed on satisfaction with work on present job and found to<br />
be a significant predictor (β = .194, p < .001). At step 2, the “self-assessment <strong>of</strong> experience” index was<br />
introduced. The beta weight for chronological age associated with satisfaction with work on present job<br />
decreased but remained significant (β = .156, p = .007). These results suggest weak, partial mediation,<br />
which was confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z = 2.138, p = .03). We repeated this analysis<br />
using the “recognition <strong>of</strong> one’s experience by others” index at step 2. The beta weight for age associated<br />
with satisfaction with work on present job decreased but remained significant (β = .1<strong>29</strong>, p = .023). These<br />
results suggest weak, partial mediation, which was confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z =<br />
2.904, p = .003).<br />
Finally, subjective age was regressed on satisfaction with work on present job and found to be a<br />
significant predictor (β = .212, p < .001). At step 2, the “self-recognition <strong>of</strong> experience” index was<br />
introduced. The beta weight for subjective age associated with satisfaction with work on present job<br />
decreased but remained significant (β = .211, p = .058). These results suggest weak, partial mediation,<br />
which was confirmed by a Sobel test <strong>of</strong> mediation (Sobel’s z = 1.790, p = 0.073). We repeated this<br />
analysis using the “recognition <strong>of</strong> one’s experience by others” index at step 2. The beta weight for age<br />
associated with satisfaction with work on present job decreased but remained significant (β = .156, p =<br />
.006). These results suggest weak, partial mediation. This was confirmed by a Sobel test <strong>of</strong> mediation<br />
(Sobel’s z = 2.15, p = .031).<br />
These results indicate that we obtained support for seven out <strong>of</strong> eight partial mediations. Therefore,<br />
our sixth hypothesis was partially supported. Generally, we concluded that there was weak but relatively<br />
consistent evidence that self-recognition about one’s work experience, as defined by the sum <strong>of</strong> the<br />
119
occurrences <strong>of</strong> three types <strong>of</strong> self-recognitions, and recognitions from others about one’s work<br />
experience, as defined by the sum <strong>of</strong> the occurrence <strong>of</strong> three types <strong>of</strong> recognitions by others, partially<br />
mediated the relationship between age and job satisfaction.<br />
Discussion<br />
Findings from this study show that the subjective evaluation individuals make about their age is an<br />
important factor in reporting occurrences <strong>of</strong> recognition <strong>of</strong> their work experience, by themselves or<br />
others. In this sample, respondents generally identified subjectively with their chronological age group.<br />
However, this general outcome may not hold across age groups or generational cohorts. Group<br />
differences were not included here due to length limitations. However, our finding <strong>of</strong> a general coherence<br />
between chronological and subjective age is nevertheless interesting since it suggests that by focusing on<br />
differences between groups, managers may lose sight <strong>of</strong> the notion that these differences may not<br />
generally matter as much as they are <strong>of</strong>ten assumed to matter.<br />
Findings also indicate that subjective age is more strongly related to both overall job satisfaction and<br />
satisfaction with work on present job. This indicates that much <strong>of</strong> the discourse about age in the<br />
workplace, which relies heavily on chronological age, may overlook issues that are important for<br />
managing key work outcomes. This perspective has important implications for human resource questions<br />
about staffing, recruitment, and retirement preparation – questions that become increasingly important as<br />
the Canadian workforce ages.<br />
The opposite direction we obtained for the relationship between subjective age and job satisfaction<br />
may be related to context. Our sample consisted <strong>of</strong> middle managers in the service industry. The samples<br />
used in previous studies (Barnes-Farrell et al., 2002; Iskra-Golec, 2002) consisted <strong>of</strong> people employed in<br />
a broader range <strong>of</strong> jobs in the health care sector. This suggests that differences in jobs and organizational<br />
sectors may influence assessments <strong>of</strong> subjective age. For example, health care workers occupy jobs that<br />
are generally more physically demanding than those occupied by managers. Furthermore, work in the<br />
health care sector may <strong>of</strong>fer a more limited range <strong>of</strong> opportunities for doing tasks that people prefer more<br />
than others. To protect patients, health care must be delivered according to strict processes. By<br />
comparison, managers in the service industry have less physically demanding work and would be likely to<br />
be able to work more creatively, and on a much broader range <strong>of</strong> tasks. Overall, they are more likely to be<br />
able to work on tasks they prefer within the scope <strong>of</strong> their jobs. It is also possible that the more physical<br />
nature <strong>of</strong> the job <strong>of</strong> health care workers would also emphasize the negative consequences <strong>of</strong> feeling<br />
subjectively older (e.g., fatigue, reduced stamina). By comparison, managers in the service industry might<br />
not be as likely to feel the negative effects <strong>of</strong> feeling subjectively older. Furthermore, because <strong>of</strong> the great<br />
reliance on tacit knowledge in managerial jobs, they may be more likely to benefit from the stereotypes<br />
that favour older people with having more wisdom. Therefore, subjectively older managers would be<br />
more likely to be satisfied with their job and their job content than subjectively older health care workers.<br />
Our findings that older individuals are more likely than younger individuals to encounter occurrences<br />
<strong>of</strong> recognition by others emphasizes the importance <strong>of</strong> recognizing and valuing work experience,<br />
especially for older workers, who may be more sensitive to this form <strong>of</strong> appreciation. In a context where<br />
human resource practitioners are becoming concerned with issues <strong>of</strong> staffing and transfer <strong>of</strong> knowledge as<br />
Baby Boomers approach retirement, it will be important to find concrete ways to recognize and validate<br />
the knowledge and work experience acquired by older employees over the course <strong>of</strong> their lives. It is<br />
unlikely that unhappy workers will want to help their employers by training and mentoring their<br />
replacements. Valuing past work experience is not only likely to lead to more satisfied workers, but it<br />
could also help stimulate the sharing <strong>of</strong> tacit knowledge within organizations.<br />
120
More generally, the positive link we found between recognition <strong>of</strong> experience and job satisfaction<br />
suggests that organizations need to be more sensitive to the role that specific types <strong>of</strong> recognition play<br />
with regards to the association between age and job satisfaction. For example, it is reasonable that<br />
individuals begin to adjust their performance levels if they start to perceive that the association between<br />
organizational rewards and their performance is weakening. If recognition <strong>of</strong> experience suggests a<br />
stronger association, or signals that employees are more appreciated over time, it is not surprising that job<br />
satisfaction increases. These relationships can be further clarified with future research that focuses on the<br />
contextual and interpretive aspects <strong>of</strong> recognition <strong>of</strong> work experience.<br />
It has also been suggested that older employees may have cognitively rationalized lower expectations<br />
from their jobs, which may explain why they are more satisfied than younger workers (Kline, Sulsky, &<br />
Rever-Moriyama, 2000). Our findings suggest an alternative or complementary view – that employees<br />
seek out sources <strong>of</strong> satisfaction. This is an interesting perspective since recognizing experience is quite<br />
manageable. Often we think <strong>of</strong> communicating this type <strong>of</strong> recognition through praise or rewards.<br />
However, there are alternative recognition strategies that could work equally well or better, such as<br />
providing opportunities for older workers to show others what they know, and seeking out older workers<br />
for job challenges that require tacit knowledge.<br />
What might help us understand the age-satisfaction dynamics more clearly? We have shown that<br />
using subjective age constitutes a valuable key to this understanding. Accordingly, we recommend more<br />
thorough investigations <strong>of</strong> the age-commitment relationship, using perceptual age measures. For example,<br />
Cleveland and Shore (1992) found that individuals who perceived themselves as subjectively older and<br />
who reported perceiving themselves as subjectively older and relatively older than the members <strong>of</strong> their<br />
work group had the most positive work attitudes. This suggests that congruence between types <strong>of</strong><br />
perceptual age is useful in explaining variance in the relationship between age and work attitudes. Due to<br />
space limitations, we focused only on subjective age as a perceptual age measure in this paper. However,<br />
future research should investigate the effects <strong>of</strong> both subjective age and perceived relative age (one’s age<br />
compared to the average age <strong>of</strong> members <strong>of</strong> one’s work group) on various forms <strong>of</strong> job satisfaction and<br />
other work attitudes.<br />
Although our findings are instructive, they must be considered in light <strong>of</strong> the limitations <strong>of</strong> this study.<br />
The sample consisted only <strong>of</strong> managers in a single industry, which raises concerns about the<br />
generalizability <strong>of</strong> the findings. Finally, this study is cross-sectional and therefore, no inferences can be<br />
made about causality. Furthermore, common method variance is always a concern with self-reports and<br />
with cross-sectional data analysis. Collecting longitudinal data on this sample in the future could help<br />
alleviate concerns with regards to common method variance. Although gender was measured in our<br />
survey, an analysis including gender extends beyond the scope <strong>of</strong> this paper. However, we plan to detail<br />
our findings regarding the effect <strong>of</strong> gender on the age-commitment relationship in future papers. In<br />
addition, our paper focuses on older workers, although the data collected is relevant to workers <strong>of</strong> various<br />
ages. A more thorough investigation <strong>of</strong> the results across ages is also planned in the future.<br />
This study extended research on the relationship between age and job satisfaction by investigating the<br />
influence <strong>of</strong> both chronological and subjective age on overall job satisfaction and satisfaction with work<br />
on present job. In light <strong>of</strong> the need for knowledge about managing an aging workforce, still relatively few<br />
studies have investigated the role <strong>of</strong> perceptual age measures on work attitudes and no studies on<br />
perceptual age have attempted to test possible explanations for the age-satisfaction relationship. Our study<br />
also contributes to managerial knowledge by providing information regarding the influence <strong>of</strong> key events<br />
on the job satisfaction <strong>of</strong> older employees. Our results contribute further to managerial knowledge by<br />
demonstrating significant links between subjective age and two types <strong>of</strong> job satisfaction, as well as<br />
between common indicators <strong>of</strong> recognition about one’s work experience, age, and job satisfaction.<br />
121
The importance <strong>of</strong> our findings lie in the widely-reported positive relationship between age and job<br />
satisfaction, and in the fact that organizational demographics will increasingly require managers to know<br />
more about managing aging workers. As the average age <strong>of</strong> employees in organizations increase with the<br />
advancing age <strong>of</strong> the Baby Boom cohort, the interplay between age and various forms <strong>of</strong> job satisfaction<br />
will become increasingly important for organizations. These relationships comport important implications<br />
for managing and retaining the most knowledgeable employees in an aging workforce. Since desirable<br />
consequences are advantageous to both organizations and employees, and since the proportion <strong>of</strong> older<br />
workers is increasing dramatically, we need to urgently devise practices that help maintain and increase<br />
job satisfaction for older workers. Investigating mediators <strong>of</strong> the age-satisfaction relationship will help<br />
identify relevant practices. In this study, we investigated the mediating role <strong>of</strong> recognition <strong>of</strong> work<br />
experience. Future studies should also examine the way these are transmitted to employees, and<br />
investigate other possibly relevant mediators, such as how organizational rewards and development<br />
activities are distributed and how opportunities to realize late-career ambitions are managed.<br />
REFERENCES<br />
Barnes-Farrell, J., & Piotrowski, M. J. (1989). Workers' perceptions <strong>of</strong> discrepancies between<br />
chronological age and personal age: You're only as old as you feel. Personality and Aging, 4(3),<br />
376 - 377.<br />
Barnes-Farrell, J., Rumery, S. M., & Swody, C. A. (2002). How do concepts <strong>of</strong> age relate to work and<br />
<strong>of</strong>f-the-job stresses and strains? A field study <strong>of</strong> health care workers in five nations. Experimental<br />
Aging Research, 28, 87 - 98.<br />
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychology<br />
research: Conceptual, strategic, and statistical considerations. Journal <strong>of</strong> Personality and Social<br />
Psychology, 51, 1173-1182.<br />
Bobko, N. A., & Barishpolets, A. T. (2002). Work ability, age and its perception, and other related<br />
concerns <strong>of</strong> Ukraine health care workers. Experimental Aging Research, 28, 59 - 71.<br />
Bono, J. E., & Judge, T. A. (2003). Self-concordance at work: understanding the motivational effects <strong>of</strong><br />
transformational leaders. Academy <strong>of</strong> Management Journal, 46, 554-571.<br />
Brayfield, A. H., & Rothe, H. F. (1951). An <strong>Index</strong> <strong>of</strong> Job Satisfaction. Journal <strong>of</strong> Applied Psychology, 35,<br />
307-311.<br />
Brush, D. H., Moch, M. K., & Pooyan, A. (1987). Individual demographic differences and job<br />
satisfaction. Journal <strong>of</strong> Occupational Behavior, 8, 139 - 156.<br />
Cleveland, J. N., & Shore, L. M. (1992). Self and supervisory perspective on age and work attitudes and<br />
performance. Journal <strong>of</strong> Applied Psychology, 77, 469-484.<br />
Cleveland, J. N., Shore, L. M., & Murphy, K. R. (1997). Person-and context-oriented perceptual age<br />
measures: Additional evidence <strong>of</strong> distinctiveness and usefulness. Journal <strong>of</strong> Organizational<br />
Behavior, 18, 239-251.<br />
Cloutier, E., Lefebvre, S., Ledoux, É., Chatigny, C., & St-Jacques, Y. (2002). Enjeux de santé et de<br />
sécurité au travail dans la transmission des savoirs pr<strong>of</strong>essionnels: Le cas des usineurs et des<br />
cuisiniers. Montreal, QC: IRSST (Institut de recherche Robert Sauvé en santé et en sécurité du<br />
travail).<br />
Conlin, M. (2003). Grandpa? He's busy at the <strong>of</strong>fice. Business Week Online, News Analysis (January 3,<br />
2003). Retrieved December 14 2005, from<br />
http://businessweek.com/bwdaily/dnflash/jan2003/nf2003013_1240.htm<br />
Crampton, S., Hodge, J., & Mishra, J. (1996). Transition - ready or not: The aging <strong>of</strong> America's work<br />
force. Public Personnel Management, 25(2), 243-256.<br />
Critchley, R. K. (2004). Doing nothing is not an option! Facing the imminent labor crisis. Mason, OH:<br />
South-Western.<br />
DeLong, D. W. (2004). Lost Knowledge. Oxford: Oxford University Press.<br />
122
Dormann, C., & Zapf, D. (2001). Job satisfaction: A meta-analysis <strong>of</strong> stabilities. Journal <strong>of</strong><br />
Organizational Behavior, 22(5), 483 - 504.<br />
Forte, C. S., & Hansvick, C. L. (1999). Applicant age as a subjective employability factor: A study <strong>of</strong><br />
workers over and under age fifty. Journal <strong>of</strong> Employment Counseling, 36, 24 - 35.<br />
Fullerton, H. N. J. (1999). Labor Force Projections to <strong>2008</strong>: Steady Growth and Changing Composition.<br />
Monthly Labor Review, 122(11), 19-33.<br />
Gibson, J. L., & Klein, S. M. (1970). Employee attitudes as a function <strong>of</strong> age and length <strong>of</strong> service: A<br />
reconceptualization. Academy <strong>of</strong> Management Journal, 13, 411-425.<br />
Glass, A. (2007). Understanding generational differences for competitive success. Industrial and<br />
Commercial Training, 39, 98.<br />
Greller, M. M., & Stroh, L. K. (1995). Careers at midlife and beyond: A fallow field in need <strong>of</strong><br />
sustenance. Journal <strong>of</strong> Vocational Behavior, 47, 232-247.<br />
Hassell, B. L., & Perrewe, P. L. (1995). An examination <strong>of</strong> beliefs about older workers: Do stereotypes<br />
still exist? Journal <strong>of</strong> Organizational Behavior, 16, 457- 468.<br />
Hughes, D., C., Blazer, D. C., & George, L. K. (1988). Age differences in life events: A multivariate<br />
controlled analysis. International Journal <strong>of</strong> Aging and Human Development, 27(3), 207-220.<br />
Iskra-Golec, I. (2002). Personal age and assessment <strong>of</strong> work stress in Polish nurses. Experimental Aging<br />
Research, 28, 51 - 58.<br />
Judge, T. A., Bono, J. E., & Locke, E. A. (2000). Personality and job satisfaction: The mediating role <strong>of</strong><br />
job characteristics. Journal <strong>of</strong> Applied Psychology, 85(2), 237 - 249.<br />
Kalleberg, A. L., & Loscocco, K. A. (1983). Aging, values and rewards: Explaining age differences in job<br />
satisfaction. American Sociological Review, 48, 78-90.<br />
Kanfer, R., & Ackerman, P. L. (2004). Aging, adult development, and work motivation. Academy <strong>of</strong><br />
Management Review, <strong>29</strong>, 440-458.<br />
Kline, T. J. B., Sulsky, L. M., & Rever-Moriyama, S. D. (2000). Common method variance and<br />
specification errors: A practical approach to detection. The Journal <strong>of</strong> Psychology, 134, 401-421.<br />
Lawrence, B. S. (1987). An organizational theory <strong>of</strong> age effects. In N. DiTomaso (Ed.), Research in the<br />
sociology <strong>of</strong> organization (Vol. 5, pp. 37-71). Greenwich, CT: JAI Press.<br />
Lawrence, B. S. (1988). New Wrinkles in the Theory <strong>of</strong> Age: Demography, Norms, and Performance<br />
Ratings. Academy <strong>of</strong> Management Journal, 31, 309-337.<br />
Lemme, B. H. (1999). Development in Adulthood (2nd Ed.). Boston, MA: Allyn & Bacon.<br />
Levinson, D. J., Darrow, C. N., Klein, E. B., Levinson, M. H., & McKee, B. (1978). The seasons <strong>of</strong> a<br />
man's life. New York: Alfred A. Knopf.<br />
Levy, B. (2003). Mind matters: Cognitive and physical effects <strong>of</strong> aging self-stereotypes. Journal <strong>of</strong><br />
Gerontology, 58B(4), 203 - 211.<br />
Lindgren, A. (2007). If 70 is the new 60, what’s your next career? Retrieved December 10, 2007, from<br />
http://www.retirement-retirement.com/story-436.php<br />
Montepare, J. M., & Lachman, M. E. (1989). "You're only as old as you feel": Self-perceptions <strong>of</strong> age,<br />
fears <strong>of</strong> aging, and life satisfaction from adolescence to old age. Psychology and Aging, 4(1), 73 -<br />
78.<br />
Nuttman-Schwartz, O. (2004). Like a High Wave: Adjustment to Retirement. The Gerontologist, 44(2),<br />
2<strong>29</strong>-236.<br />
Peterson, C. (1996). The ticking <strong>of</strong> the social clock: Adults' beliefs about the timing <strong>of</strong> transitional events.<br />
International Journal <strong>of</strong> Aging and Human Development, 42(3), 189 - 203.<br />
Rhodes, S. R. (1983). Age-related differences in work attitudes and behavior: A review and conceptual<br />
analysis. Psychological Bulletin, 93, 323-367.<br />
Rubin, D. C., & Berntsen, D. (2006). People over 40 feel 20% younger than their age: Subjective age<br />
across the lifespan. Psychonomic Bulletin & Review, 13(5), 776 - 780.<br />
Shaw, J. B., & Grubbs, L. L. (1981). The process <strong>of</strong> retiring: Organizational entry in reverse. Academy <strong>of</strong><br />
Management Review, 6, 41-47.<br />
123
Smith, P., Kendall, L., & Hulin, C. (1969). The measurement <strong>of</strong> satisfaction in work and retirement.<br />
Chicago, IL: Rand McNally.<br />
Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equation models. In S. Leinhart<br />
(Ed.), Sociological Methodology. San Francisco, CA: Jossey-Bass.<br />
Spector, P. E. (1997). Job satisfaction: Application, assessment, causes and consequences. London: Sage.<br />
Super, D. E. (1957). The psychology <strong>of</strong> careers. New York: Harper.<br />
Uotinen, V., Rantanen, T., Suutama, T., & Ruoppila, I. (2006). Change in subjective age among older<br />
people over an eight-year follow-up: "Getting older and feeling younger?" Experimental Aging<br />
Research, 32(4), 381 - 393.<br />
White, A. T., & Spector, P. E. (1987). An investigation <strong>of</strong> age-related factors in the age-job-satisfaction<br />
relationship. Psychology and Aging, 2(3), 201 - 265.<br />
Williams Walsh, M. (2001, Feb 26, 2001). Reversing Decades-Long Trend, Americans Retiring Later in<br />
Life. New York Times, p. A1.<br />
Wright, H., & Hamilton, R. (1978). Work satisfaction and age: Some evidence for the job change<br />
hypothesis. Social Forces, 59(4), 1140-1158.<br />
124
ABSTRACTS / RÉSUMÉS<br />
ASAC <strong>2008</strong> Mardi Witzel (student)<br />
Halifax, Nova Scotia School <strong>of</strong> Business and Economics<br />
Wilfrid Laurier University<br />
LEADERSHIP AND CREATIVITY: THE MEDIATING EFFECTS OF INTRINSIC<br />
MOTIVATION, DOMAIN-RELEVANT SKILLS, AND CREATIVITY-RELEVANT<br />
PROCESSES<br />
The literature on organizational behavior seeks to identify the human and contextual dynamics<br />
that explain creativity. This research investigates the role <strong>of</strong> transactional and transformational<br />
leadership in predicting creative outcomes, with intrinsic task motivation, domain-relevant skills<br />
and creativity-relevant processes considered as mediators. A conceptual model and hypotheses<br />
are presented, with plans for empirical testing.<br />
ASAC <strong>2008</strong> Sara L. Mann<br />
Halifax, Nova Scotia Department <strong>of</strong> Business<br />
University <strong>of</strong> Guelph<br />
GaryP. Latham<br />
Rotman School <strong>of</strong> Management<br />
University <strong>of</strong> Toronto<br />
Who receives a performance appraisal and does it matter? An empirical investigation <strong>of</strong><br />
the determinants <strong>of</strong> the receipt <strong>of</strong> a performance appraisal and its effect on job satisfaction<br />
An analysis <strong>of</strong> Statistic Canada’s Workplace and Employee Survey (WES), based on a five-year<br />
period, revealed that only 60% <strong>of</strong> the 20,834 respondents received a performance appraisal.<br />
Higher levels <strong>of</strong> job satisfaction were reported by individuals who perceived that their appraisal<br />
impacted their pay, benefits and most recent promotion.<br />
150
ASAC <strong>2008</strong> Saba Sharih (PhD Student)<br />
Halifax, Nova Scotia Simon Taggar<br />
School <strong>of</strong> Business & Economics<br />
Wilfrid Laurier University<br />
STRATEGIC RESPONSIVENESS WITHIN AN IMSA FRAMEWORK-<br />
INSTITUTIONAL AND RESOURCE BASED PERSPECTIVES<br />
Organizational scholars have conceptualized organizational responsiveness to institutional<br />
pressures as a strategic choice. This thesis highlights a number <strong>of</strong> institutional and resourcebased<br />
determinants <strong>of</strong> this critical strategic choice from an intrafirm perspective. Cause,<br />
constituents and content are proposed as antecedents <strong>of</strong> strategic responses employed by the<br />
inner core employee segment to pressures exerted by other employee segments within an<br />
organization. Implications for theory building and practice are discussed.<br />
ASAC <strong>2008</strong> Christa L. Austin (student)<br />
Halifax, Nova Scotia Catherine E. Connelly<br />
Human Resources and Management<br />
DeGroote School <strong>of</strong> Business, McMaster University<br />
RECRUITERS’ PERCEPTIONS OF APPLICANTS’ VOLUNTEER EXPERIENCE: ARE ALL<br />
VOLUNTEER ASSIGNMENTS EQUALLY RELEVANT?<br />
An ongoing debate exists in the practitioner literature as to how recruiters interpret and use<br />
applicants’ volunteer experiences. This study found that recruiters perceived certain volunteer<br />
experiences to be more indicative <strong>of</strong> various desirable traits (i.e., motivation, interpersonal skills,<br />
leadership qualities). Practical implications and directions for future research are discussed.<br />
151
ASAC <strong>2008</strong> Travor C. Brown<br />
Halifax, NS Memorial University<br />
Martin McCracken<br />
University <strong>of</strong> Ulster<br />
Sergio De Leon (student)<br />
Memorial University<br />
BARRIERS TO TRANSFER: PARTICIPANT’S PERSPECTIVE: PRELIMINARY RESULTS<br />
Transfer <strong>of</strong> training is a critical issue for training programs. While considerable quantitative work has<br />
examined barriers to transfer, few studies have assessed transfer from the perspective <strong>of</strong> traineesthemselves.<br />
In this paper, we review open-ended qualitative data, from union and managerial participants<br />
enrolled in a leadership skills program, to examine barriers to transfer.<br />
ASAC <strong>2008</strong> Sunny Marche, PhD, CMC<br />
HALIFAX, NOVA SCOTIA John F. Duffy, PhD<br />
Geneviève Perron (Ph.D. Candidate)<br />
Faculty <strong>of</strong> Management / Faculty <strong>of</strong> Graduate Studies<br />
Dalhousie University<br />
SERIOUS THREATS TO CANADIAN UNIVERSITY FACULTY MEMBERS –SURVEY<br />
OUTCOMES<br />
Serious workplace threats have been reported for doctors, nurses, judges, lawyers<br />
and elementary school teachers. This is the first survey <strong>of</strong> workplace threats for<br />
university pr<strong>of</strong>essors in Canada, and to the best <strong>of</strong> our knowledge, in the world.<br />
We report on a selection <strong>of</strong> existing related literature and the results from an<br />
online survey.<br />
152
ASAC <strong>2008</strong> James Doyle<br />
Halifax, Nova Scotia Alia El Banna<br />
Linda Duxbury<br />
Chris Higgins<br />
Carlton University<br />
A STATISTICAL INVESTIGATION OF ANTECEDENTS AND OUTCOMES OF JOB-<br />
RELATED STRESS FOR TEACHERS<br />
Developing a healthy workplace is an important concern for organizations, as the implications <strong>of</strong><br />
job-related stress are extensive and have negative effects on both individuals and organizations.<br />
This research study explores job-related stress factors and the mental-health outcomes<br />
experienced by a sample <strong>of</strong> school teachers in Ontario.<br />
ASAC <strong>2008</strong><br />
Halifax, Nova Scotia<br />
Shawn Komar (student)<br />
Douglas J. Brown<br />
Department <strong>of</strong> Psychology<br />
University <strong>of</strong> Waterloo<br />
Chet Robie<br />
School <strong>of</strong> Business and Economics<br />
Wilfrid Laurier University<br />
EXPLORING FAKING IN THE CONTEXT OF MULTI-STAGE SELECTION SYSTEMS<br />
The current study examines the impact that faking on a conscientiousness measure has on<br />
performance in multi-stage selection contexts. The results suggest that combining personality<br />
measures with other measures will not mitigate the negative effect <strong>of</strong> faking. The results are<br />
discussed in terms <strong>of</strong> the implications for employee selection.<br />
153
ASAC <strong>2008</strong> Akanksha Bedi (student)<br />
Halifax, Nova Scotia Mark S. Skowronski (student)<br />
Dr. Aaron C. H. Schat<br />
DeGroote School <strong>of</strong> Business<br />
McMaster University<br />
POLITICAL SKILL: A META –ANALYSIS OF ITS<br />
PREDICTORS AND OUTCOMES<br />
This paper describes a meta-analysis <strong>of</strong> the antecedents, correlates, and outcomes <strong>of</strong> political skill. The<br />
results suggest negligible relationships between political skill and demographic variables and moderate<br />
relationships with job satisfaction and job performance. Practical and research implications <strong>of</strong> the findings<br />
are discussed.<br />
ASAC <strong>2008</strong> Wendy R. Carroll<br />
Halifax, Nova Scotia Ph.D. Student (ABD<br />
Saint Mary’s University<br />
Terry H. Wagar<br />
Saint Mary’s University<br />
Kent V. Rondeau<br />
University <strong>of</strong> Alberta<br />
EXPLORING THE RELATIONSHIP BETWEEN ORGANIZATIONAL CULTURE AND<br />
QUIT BEHAVIOUR: EVIDENCE FROM CANADIAN CALL CENTRES<br />
Call centres are integral business operations for many organizations that have become<br />
characterized by high levels <strong>of</strong> voluntary turnover. In this paper, we examine the relationship<br />
between quit behaviour and organizational culture using data from a national survey <strong>of</strong> call centres<br />
and a case study <strong>of</strong> employees at one call centre. We found that call centres with a more social<br />
organizational culture type (clan) were associated with lower quit rates and longer periods <strong>of</strong><br />
employee intention to stay with the organization. However, call centres with a more market<br />
oriented culture type were related to higher quit rates and shorter periods <strong>of</strong> intended stay with an<br />
organization.<br />
154
ASAC <strong>2008</strong> Jim D. Grant (PhD Candidate)<br />
Halifax, Nova Scotia School <strong>of</strong> Business and Economics<br />
Nipissing University<br />
Terry H. Wagar<br />
Sobey School <strong>of</strong> Business<br />
Saint Mary’s University<br />
DEVELOPMENTS IN CANADIAN EMPLOYMENT LAW AND THE SUPREME COURT OF<br />
CANADA: EVIDENCE FROM AN EXAMINATION OF REASONABLE NOTICE IN THE<br />
WRONGFUL DISMISSAL COMMON LAW<br />
This study examined the determinants <strong>of</strong> reasonable notice in wrongful dismissal law in a way which<br />
incorporates the effect <strong>of</strong> Supreme Court <strong>of</strong> Canada (SCC) decisions such as in Wallace v. United Grain<br />
Growers (1997). We explored the relationship between reasonable notice awards and employees’<br />
performance, vulnerability, and changing employment circumstances, as well as employers’ allegations <strong>of</strong><br />
misconduct over and above poor performance and failure to employ progressive discipline. In addition,<br />
we wanted to determine if the Wallace decision, in which the SCC affirmed the bargaining inequality<br />
between employee and employer, has had an effect on notice periods in general.<br />
155