LEVEL AND DISPERSION OF SATISFACTION IN TEAMS: USING

LEVEL AND DISPERSION OF SATISFACTION IN TEAMS:USING FOCI AND SOCIAL CONTEXT TO EXPLAIN THE

SATISFACTION-ABSENTEEISM RELATIONSHIP

BRIAN R. DINEENUniversity of Kentucky

RAYMOND A. NOEThe Ohio State University

JASON D. SHAWMICHELLE K. DUFFY

University of Minnesota

CAROLYN WIETHOFFIndiana University

We develop an integrative theory regarding the effects of mean levels and dispersion ofsatisfaction predicting absenteeism. Differential interactive predictions are derived fortwo satisfaction foci and tested in two distinct samples. Among student teams, absen-teeism from team meetings was highest when team (internally focused) satisfactionmean and dispersion were both lower, but low when course (externally focused)satisfaction mean and dispersion were both lower. Moreover, given lower dispersion,the mean team satisfaction-absenteeism relationship appeared stronger, whereas thesame relationship involving course satisfaction appeared weaker than meta-analyzedindividual-level relationships. We replicated these results among manufacturingteams using team and job satisfaction foci.

The nature of the relationship between satisfac-tion and withdrawal behaviors such as absenteeismcontinues to captivate researchers. Yet most of thestudies of the satisfaction-absence relationshiphave been conducted at the individual level, withmeta-analyses indicating a negative and moderaterelationship (Hackett, 1989; Harrison, Newman, &Roth, 2006). Although informative, these cumula-tive results and near-singular focus on the individ-ual level leave at least two fundamental issues un-resolved. First, despite wide acceptance of thepremises that workplace interactions play a sub-stantial role in the development of individuals’ at-titudes and behaviors (e.g., Bliese & Halverson,1998; Salancik & Pfeffer, 1978) and that absentee-ism has social and normative roots (e.g., Gellatly,1995; Harrison & Martocchio; Mathieu & Kohler,1990), the literature generally fails to capture the

social-contextual underpinnings of the satisfaction-absenteeism relationship, prompting a need formore thorough investigations (Johns, 2006).

Second, and in line with a social-contextual per-spective, it is uncertain whether the satisfaction-absenteeism findings obtained at the individuallevel would be expected to materialize at higherlevels of analysis, such as the team level. Led byMorgeson and Hofmann (1999), Kozlowski andKlein (2000), and Chen, Mathieu, and Bliese (2004),there is a growing interest in the question ofwhether typical individual-level findings are rep-resentative of corresponding higher-level relation-ships. Concerning satisfaction relationships, Judge,Thoresen, Bono, and Patton wrote, “Comparing therelative predictive power of job satisfaction at var-ious levels of analysis would be a worthwhile topicfor future research, as would further theoreticaldevelopment underlying expected differences”(2001: 391).

At first glance, it might seem reasonable to expecta vertical synthesis of the satisfaction-absenteeismrelationship—that is, mean levels of satisfaction inteams would relate negatively to mean levels ofabsenteeism in teams. However, Ostroff and Harri-

We thank Jason Colquitt, Dan Brass, Philip Podsakoff,Krish Muralidhar, Jackie Thompson, Associate EditorBrad Kirkman, and three anonymous reviewers for theirassistance. Partial funding was provided by the Teachingand Learning Center and the Forester Professorship fundat the University of Kentucky.

� Academy of Management Journal2007, Vol. 50, No. 3, 623–643.

623

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download or email articles for individual use only.

son (1999) argued that social or normative teamfactors might cause this relationship to differ at theteam level of analysis, whereas Rentsch and Steel(2003) dissuaded researchers from assuming thatindividual-level absence theories would necessar-ily transfer across levels. For example, individualsenjoy interacting with others who hold similar psy-chological characteristics such as attitudes, andthis attraction may occur whether shared attitudesare negative or positive (Harrison, Price, Gavin, &Florey, 2002). Thus, although research indicates anegative relationship between satisfaction and ab-senteeism at the individual level, a common levelof dissatisfaction among team members may renderthose members less rather than more prone to ab-senteeism if they develop a sense of unity in theface of the dissatisfying stimulus. Indeed, it is pos-sible that some of the individuals included in meta-analyzed studies have indicated low job satisfac-tion but have worked alongside others withsimilarly low job satisfaction. Common in-groupdynamics (Brewer & Brown, 1998) may have atten-uated absenteeism levels in these studies, leadingto attenuated meta-analyzed relationships betweensatisfaction and absenteeism. It is clear, then, thatboth the mean level of satisfaction and the distri-bution or dispersion of satisfaction are important toconsider if a social-contextual view of the satisfac-tion-absenteeism relationship is to be taken.

Thus, new theory development and empiricaltesting that accounts for the social context in whichsatisfaction is experienced by investigating rela-tionships at a higher level of analysis is warranted.We take steps in this direction here. Specifically,we address whether the negative but moderate re-lationship found in individual-level meta-analyses(Hackett, 1989; Harrison et al., 2006) extrapolates tothe team level and, further, we explore how levelsof similarity might differ depending on satisfactionfoci and levels of dispersion. Previous theoretical(Lindsley, Brass, & Thomas, 1995) and empirical(Kirkman & Rosen, 1999; Prussia & Kinicki, 1996)work has addressed likely similarities in the func-tion of constructs such as empowerment, efficacy,and affective evaluations across levels of analysis.In keeping with these studies, our focus is not on anexplicit within-study comparison of individual-and team-level relationships. Rather, we adoptfunctional homology as a guiding perspective (e.g.,Chen et al., 2004) and attend to how homology, orsimilarity in the functional relationship betweensatisfaction and absenteeism across levels, is likelyto manifest. Specifically, we compare our team-level results with the individual-level results pre-viously obtained via meta-analysis. Thus, we spe-cifically assess the function of satisfaction in

predicting absenteeism at the team level of analysiswhen dispersion levels and foci are taken into ac-count, rather than how its specific structure may ormay not differ (Morgeson & Hofmann, 1999).

We define dispersion as differences in satisfac-tion among team members, a definition that is con-sistent with Chan’s (1998) typology. According toChan (1998), dispersion is a group-level propertyconsisting of variability originating at the individ-ual level that may include mutable underlying at-tributes (e.g., satisfaction levels; see also Jackson,May, and Whitney [1995] and Milliken and Martins[1996]). This view is also consistent with researchconceptualizing within-team attitudinal differ-ences as a form of “deep-level,” or nonvisible, di-versity (Harrison, Price, & Bell, 1998; Kirkman &Shapiro, 2005). We define and differentiate satis-faction foci following Siders, George, and Dhar-wadkar’s (2001) dichotomy of internal/externalcommitment foci. That is, we argue that the jointeffects of mean and dispersion levels of team mem-ber satisfaction with fellow teammates (internallyfocused) and team member satisfaction with otherpeople or entities exclusive of fellow teammates(externally focused) differentially predict team ab-senteeism levels.

The article is organized as follows. First, we re-view research that has examined attitudinal meansand dispersion in relative isolation and argue forthe importance of considering them simulta-neously when examining relationships involvingsatisfaction at the team level of analysis. Second, toprovide additional clarity and integration, we arguethat the nature of the satisfaction-absenteeism rela-tionship differs depending on internal or externalsatisfaction focus. Third, we develop an integrativeframework and hypothesize how mean and disper-sion levels of internally focused (team) and exter-nally focused (course or job) satisfaction levels re-late to absenteeism. This framework suggests thatlevels of functional homology with meta-analyzedindividual-level relationships vary depending onlevels of dispersion and satisfaction foci. Fifth, wereport tests of the hypotheses in a study of class-room teams and replication tests using manufactur-ing teams. Finally, we discuss implications of theresults for theory and management practice.

THEORETICAL FOUNDATION ANDHYPOTHESIS DEVELOPMENT

As noted above, meta-analytic evidence suggeststhat satisfaction and absenteeism are negatively re-lated at the individual level (Hackett, 1989; Harri-son et al., 2006). Although this relationship hasreceived little research attention at the team level,

624 JuneAcademy of Management Journal

there is similar reason to expect that average ormean satisfaction levels within a team will relatenegatively to mean levels of absenteeism. For ex-ample, Ostroff (1992) and Ryan, Schmit, and John-son (1996) found positive relationships betweenmean levels of satisfaction and unit-levelperformance.

The research on attitudinal dispersion in teamsalso provides a rather consistent pattern of evi-dence. For example, the deep-level diversity per-spective suggests that lower levels of satisfactiondispersion are associated with greater cohesive-ness, with relationships being stronger among long-er-tenured teams (Harrison et al., 1998). Teamswith lower dispersion on associated characteristicsare also argued to have greater social integration, anenhanced sense of common in-group identity, andreduced fragmentation (Boone, Van Olffen, & VanWitteloostuijn, 2005; Brewer & Brown, 1998; Har-rison et al., 2002), all factors that should be associ-ated with lower team absenteeism levels (Gellatly,1995; Somers, 1995).

Few studies simultaneously examine mean lev-els and dispersion of team characteristics, despitegrowing recognition that models may be under-specified if both are not incorporated (e.g., Kirkman& Shapiro, 2005). Indeed, from a broad perspective,the results from studies focusing only on meanattitudinal levels (Ostroff, 1992) or only on disper-sion (e.g., Harrison et al., 1998) are not fully com-patible and have somewhat paradoxical (Poole &Van De Ven, 1989) predictions. On the one hand,the literature suggests that lower mean levels ofsatisfaction in teams should be associated with in-ferior team-level outcomes, including higher ab-sence levels. Yet research also suggests that lowerdispersion or greater agreement about satisfactionis associated with more optimal group outcomes.This latter conclusion holds presumably for allmean levels of satisfaction (low to high), a view thatis not fully compatible with the mean-level results.An open question, for example, is what absentee-ism levels would be expected when mean levelsand dispersion of satisfaction within a team aresimultaneously low. Addressing this issue, Weeksrecently made the intriguing suggestion that lowerlevels of job satisfaction may not always lead tonegative outcomes such as absenteeism, because ifteam members share a sense of job dissatisfaction,“these complaints can help strengthen social bondsand build a sense of community” (2004: 20).

A plausible reason for this lack of integration isthe fact that many unit-level studies are based on adirect consensus approach to aggregation, wherebyminimum standards of within-group agreementmust be met for aggregation (Gonzalez-Roma, Peiro,

& Tordera, 2002; Kozlowski & Klein, 2000; Ostroff& Bowen, 2000). Although informative, these stud-ies ignore potential effects of attitudinal dispersionor “climate strength” by minimizing within-groupvariation in predictors (Chan, 1998). A growingnumber of authors have argued that direct consen-sus models have hidden the substantive impor-tance of dispersion in predicting work-related out-comes (Bliese & Halverson, 1998; Boone et al.,2005; Brown, Kozlowski, & Hattrup, 1996; Chan,1998; Kirkman & Shapiro, 2005).

In a team context, individual satisfaction levelsare experienced relative to the satisfaction levels ofother team members. Because satisfaction isthought to be influenced by both social-contextualand dispositional elements (e.g., Bowling, Beehr,Wagner, & Libkuman, 2005; Staw, Bell, & Clausen,1986), these satisfaction levels may be widely ornarrowly dispersed. Hackman (1992) argued, forexample, that team members may have differentsatisfaction levels because the stimuli they experi-ence are either ambient or discretionary. In partic-ular, ambient stimuli (commonly experienced byeveryone on a team) should lead to less dispersionin satisfaction levels, whereas discretionary stimuli(experienced differently by individuals on a team)might lead to greater dispersion in satisfaction.Other scholars have focused on satisfaction’s dis-positional roots (Staw et al., 1986), which suggestsa preexisting level of natural dispersion amongteam members. Still others have viewed social in-formation processing (Salancik & Pfeffer, 1978) orcontagion processes (Mason & Griffin, 2002) as re-sulting in decreased satisfaction dispersion inteams over time. For example, Bowling and col-leagues recently highlighted the stability of job sat-isfaction yet went on to claim that it is influencedby both dispositions and workplace events. Giventhese dispositional and social-contextual determi-nants, the need to examine both mean and disper-sion levels of satisfaction is apparent. Our theoret-ical development accounts for both simultaneouslyand suggests conditions under which different lev-els of homology might be expected in comparingteam-level satisfaction-absenteeism relationshipsto typical individual-level relationships (Harrisonet al., 2006).

In particular, by examining the satisfaction-ab-senteeism relationship at the team level, we extend(1) research linking mean levels of satisfaction toaggregate outcomes (e.g., Ostroff, 1992), (2) deep-level diversity research (e.g., Harrison et al., 1998),(3) team climate research (e.g., Schneider, Salvag-gio, & Subirats, 2002), and (4) theories of commonin-group identity or “community of fate” (e.g.,Brewer & Brown, 1998; Shaw, Duffy, & Stark, 2000).

2007 625Dineen, Noe, Shaw, Duffy, and Wiethoff

A key to integrating these literatures lies in theconsideration of attitudinal focus (i.e., internal orexternal). A growing body of work suggests thenecessity of differentiating between internal andexternal attitudinal foci (e.g., Bishop & Scott, 2000;Siders et al., 2001). For example, Siders and col-leagues (2001) found that employees’ commitmentto their supervisor and organization related to or-ganizationally relevant performance outcomes,whereas employees’ commitment to customers re-lated to customer-relevant outcomes. The authorsviewed their distinction between commitment focias useful in explaining the typically weak relation-ships found between broader commitment mea-sures and performance. Building on Becker’s (1992)work, Clugston, Howell, and Dorfman (2000) alsoargued that it is necessary to differentiate not onlyamong various bases, but also among various foci ofcommitment.

We propose two possible patterns of relation-ships between absenteeism and mean and disper-sion levels of team member satisfaction. Figure 1graphically depicts the two proposed patterns. Spe-cifically, Figure 1a shows the pattern that is morepredictive of absenteeism when internally focusedsatisfaction is considered, whereas Figure 1b’s pat-tern is more predictive when externally focusedsatisfaction is considered. In terms of our homologyperspective, Figure 1a (an internal satisfaction fo-cus) suggests that when dispersion is lower, astronger relationship between satisfaction and ab-senteeism is expected at the team level than hastypically been found at the individual level (e.g.,Hackett, 1989), whereas a weaker relationship thatis more homologous with individual-level findingsis expected when dispersion is higher. Figure 1balso suggests, similarly to individual-level find-ings, a weak relationship between mean levels ofsatisfaction and absenteeism when dispersion ishigher. However, satisfaction and absenteeism arenot expected to be related when dispersion islower, suggesting a lack of homology with what hastypically been found at the individual level.

In team-based work situations, individualsclearly develop attitudes toward their fellow teammembers as well as toward their broader job ororganization. Thus, in Study 1 (a sample of studentteams), we examined team members’ satisfactionwith others on their team (internally focused satis-faction) as well as their satisfaction with the coursethey were taking (externally focused). In Study 2(manufacturing teams), we again examined teammembers’ satisfaction with others on their team(internally focused) and also examined members’satisfaction with their jobs (externally focused). In

the ensuing sections, we discuss how prior theoryand research support the predicted patterns.

Internally Focused Satisfaction Configurationsand Absenteeism

The few studies that have looked at both meanand dispersion levels of team characteristics havedone so primarily through a team climate lens (e.g.,Colquitt, Noe, & Jackson, 2002; Gonzalez-Roma etal., 2002; Schneider et al., 2002). This researchaddresses the possibility that lower dispersion ofteam characteristics (i.e., greater climate strength)augments the relationship between mean levelsand outcomes. For example, Colquitt et al. (2002)found that a strong procedural justice climate (lowdispersion of justice) heightened the negative ef-fects of low mean levels of procedural justice. Thisexample, and others (e.g., Boone et al., 2005;George, 1990) provides evidence suggesting thenegative relationship between satisfaction and ab-senteeism will be homologous at the team level andthat this relationship will be even stronger whensatisfaction dispersion is low. More generally, thisresearch is similar to the classic work showing thatthe relationship between cohesion and team perfor-mance is not straightforward (Stodgill, 1972) butdepends on task norms, with cohesion leading tomore negative performance outcomes if task normsare low (e.g., Gammage, Carron, & Estabrooks, 2001;Goodman, Ravlin, & Schminke, 1987).

A closer look at these studies, however, reveals akey to their interpretation. Specifically, theColquitt et al. (2002) study and other similar stud-ies have tended to focus on perceptions related tofellow team members. Extending these findings, weargue that this pattern—stronger relationships be-tween satisfaction and absenteeism when disper-sion is lower—will hold only when internally fo-cused satisfaction (i.e., team satisfaction) isconsidered. For example, per Figure 1a, if teammembers all tend to be dissatisfied with each other(lower team satisfaction mean, lower dispersion),the least amount of social integration seems likely,and it is reasonable to expect an augmentation ef-fect of dispersion on means that is such that absen-teeism among those team members will be rela-tively higher. That is, although the deep-leveldiversity literature would imply higher levels ofsocial integration when satisfaction diversity is low(e.g., Harrison et al., 2002), we argue that consensusdoes not always translate into solidarity, and in thiscase social integration should be lower because thelower satisfaction dispersion concerns negative at-titudes toward fellow teammates. In this scenario,lower mean levels of team satisfaction should be


associated with higher absenteeism. In contrast,when teammates all tend to report high levels ofsatisfaction with their team (higher mean, lowerdispersion), social integration should be higher, re-sulting in lower absenteeism.

Finally, evidence from team climate researchsuggests that the relationship between mean levelsof internally focused satisfaction and absenteeismwill be weaker when satisfaction dispersion among

teammates is higher, a finding that would be con-sistent with typical individual-level findings. Inparticular, a higher level of team satisfaction dis-persion will result in a moderately negative satis-faction-absenteeism relationship (see the top twocells of Figure 1a). Thus, we propose:

Hypothesis 1. Mean and dispersion levels ofinternally focused (i.e., team) satisfaction ex-

FIGURE 1Absenteeism Levels Predicted by Internally and Externally Focused Satisfaction Configurations in Teams,

with Suggested Levels of Associated Functional Homology


hibit a joint relationship with absenteeism thatis such that mean team satisfaction exhibits aweaker, negative relationship with absentee-ism when dispersion is higher and a stronger,negative relationship with absenteeism whendispersion is lower.

Externally Focused Satisfaction Configurationsand Absenteeism

Despite advances in the team climate literature,its basic findings are not fully compatible with themain effects of attitudinal agreement proposed inother literatures (e.g., the deep-level, or nonvisible,diversity literature). In particular, we expect thecombined effects of mean and dispersion levels ofexternally focused satisfaction (i.e., course or jobsatisfaction) to produce a differential pattern offindings (see Figure 1b).

First, we propose, in contrast to the augmenta-tion effect resulting from lower internally focusedsatisfaction dispersion, a neutralization effect, re-sulting from lower externally focused satisfactiondispersion, on the relationship between mean ex-ternally focused satisfaction and absenteeism, sothat social integration is higher and absenteeismlower over all mean levels of satisfaction whendispersion is lower. In particular, the lower leftquadrant of Figure 1b shows a predicted lowerlevel of absenteeism when mean levels of exter-nally focused (i.e., job or course) satisfaction arehigher and dispersion is lower. This combinationsuggests that lower absenteeism results from favor-able evaluation of a context (e.g., higher mean jobsatisfaction), stronger in-group identity resultingfrom a common attitude toward an external entity,and increased social integration resulting from alack of deep-level diversity.

Unlike Figure 1a, Figure 1b also shows predictedlow absenteeism when mean levels and dispersionof externally focused satisfaction are both lower(lower right quadrant). This prediction is groundedin the deep-level diversity and common in-groupidentity literatures (Brewer & Brown, 1998; Harri-son et al., 1998). In addition, the community of fateperspective suggests that lower dispersion neutral-izes mean-level effects, despite lower mean satis-faction levels (Shaw et al., 2000). When individualswithin a team hold uniformly unfavorable attitudestoward an external entity that exists beyond theirimmediate experience in the team (such as theirjobs), in-group salience likely increases, and theymay become more committed to each other in anattempt to deal effectively with the common en-emy. This consequence is similar in nature to thelower performance that might result from highly

cohesive teams if performance norms are low. Inessence, this argument suggests that lower job orcourse satisfaction dispersion can “compensate”for lower mean levels of a characteristic (e.g., Bar-sade, Ward, Turner, & Sonnenfeld, 2000; Harrisonet al., 2002; Harrison & Klein, in press). Thus, in-stead of enhancing the negative effects of lowermean levels of externally focused satisfaction, asthe team climate literature would indicate, lowerdispersion should create team solidarity, foster theformation of a stronger in-group identity (Brewer &Brown, 1998), and reduce absenteeism, regardlessof the lower mean satisfaction level. Indeed, Weeks(2004) suggested that people are drawn togetherthrough affirmations of shared suffering. This ideafurther suggests that the potential to commiseratewith teammates acts more strongly than the desireto avoid an aversive situation by withdrawing,which is logical given that withdrawal negatesone’s ability to commiserate with and receive affir-mation from those teammates. In such a case then,lower dispersion ameliorates rather than enhancesthe otherwise deleterious effects of low job satisfac-tion by providing a functional, expressive outlet forthe dissatisfaction. This argument further suggestsa lack of homology with typical individual-levelrelationships when dispersion in externally fo-cused satisfaction is lower.

By contrast, when job or course (externally fo-cused) satisfaction dispersion is higher, commonin-group dynamics do not materialize and, thus,identity and community of fate effects are less ap-plicable. Rather, predictions based on the deep-level diversity literature and literature examiningmean levels of attitudes can be used to generateabsenteeism predictions for these cells rangingfrom moderate to high (see the top two quadrants ofFigure 1b). First, as in the cohesiveness-perfor-mance findings reviewed earlier, moderate absen-teeism is predicted when an externally focused sat-isfaction mean and dispersion are both higher.Specifically, the literature relating mean levels ofsatisfaction to outcomes suggests, on the one hand,a lower level of absenteeism based on the highermean level of course or job satisfaction in a team(e.g., Ostroff, 1992). On the other hand, dispersionin satisfaction is relatively higher, suggesting lowersocial integration and thus higher absenteeism, ac-cording to the deep-level diversity literature.

Combined, these two perspectives suggest over-all moderate absenteeism for the mean and disper-sion levels represented in the top left quadrant ofFigure 1b. Finally, when mean levels of course orjob satisfaction are lower and dispersion is higher(top right quadrant of Figure 1b), higher absentee-ism is predicted. That is, these teams experience


lower social integration owing to the greater deep-level diversity (Harrison et al., 1998). Lower socialintegration along with lower course (job) satisfac-tion contributes to higher absenteeism. Accordingly:

Hypothesis 2. Mean and dispersion levels ofexternally focused (i.e., job or course) satisfac-tion exhibit a joint relationship with absentee-ism that is such that absenteeism is loweracross mean levels of job (course) satisfactionwhen dispersion is lower, whereas mean levelsof job (course) satisfaction exhibit a moder-ately negative relationship with absenteeismwhen dispersion is higher.

STUDY 1 METHODS

Sample and Procedures

Participants in this study were 450 upper-divi-sion undergraduate students enrolled in 11 differ-ent business administration classes at a large south-ern U.S. university. Participants’ average age was22 years, and 63 percent were male. Participantswere guaranteed confidentiality and assured thatparticipation was voluntary. Teams (n � 103) re-mained intact throughout the term and averagedfour members who collaborated on projects in andoutside class, including textbook cases, exercises,and written term projects. Questionnaires were ad-ministered at the end of the course (16 weeks afterteams were formed).

Measures

We operationalized externally focused satisfac-tion as an individual’s evaluation of his or herlevel of satisfaction with the course. Course sat-isfaction was assessed with three items adaptedfrom Cammann, Fichman, Jenkins, and Klesh’s(1983) global job satisfaction measure (seven-point Likert-type scale; � � .89; e.g., “All in all, Iam satisfied with this class”). Team satisfactionwas assessed with three parallel items (� � .85;e.g., “All in all, I am satisfied with my group inthis class”). We assessed the convergent validityof the adapted course satisfaction scale in a pilotstudy of 29 graduate business students. Correla-tions between this measure and measuresadapted from the Minnesota Satisfaction Ques-tionnaire (MSQ; Weiss, Dawis, England, &Lofquist, 1967) and Overall Job Satisfaction Scale(OJS; Brayfield & Rothe, 1951) were .82 and .94,respectively, indicating strong convergence.

Our satisfaction constructs were assessed at theindividual level and used to generate descriptive

team-level satisfaction mean and dispersion vari-ables (Barrick, Stewart, Neubert, & Mount, 1998;Chen et al., 2004). First, mean course and teamsatisfaction were conceptualized as additive team-levelconstructs inChan’s (1998) typologyandopera-tionalized as the mean level of course and teamsatisfaction among team members. We did notadopt Chan’s consensus and referent shift models,which mandate agreement among team members,given our expectation of a certain level of disper-sion due to dispositional or stimulus-based differ-ences among members (Hackman, 1992; Staw et al.,1986). Given this expectation, satisfaction disper-sion was also substantively meaningful accordingto our theorizing, and it was conceptualized underChan’s (1998) dispersion model as a configuralproperty of a team (Kozlowski & Klein, 2000). Inparticular, our conceptualization is what Harrisonand Klein (in press) termed “separation,” meaningdifferences in position along a continuum that rep-resent dissimilarity in characteristics such asattitudes.

Team and course satisfaction dispersion con-structs were operationalized as the standard devi-ation among team member satisfaction levels. Har-rison and Klein (in press) recommended thestandard deviation as an appropriate index for sep-aration-type diversity. Klein, Conn, Smith, andSorra (2001) and Harrison et al. (2002) have alsoprovided convergent validity evidence supportinguse of a standard deviation dispersion index, anduse of the standard deviation is consistent withprocedures followed in several other studies of thistype (e.g., Boone et al., 2005; Kirkman & Shapiro,2005; Schneider et al., 2002). Thus, conceptual andoperational support exists for the construct validityof this dispersion index. Finally, absenteeism wasconceptualized under Chan’s (1998) additivemodel and operationalized at the team level as themean of team member responses to the question,“About how many times did you miss group meet-ings?”

Given that those with heavier course loadsmight have experienced external conflicts thatcaused unintentional absences, we controlled forteam member average course load by asking par-ticipants to indicate the number of hours theywere taking during the semester. We also con-trolled for team size, because it is associated withabsenteeism (Markham, Dansereau, & Alutto,1982), and for gender and ethnic diversity (usingTeachman’s index), given the possible effects ofsurface-level diversity in these relatively short-tenured teams (Harrison et al., 1998).


Analyses

We used polynomial regression analysis (Cohen,Cohen, West, & Aiken, 2003; Edwards & Parry,1993) because of its added precision in examiningjoint relationships between variables (Kristof-Brown, Barrick, & Stevens, 2005; Van Vianen, DeParter, Kristof-Brown, & Johnson, 2004), and be-cause curvilinearity was possible in the relation-ships examined. For example, in regard to climate,Lindell and Brandt noted that “in addition to thefact that climate quality and climate consensus arenot statistically independent of each other, theirjoint effects on organizational outcomes do not nec-essarily take the form of simple linear relations”(2000: 336; see also Van der Vegt & Bunderson,2005). Specifically, the restriction in range ofmeans at higher levels of dispersion suggests thatrelationships between mean levels of satisfactionand absenteeism may be attenuated when higherlevels of dispersion exist, which in turn suggestspossible curvilinearity in the joint relationship ofsatisfaction mean and dispersion and absenteeism.

Cortina (1993) suggested that when the compo-nent terms of an interaction are interdependent orwhen curvilinearity is possible, squared terms cor-responding to each component should be entered.He argued that what appears to be an interactioneffect may actually be a curvilinear effect of one orboth independent variables if those variables arethemselves highly related (see also Edwards, 1994;Ganzach, 1997; Lubinski & Humphreys, 1990). Thisapproach allows a researcher to interpret signifi-cant interaction terms after having accounted forthis alternative explanation. However, because in-sufficient theoretical clarity and empirical evi-dence exist to predict the specific nature of poten-tial curvilinearity in the relationship between

satisfaction mean and dispersion and absenteeism,we followed Edwards’s (1994) suggestion by pro-ceeding in an exploratory analytic fashion in Study1, with the intent of cross-validating results inStudy 2. Prior to conducting analyses, we standard-ized variables to reduce collinearity and facilitateinterpretation (Cohen et al., 2003). Because our an-alytical approach in Study 1 was exploratory, weassessed the significance of each block of regres-sion terms and created and analyzed surface plotscorresponding to the final significant step in eachequation (Edwards & Parry, 1993; Kristof-Brown &Stevens, 2001). The final step included a mean bydispersion product term and squared mean anddispersion terms, with full support for our hypoth-eses resting on a significant interaction term (VanVianen et al., 2004).

STUDY 1 RESULTS

Table 1 presents the means, standard devia-tions, and correlations among Study 1 variablesat the team level of analysis. Prior to analyses, weexamined the distribution of residuals in theteam-level absenteeism variable, given that theoverall distribution was positively skewed. Q-Qplots and the Cook-Weisberg procedure failed touncover evidence of nonnormality or heterosce-dasticity of residuals; thus, we proceeded usingthe raw absenteeism scores. Also, because influ-ential observations can disproportionately affectpolynomial regression results (Edwards & Parry,1993), we screened each equation for these obser-vations by examining diagonal values of the hatmatrix (i.e., leverage values), studentized residu-als, and Cook’s D-statistic (see also Edwards &Rothbard, 1999). Two such observations ex-

TABLE 1Means, Standard Deviations, and Correlations among Study 1 Variablesa

Variable Mean s.d. 1 2 3 4 5 6 7 8

1. Team size 4.37 0.792. Gender diversity 0.48 0.26 .113. Ethnic diversity 0.26 0.34 .19 .054. Course loadb 13.23 2.55 .07 .09 �.005. Team satisfaction level 5.70 0.69 �.09 �.11 �.10 �.036. Team satisfaction dispersion 0.90 0.62 �.08 .02 .15 .04 �.55**7. Course satisfaction level 5.56 0.70 �.02 �.09 .03 .00 .42** �.33**8. Course satisfaction dispersion 0.93 0.56 .13 .08 �.06 �.07 �.20* .31** �.67**9. Absenteeism 0.24 0.29 .22* �.05 �.04 �.13 �.23* .04 �.17 .20

a n � 101 teams.b In hours.

* p � .05** p � .01

Two-tailed tests.


ceeded Bollen and Jackman’s (1990) minimumlevels on all three screening criteria and fell inthe tail of the absenteeism distribution; they werethus deleted from analyses.

Table 2 presents results of the regression analy-ses for Study 1. The final step of the regressionequations involving internally (team) and exter-nally (course) focused satisfaction explained an ad-ditional 8 and 7 percent (p � .05) of the variation inabsenteeism, respectively. However, although vari-ance explained was similar, coefficients from step 3were opposite in sign for the two foci (see Table 2).First, the interaction term was significant in theteam satisfaction analysis, suggesting the relation-ship between mean team satisfaction and absentee-

ism depends on team satisfaction dispersion (VanVianen et al., 2004). We plotted the form of thisrelationship using a response surface graph to morespecifically assess support for Hypothesis 1. Figure2a, the response surface graph for the relationshipsof absenteeism with the mean and dispersion ofteam satisfaction, our internal focus, indicates ahigh level of consistency with the pattern of rela-tionships predicted in Figure 1a and Hypothesis 1.Specifically, a stronger, negative relationship be-tween team satisfaction and absenteeism is evidentwhen dispersion in team satisfaction is lower (i.e.,when there was strong agreement regarding teamsatisfaction: points A to B in Figure 2a). A weakernegative relationship suggesting homology with in-

TABLE 2Study 1 Regression Analysis Resultsa

Variables Model 1 Model 2 Model 3

ControlTeam size .25* .23* .22*Gender diversity �.06 �.09 �.10Ethnic diversity �.08 �.09 �.10Course load �.15 �.15 �.19

IndependentTeam satisfaction mean �.27* �.23Team satisfaction dispersion �.07 .04

Higher-order termsTeam satisfaction mean squared .03Team satisfaction dispersion squared .07Team satisfaction mean � dispersion .34*

R2 .08 .14* .21**�R2 .08 .05 .08*

Control

Team size .25* .24* .21*Gender diversity �.06 �.08 �.11Ethnic diversity �.08 �.07 �.08Course load �.15 �.14 �.12

IndependentCourse satisfaction mean �.12 �.20Course satisfaction dispersion .07 .19

Higher-order termsCourse satisfaction mean squared �.21Course satisfaction dispersion squared �.38*Course satisfaction mean � dispersion �.19

R2 .08 .11 .19*�R2 .08 .03 .07*

a The dependent variable is absenteeism. Standardized regression coefficients are shown.* p � .05

** p � .01


dividual-level findings is also apparent when dis-persion is higher (points D to C). Overall, absentee-ism is lowest when mean team satisfaction ishigher and dispersion is lower (point B), and it ishighest when both mean team satisfaction and dis-persion are lower (point A). Thus, Hypothesis 1received support.

By contrast, the interaction term relating coursesatisfaction mean and dispersion to absenteeism

did not reach a conventional level of significance,failing to fully support Hypothesis 2. This may be aconsequence of the small sample size and lack ofpower to detect the effect. However, because thefinal step was significant (explaining 7 percent ofthe variance in absenteeism) and because compar-ison with Study 2 results was desirable, we plottedthe form of the relationship, which is seen in Figure2b. As predicted, the pattern relating course satis-

FIGURE 2Study 1 Relationships of Satisfaction Mean and Dispersion with Absenteeism by Focia

a For interpretation, graphs rotated with means shown right to left from �1.0 (lower) to �1.0 (higher) standard deviations. Dispersionshown front to back from �1.0 (lower) to �0.5 (higher) standard deviations to account for the restricted range of means at higher dispersionlevels.


faction mean and dispersion to absenteeism ismarkedly different from the pattern exhibited forteam satisfaction. When course satisfaction disper-sion is higher, absenteeism is moderate when meancourse satisfaction is higher (point C), and it ishigher when mean course satisfaction is lower(point D). Thus, a moderate satisfaction-absentee-ism relationship is indicated, suggesting homologywith meta-analyzed individual-level results. Bycontrast, when dispersion is lower, absenteeismremains lower over all levels of mean course satis-faction (points A to B). This pattern of findingssupports both the common in-group and commu-nity of fate perspectives while suggesting a weakeroverall relationship than typically found at the in-dividual level. Overall, this pattern of relationshipsis highly consistent with the hypothesized patterndepicted in Figure 1b.

STUDY 2 CONTEXT AND REPLICATIONANALYSES

The Study 1 results contribute to our understand-ing of attitudinal diversity in teams in general andshed light on the satisfaction-absenteeism relation-ship in particular (e.g., Hackett, 1989; Judge et al.,2001) by demonstrating that different patterns ofabsenteeism are evident at the team level of analy-sis when satisfaction focus and the social context inwhich satisfaction is experienced are considered.However, Study 1 examined student-based teams;thus, the findings may have limited generalizabilityto work teams in organizational settings. To inves-tigate the external validity of the Study 1 findingsand cross-validate the pattern of results, in theStudy 2 sample we included manufacturing teamswith longer intrateam tenure, no predefined dis-banding date, and team operation in a paid (asopposed to a graded) context.

STUDY 2 METHODS

Sample and Procedures

Participants were 1,457 employees in 70 workteams in seven plants of an automobile parts man-ufacturing company. Teams performed a wide va-riety of production-related tasks and had an aver-age tenure of 32 months. Participants completed aquestionnaire during work time and were assuredconfidentiality. These data were part of a largerdata set that Colquitt and his colleagues (2002)used in a different set of analyses. Whereas theyexamined issues related to procedural justice cli-mate level and strength, the present study extendstheir approach and cross-validates the Study 1

findings. Furthermore, Colquitt et al. did not exam-ine job or team satisfaction (the primary focus ofthe present research), did not use polynomial re-gression, and overlapped with the present studyonly in regard to the absenteeism variable.

Measures

Measures were adapted from the literature andtailored to the manufacturing context with inputfrom organizational representatives. All items wereassessed using five-point scales. Team satisfaction(internally focused) was measured with the item,“Are you satisfied with the members of your workteam?” Externally focused satisfaction in this con-text was job satisfaction (assessed with “Are yousatisfied with your job?”). Whereas the reliability ofone-item measures cannot be definitively estab-lished, Wanous, Reichers, and Hudy (1997) en-dorsed their use and estimated a minimum reliabil-ity of .70 for these types of measures. Their meta-analysis showed that single-item measures ofoverall job satisfaction exhibited a corrected corre-lation of .72 with the group of scale measures thatfocus on overall job satisfaction rather than a sum-mation of facets. Moreover, the pilot study de-scribed in Study 1 revealed that the job satisfactionitem correlated .77 with the MSQ, .80 with the OJS,and .90 with the satisfaction scale used in Study 1.Finally, team absenteeism data were obtained fromfacility records for the three months following sur-vey administration (they were not available foreach individual employee). Absenteeism was opera-tionalized as average hours of absence from workper team member over this period. Because plantoperations were team-based, work absences werethus comparable to the absences from team meet-ings in Study 1.

We sought to account for other potential sourcesof variance in absenteeism by including severalcontrol variables. These included mean and disper-sion of perceived status in the team, operation-alized using Luhtanen and Crocker’s (1992) collec-tive self-esteem construct (three items based ontheir membership subscale, including, “I feel I aman important member of this team,” � � .82) andself-efficacy (mean and standard deviation, as-sessed with the item, “I think I am able to meet thechallenges posed by my work”). Both have beenlinked to cognitive withdrawal or withdrawal be-havior, including absenteeism, in prior work (e.g.,Harrison & Klein, in press; McDonald & Siegall,1993; Phillips, 2002; Phillips, Douthitt, & Hyland,2001; Van Dick & Wagner, 2002). For example,levels of self-efficacy relate to persistence of effort,which could extend to absenteeism. And disper-


sion of self-efficacy among teammates suggestshigher absenteeism if some members do not believethey can adequately contribute to the team effort.Finally, both self-esteem and efficacy have beenidentified as core self-evaluation facets, with coreself-evaluations being dispositional predictors ofsatisfaction (Judge & Bono, 2001).

Because two of the plants were considerablysmaller than the other five, we controlled forwhether a team was from a small or large plant. Useof these controls was consistent with work showingsignificant effects of contextual/environmental fac-tors on team outcomes (e.g., Kirkman, Tesluk, &Rosen, 2004) or specific relationships between fac-tory size and absenteeism (Porter & Lawler, 1965).We also controlled for team size and median teamtenure. We did not control for gender or ethnicdiversity because teams were of considerablylonger tenure in Study 2 than in Study 1 (Harrisonet al., 1998), and these controls were not significantin Study 1.

STUDY 2 RESULTS

Table 3 presents the means, standard deviations,and correlations among Study 2 variables at theteam level of analysis. The distribution of residualsagain suggested retaining the raw absenteeism mea-sure. Following Bollen and Jackman’s (1990) guide-lines, we removed two influential data points priorto conducting analyses.

Table 4 gives the results of our Study 2 regressionanalyses. The final step of the regression equationpredicting absenteeism from team satisfactionmean and dispersion was significant and explainedan additional 13 percent of the variance in absen-teeism (p � .01) in a pattern consistent with Figure1a. Supporting Hypothesis 1, a significant interac-tion term indicated that the mean team satisfaction-absenteeism relationship depended on team satis-faction dispersion. Figures 3a and 3b presentresponse surface graphs for the relationships of sat-isfaction mean and dispersion with absenteeism forour Study 2 internal and external foci. Figure 3aindicates that the relationship between mean teamsatisfaction and absenteeism is negative when teamsatisfaction dispersion is lower (points A to B), aspredicted, but relatively flat when dispersion ishigher (points D to C). Of particular interest is thatabsenteeism levels are again highest when teammembers are dissatisfied with each other and ex-hibit lower dispersion (point A) and lowest whenmembers are satisfied with each other and exhibitlower dispersion (point B); this pattern suggestsstronger satisfaction-absenteeism relationshipsamong less dispersed teams than is typically foundat the individual level (e.g., Harrison et al., 2006).

With respect to job satisfaction, the final stepexplained an additional 12 percent of the variancein absenteeism (p � .05; see Table 4), with a signif-icant mean by dispersion interaction term provid-ing support for Hypothesis 2. Figure 3b shows the

TABLE 3Means, Standard Deviations, and Correlations among Study 2 Variablesa

Variable Mean s.d. 1 2 3 4 5 6 7 8 9 10 11

1. Team size 21.21 11.552. Team tenure medianb 23.78 17.49 �.203. Collective self-esteem mean 3.39 0.32 �.24* �.154. Collective self-esteem dispersion 0.83 0.22 .40** �.17 �.27*5. Self-efficacy mean 4.25 0.24 �.01 �.32** .28* .126. Self-efficacy dispersion 0.82 0.22 .32** �.10 �.32* .28* �.55**7. Team from a small plantc 0.03 0.17 �.02 �.05 .29* �.02 .20 �.098. Team satisfaction level 3.62 0.36 �.31* .08 .66** �.40** �.03 �.16 .32**9. Team satisfaction dispersion 0.97 0.26 .38** �.15 �.36** .58** .12 .24 �.15 �.60**

10. Job satisfaction level 3.62 0.30 �.17 �.21 .35** .01 .19 �.20 .19 .40** �.1811. Job satisfaction dispersion 0.95 0.21 .35** �.17 .03 .39** .16 .20 �.03 �.22 .34** �.34**12. Absenteeismd 24.43 8.67 .18 �.13 �.13 .15 .02 .22 �.29* �.28* .06 �.11 .14

a n � 68 teams.b Tenure (in months) was assessed using the median given that over half the teams had a negatively skewed within-team tenure

distribution (Harrison et al., 1998).c 0 � “no,” 1 � “yes.”d In days.

* p � .05** p � .01

Two-tailed tests.


surface plot, which is generally consistent with thepattern of results involving course satisfaction inStudy 1. Absenteeism is relatively low across levelsof job satisfaction when dispersion is lower (pointsA to B) and actually lowest when team membersexhibit lower dispersion regarding job dissatisfac-tion (point A). In contrast, absenteeism ranges froma lower level to its highest level when job satisfac-tion dispersion is higher (points C to D). Theseresults suggest that the team-level relationship be-tween externally focused satisfaction and absentee-ism is more homologous with individual-levelfindings when dispersion is higher but less homol-ogous when dispersion is lower.

SUPPLEMENTAL MONTE CARLO SIMULATION

An issue worthy of additional exploration is thepotentially problematic interdependence betweenmean and dispersion constructs in team-level re-search. Scholars have suggested that controlling formean and dispersion main effects prior to examin-ing higher-order terms accounts for variance attrib-utable to the interdependence between these con-structs (Bliese & Halverson, 1998; Lindell & Brandt,2000). Yet it is possible that this interdependencestill makes it more difficult to detect interactionsbetween the two.

The Cortina (1993) approach discussed earlier is

TABLE 4Study 2 Regression Analysis Resultsa

Variables Model 1 Model 2 Model 3

ControlTeam size .09 .09 .15Team tenure �.02 .01 �.12Collective self-esteem mean .00 .17 .15Collective self-esteem dispersion �.00 .06 .01Self-efficacy mean .24 .25 .19Self-efficacy dispersion .29 .35* .30Team from a small plant �.32* �.28* �.25*

IndependentTeam satisfaction mean �.38* �.40*Team satisfaction dispersion �.33 �.34*

Higher-order termsTeam satisfaction mean squared .52*Team satisfaction dispersion squared .50**Team satisfaction mean � dispersion .82**

R2 .18 .25* .38**�R2 .18 .08 .13*

ControlTeam size .09 .08 �.04Team tenure �.02 �.03 �.08Collective self-esteem mean .00 .02 .04Collective self-esteem dispersion �.00 .01 �.16Self-efficacy mean .24 .24 .39*Self-efficacy dispersion .29 .29 .35*Team from a small plant �.32* �.32* �.36**

IndependentJob satisfaction mean �.04 �.06Job satisfaction dispersion �.02 .08

Higher-order termsJob satisfaction mean squared �.36**Job satisfaction dispersion squared �.39*Job satisfaction mean � dispersion �.38*

R2 .18 .18 .30*�R2 .18* .00 .12*

a The dependent variable is absenteeism. Standardized regression coefficients are shown.* p � .05

** p � .01


useful in dealing with linearly correlated predic-tors when examining interactions between them,but it is uncertain how the approach might workwhen predictors exhibit the theoretical inverse U-shaped relationship that exists between means andstandard deviations (Lindell & Brandt, 2000) or anonsymmetrical inverse U-shaped relationshipsuch as the one we discovered in our data. Todirectly examine this issue, we conducted a MonteCarlo simulation where, for each of the four analy-

ses reported in Tables 2 and 4, we assessed theincremental increase in variance explained (R2) forthe interaction term over 10,000 samples generatedunder three different distributional assumptions.Each of the simulated samples was the same size asthose in the respective studies (n � 101 in Study 1and 68 in Study 2). In particular, we first generated10,000 samples that replicated, on average, the ac-tual distributional properties of our mean and stan-dard deviation constructs (X and Z). Then, for each

FIGURE 3Study 2 Relationships of Satisfaction Mean and Dispersion with Absenteeism by Focia

a Means are shown right to left from �1.0 (lower) to �1.0 (higher) standard deviations. Dispersion is shown front to back from �1.0(lower) to �0.5 (higher) standard deviations to account for the restricted range of means at higher dispersion levels.


of these 10,000 samples, we computed slope esti-mates for the five regression terms (X, Z, X2, Z2,XZ), as well as an overall R2 value, and comparedthis value to one where the interaction term wasexcluded. Incremental R2 values (averaged over all10,000 samples) were then compared to incremen-tal R2 values that resulted when distributionalproperties were changed in such a way that (1) Xand Z were normally distributed and completelyuncorrelated and (2) X was normally distributed,with Z a symmetrical inverse U-shaped function ofX, reflecting the theoretical relationship betweenmean and standard deviation (Lindell & Brandt,2000).

Taking the team satisfaction mean-dispersion re-lationship with absenteeism from Study 1 as anexample, one sees the mean and standard deviationof sampling means (from Table 1) were 5.70 and0.69, whereas the mean and standard deviation ofsampling standard deviations were 0.90 and 0.62.Team satisfaction mean and standard deviationwere related in an inverse U-shaped fashion with acorrelation of –.55, but the relationship was non-symmetrical, with fewer cases in the left half of thedistribution. The 10,000 simulated samples eachincluded a normally distributed random error term(with a mean of zero and variance commensuratewith that in the actual data) that allowed us toreplicate these distributional properties. We thenassessed the impact of these properties over the10,000 samples on our ability to detect an interac-tion effect by comparing incremental R2 values tosimilar values arrived at when using the uncorre-lated and symmetrical inverse U-shaped distribu-tions (10,000 samples each).

Results for the four sets of analyses were consis-tent in showing that incremental R2 values for theinteraction term were nearly identical when X andZ variables that were either uncorrelated or relatedin a symmetrical inverse U-shaped fashion wereconsidered. However, given samples averaging theactual, nonsymmetrical distributional propertiesapparent in our data, incremental R2 values for theinteraction term were lower for each of the fouranalyses. For example, for team satisfaction inStudy 1, incremental R2 values were .11 (uncorre-lated X and Z distributions), .10 (symmetrical in-verse U), and .05 (replication of our actual data).1

These simulations indicate that our study resultsare conservative. Thus, to the degree that meansand dispersion are related in a symmetrical inverseU-shaped fashion, confidence in being able to de-

tect an interaction effect between the two whenusing the Cortina (1993) approach appears justi-fied, whereas a nonsymmetrical curvilinear rela-tionship (like that found in our samples) appears toincrease the difficulty of detecting interactioneffects.

GENERAL DISCUSSION

The study results indicate that relationships be-tween satisfaction and absenteeism typically foundat the individual level are not necessarily homolo-gous at the team level of analysis when satisfactionfoci and dispersion are taken into account. In twodiverse samples, mean levels of course or job (ex-ternally focused) satisfaction were unrelated to ab-senteeism when satisfaction dispersion amongteam members was lower. By contrast, when dis-persion in team (internally focused) satisfactionwas lower, mean levels of team satisfactionstrongly related to absenteeism. For both satisfac-tion foci, results more homologous with individu-al-level findings tended to materialize only whensatisfaction dispersion was higher.

These findings advance understanding of satis-faction and absenteeism at the team level by ex-plaining and then demonstrating how their rela-tionship differs across dispersion levels andsatisfaction foci. Our research answers calls to in-corporate the social context into studies of organi-zational phenomena (Johns, 2006; Kozlowski &Klein, 2000) and offers a potential explanation forwhy the negative relationship between satisfactionand absenteeism at the individual level is not stron-ger. Indeed, the widespread use of teams in organ-izations makes studying relationships involving in-dividually held attitudes such as job satisfaction inteam contexts even more critical, given that theseattitudes are shaped, in part, by the attitudes ofother team members. Previously, the effects of co-worker satisfaction levels on an individual’s ownsatisfaction might have been weaker because con-tact and information exchange between these em-ployees were lower.

The study further addresses the likelihood thateffects on absenteeism vary differentially depend-ing on satisfaction focus and that the functionalhomology of the satisfaction-absenteeism relation-ship at the team level may differ with both disper-sion and satisfaction focus. These considerationsappear to be important, as they provide more de-tailed information regarding the satisfaction-absen-teeism relationship that has previously been veiledby a near-singular focus on individual-level stud-ies. Indeed, Hackett concluded his meta-analysisby acknowledging the oversimplicity of a direct

1 Additional details regarding this simulation areavailable from the first author upon request.


relationship between absence and work attitudes,and it is possible that social-contextual factors havebeen present but unaccounted for in studies con-ducted at the individual level, leading to attenu-ated results. It is thus encouraging that researchersare starting to recognize the dearth of studies usinga dispersion model of aggregation as well as thecriticality of extending research in this area (e.g.,Boone et al., 2005; Klein et al., 2001; Ostroff & Bowen,2000), and we endorse future work of this type.

We expected that when satisfaction judgmentswere focused internally on other team members,mean team satisfaction would be strongly and neg-atively related to absenteeism only when team sat-isfaction dispersion was lower and that this rela-tionship would be weaker and fairly homologouswith individual-level findings when dispersionwas higher. We expected, however, to see a contrastwith individual-level findings for externally fo-cused satisfaction judgments. In view of commonin-group and community of fate arguments, we didnot expect mean satisfaction and absenteeism torelate when satisfaction judgments were externallyfocused (i.e., on a course or job) and satisfactiondispersion was lower. We found considerable sup-port for our hypotheses in a college class settingand strong support among manufacturing teams.The divergent findings for the two satisfaction fociare particularly remarkable given the positive cor-relations between team and course (job) satisfac-tion, which ranged from .40 to .42 in the samples.

Theoretical Implications

First, the pattern of results for externally focusedsatisfaction was highly consistent with predictionsbased on the combination of common in-group andcommunity of fate perspectives, as well as thedeep-level diversity literature and studies linkingmean levels of attitudes to outcomes. A particularlyinteresting finding, generally consistent over thetwo samples, was that individuals who were dis-satisfied with the external entity—their course orjob—were not the most likely to be absent fromteam meetings or work, but rather were less likelyto be absent from team meetings or work if theirteammates held similarly negative evaluations oftheir external work context. To our knowledge, thisis the first study to conceptually differentiate andempirically demonstrate such a pattern of relation-ships. Although seemingly counterintuitive andlacking homology with individual-level findings,the pattern evidenced here is consistent with recentaccounts of actual employee behavior (Weeks,2004) as well as with the common in-group identityand community of fate perspectives (Brewer &

Brown, 1998; Shaw et al., 2000). These perspec-tives all suggest that the creation of a “commonenemy” through shared negative attitudes (viz., anundesirable job) can actually create bonding amongteam members. Given this social-contextual influ-ence, members may enjoy showing up for teammeetings or for work, perhaps to commiserate withteammates or simply to support one another andreceive affirmation for their sense of dissatisfac-tion. This finding is also similar to the classic neg-ative cohesion–performance relationship in whichteams adopt low performance norms.

By contrast, lower dispersion regarding levels ofinternally focused satisfaction augments the nega-tive effects of means on absenteeism, resulting in agreater likelihood of absenteeism when teammatesare similarly dissatisfied with each other. This pat-tern suggests that team-level findings might bestronger than individual-level findings when dis-persion is lower, which is consistent with the teamclimate literature. However, it also calls into ques-tion the notion that attitudinal consensus is alwaysbeneficial (Harrison et al., 1998).

Taken together, the results emphasize the impor-tance of Chan’s (1998) typology of compositionmodels. In particular, we provide a conceptualfoundation for treating dispersion as a unit-levelconstruct with meaningful variation for team-leveltheory building. It is likely that our pattern of re-sults has not previously been demonstrated be-cause prior research has tended either to adoptdirect consensus models of aggregation (Chan,1998), thereby treating dispersion as error varianceand requiring sufficient within-group agreement tojustify aggregation (Ostroff, 1992), or to relegatemean attitude levels to control variable status indiversity studies (Harrison et al., 1998). Yet, from atheoretical standpoint, it is reasonable to expectthat dispersion in satisfaction will exist, given thatindividuals might experience discretionary stimulior bring dispositions toward varied levels of satis-faction with them to a team context (Hackman,1992; Staw et al., 1986).

Another important theoretical contribution is ourdifferentiation between internally and externallyfocused satisfaction among team members. Thisdifferentiation parallels that made in research dis-tinguishing between foci of commitment (e.g.,Becker, 1992; Clugston et al., 2000; Siders et al.,2001). Along with social-contextual consider-ations, our differentiation helps to further elucidatethe precise nature of the satisfaction-absenteeismrelationship. In particular, whereas team climateresearch and our results suggest that agreement re-garding lower levels of internally focused satisfac-tion is detrimental in terms of higher absenteeism,


we find strikingly different results for externallyfocused satisfaction. Here, a unified sense of dis-satisfaction directed toward an external entity ap-pears to draw team members together, causingthem to attend work or team meetings to a greaterextent.

Also of interest is that the relative impact ofdispersion or mean levels of satisfaction on absen-teeism appears to differ for the two satisfaction foci.For example, Tables 2 and 4 (model 2) both indi-cate main effects for mean levels when internally—but not externally—focused satisfaction is underconsideration. Similarly, correlational evidencefrom Tables 1 and 3 shows stronger relationshipsbetween mean team satisfaction and absenteeismthan between mean course (job) satisfaction andabsenteeism. Correlational evidence from teamswhose dispersion was below the median also mightfacilitate understanding of weak individual-levelmeta-analytic results while illuminating the degreeto which homology with individual-level relation-ships varies by the foci. For example, among teamswith lower dispersion, the average team satisfac-tion-absenteeism relationship was –.42 across stud-ies, and the average corresponding course (job) sat-isfaction relationship was –.09, whereas meta-analytic estimates at the individual level are in the–.21 to –.23 range (Hackett, 1989). In keeping withFigure 1, these findings suggest that when inter-nally focused satisfaction is considered and teamsexhibit lower satisfaction dispersion, the satisfac-tion-absenteeism relationship is stronger than it isat the individual level, whereas when externallyfocused satisfaction is considered and dispersion islower, the relationship is weaker than it is at theindividual level. Indeed, if one were to overlayFigure 2a onto Figure 2b, and 3a onto 3b, it wouldbecome apparent that satisfaction-absenteeism ef-fects are cancelled out.

Finally, it is possible that the determinants ofsatisfaction among team members may differ de-pending on its focus. For example, average teamtenure levels were correlated –.23 with job satisfac-tion dispersion, yet only –.03 with team satisfac-tion dispersion in Study 2. This pattern suggeststhat, relative to the dispositional aspect, the social-contextual aspect of satisfaction may be more in-fluential over time for externally but not internallyfocused satisfaction, an idea that is consistent withthe solidifying influence of common in-group dy-namics. Continued distinction between these twofoci is therefore encouraged. Researchers mightalso strive to refine this distinction to account forpotential differences in relationships involving ex-ternally or internally focused attitudes that are ei-ther interpersonal or not interpersonal in nature.

Strengths, Limitations, and Directions for FutureResearch

The study conclusions are particularly robust be-cause divergent patterns of results were replicatedin two extremely diverse samples. The samplesdiffered on setting (students at a university versusadults working full-time in industry), rewards(graded versus paid), task type (knowledge-basedversus manufacturing-based), and expected tenure(16 weeks versus ongoing). In addition, concernsover measurement limitations—for example, theself-report absenteeism measure in Study 1 or thesingle-item satisfaction measures in Study 2—arereduced by the replicated results across the twosamples.

Despite the strengths of these two studies in com-bination, limitations exist. For example, course sat-isfaction, team satisfaction, and absences fromteam meetings were assessed on the same survey inStudy 1. This raises concerns about commonmethod variance and our presumed causal se-quence, although common method bias does notprovide a logical explanation for differential high-er-order findings. Moreover, some recent researchprovides initial evidence suggesting the prece-dence of attitudes to behaviors (Harrison et al.,2006). Also, the limited team tenure in Study 1likely contributed to the weaker overall effects (e.g.,the effects of deep-level diversity may grow stron-ger over time [Harrison et al., 1998]), and neitherstudy explicitly accounted for unintentional ab-sences (e.g., legitimate illnesses).

It is quite possible that the modest relationshipbetween satisfaction and other outcomes such asperformance (the “holy grail” of organizational re-search [cf. Judge et al., 2001]) might be due in partto a lack of consideration of the joint effects ofmean levels of satisfaction as they exist in a team,as well as the dispersion in satisfaction amongmembers. We also encourage researchers to lookmore closely at the black box surrounding the rela-tionship between attitudinal constellations andoutcomes. For example, the social support andcommon in-group identity thought to developwhen attitudes converge should be scrutinized,perhaps by investigating specific patterns of socialinteraction among team members. In addition,common in-group or community of fate dynamicsmight differ by cultural type (e.g., degree of teammember collectivism). To expand the criterion do-main, researchers should also examine mean levelsand dispersion in relation to an overall effective-ness construct that includes withdrawal and in-and extra-role performance (see Harrison et al.,2006). Efforts to more precisely examine the homol-


ogy of other attitude-outcome relationships interms of differences in specific parameters acrosslevels would be valuable, as would efforts to betterelucidate the specific structure of attitudes in orderto justify assessing mean levels (e.g., if an attitudeis more of a shared-unit property of a team), dis-persion (if a configural property), or both (if anattitude is thought to contain both shared unit andconfigural elements, or as a “conservative” ap-proach if their relative balance is uncertain). Ingeneral, researchers are advised to ground simulta-neous examinations of team-level means and dis-persion in relevant theory (Harrison & Sin, 2005).

Researchers might also pursue other means ofincorporating dispersion as a substantive construct.In the present study, level and dispersion wereboth derived from the same measures. Althoughmutual derivation is the most common practice inresearch in this stream, it introduces the issue ofinterdependence discussed earlier. Those makingfuture efforts might consider using a more directapproach. For example, respondents could be di-rectly asked about their satisfaction levels and theirperceptions of how strongly they believe othermembers of their team share their attitude. A pos-itive correlation between a dispersion index andthe mean of the perceived shared attitude indexmight provide construct validity evidence for thedispersion index.

Managerial Implications

The study results suggest that, in attempting toinfluence outcomes such as absenteeism, managersshould look beyond individual worker satisfactionor mean satisfaction and instead focus on relativesatisfaction among team members. Indeed, this rec-ommendation suggests a shift of attention from try-ing to predict whether a happy worker shows up towork to whether a worker’s happiness relative toteammates’ predicts attendance. By extension,managers should consider that enacting change tocounter dissatisfaction may not always be immedi-ately necessary, a sentiment echoed by Weeks(2004). In fact, shared dissatisfaction does not ap-pear to be problematic if it is externally focused, atleast in terms of absences.

Perhaps even more interesting, however, is thesuggestion that managers dedicate more effort toenhancing team satisfaction rather than job satis-faction. Indeed, the results depicted in Figures 2and 3 imply that attempts to raise average teamsatisfaction levels have a greater payoff than doattempts to raise course or job satisfaction levels,especially when dispersion in satisfaction is lower.For example, absenteeism decreases substantially

from points A to B in Figures 2a and 3a as teamsatisfaction is enhanced, but points A to B in Fig-ures 2b and 3b show that absenteeism remains rel-atively unchanged as course (job) satisfaction levelsare enhanced. Thus, managers might benefit moreby focusing on appropriate team compositionrather than on job design. For example, they mightvary team composition to ensure similarity in val-ues among team members, without having tochange the nature of the jobs performed. Doing somight decrease relationship conflict (Jehn, North-craft, & Neale, 1999) and promote a consistentsense of team satisfaction.

REFERENCES

Barrick, M. R., Stewart, G. L., Neubert, M. J., & Mount,M. K. 1998. Relating member ability and personalityto work-team processes and team effectiveness. Jour-nal of Applied Psychology, 83: 377–391.

Barsade, S., Ward, A., Turner, J., & Sonnenfeld, J. 2000.To your heart’s content: A model of affective diver-sity in top management teams. Administrative Sci-ence Quarterly, 45: 802–836.

Becker, T. E. 1992. Foci and bases of commitment: Arethey distinctions worth making? Academy of Man-agement Journal, 35: 232–244.

Bishop, J., & Scott, K. D. 2000. An examination of organ-izational and team commitment in a self-directedteam environment. Journal of Applied Psychology,85: 439–450.

Bliese, P. D., & Halverson, R. R. 1998. Group consensusand psychological well-being: A large field study.Journal of Applied Social Psychology, 28: 563–580.

Bollen, K. A., & Jackman, R. W. 1990. Regression diag-nostics: An expository treatment of outliers and in-fluential cases. In J. Fox & J. S. Long (Eds.), Modernmethods of data analysis: 257–291. Newbury Park,CA: Sage.

Boone, C., Van Olffen, W., & Van Witteloostuijn, A. 2005.Team locus-of-control composition, leadershipstructure, information acquisition, and financial per-formance: A business simulation study. Academy ofManagement Journal, 48: 889–909.

Bowling, N. A., Beehr, T. A., Wagner, S. H., & Libkuman,T. M. 2005. Adaptation-level theory, opponent pro-cess theory, and dispositions: An integrated ap-proach to the stability of job satisfaction. Journal ofApplied Psychology, 90: 1044–1053.

Brayfield, A., & Rothe, H. 1951. An index of job satisfac-tion. Journal of Applied Psychology, 35: 307–311.

Brewer, M., & Brown, R. 1998. Intergroup relations. In D.Gilbert, S. Fiske, & L. Gardner (Eds.), The handbookof social psychology (4th ed.), vol. 2: 554–594. NewYork: McGraw-Hill.

Brown, K., Kozlowski, S. W. J., & Hattrup, K. 1996. The-


ory, issues, and recommendations in conceptualiz-ing agreement as a construct in organizational re-search: The search for consensus regardingconsensus. Paper presented at the annual meeting ofthe Academy of Management.

Cammann, C., Fichman, M., Jenkins, G. D., & Klesh, J. R.1983. Assessing the attitudes and perceptions of or-ganizational members. In S. E. Seashore, E. E.Lawler, P. H. Mirvis, & C. Cammann (Eds.), Assess-ing organizational change: A guide to methods,measures, and practices: 71–138. New York: Wiley.

Chan, D. 1998. Functional relations among constructs inthe same content domain at different levels of anal-ysis: A typology of composition models. Journal ofApplied Psychology, 83: 234–246.

Chen, G., Mathieu, J. E., & Bliese, P. D. 2004. A frame-work for conducting multilevel construct validation.In F. J. Yammarino & F. Dansereau (Eds.), Researchin multilevel issues: Multilevel issues in organiza-tional behavior and processes, vol. 3: 273–303. Ox-ford, U.K.: Elsevier.

Clugston, M., Howell, J. P., & Dorfman, P. W. 2000. Doescultural socialization predict multiple bases and fociof commitment? Journal of Management, 26: 5–30.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. 2003.Applied multiple regression/correlation analysisfor the behavioral sciences (3rd ed.). Mahwah, NJ:Erlbaum.

Colquitt, J., Noe. R., & Jackson, C. 2002. Justice in teams:Antecedents and consequences of procedural justiceclimate. Personnel Psychology, 55: 83–110.

Cortina, J. M. 1993. Interaction, nonlinearity, and multi-collinearity: Implications for multiple regression.Journal of Management, 19: 915–922.

Edwards, J. 1994. The study of congruence in organiza-tional behavior research: Critique and proposed al-ternative. Organizational Behavior & Human Deci-sion Process, 58: 51–100.

Edwards, J. R., & Parry, M. 1993. On the use of polyno-mial regression equations as an alternative to differ-ence scores in organizational research. Academy ofManagement Journal, 36: 1577–1613.

Edwards, J. R., & Rothbard, N. P. 1999. Work and familystress and well-being: An examination of person-environment fit in the work and family domains.Organizational Behavior & Human Decision Pro-cesses, 77: 85–129.

Gammage, K. L., Carron, A. V., & Estabrooks, P. A. 2001.Team cohesion and individual productivity: The in-fluence of the norm for productivity and the identi-fiability of individual effort. Small Group Re-search, 32: 3–18.

Ganzach, Y. 1997. Misleading interaction and curvilinearterms. Psychological Methods, 2: 235–247.

Gellatly, I. 1995. Individual and group determinants of

employee absenteeism: Test of a casual model. Jour-nal of Organizational Behavior, 16: 469–485.

George, J. 1990. Personality, affect, and behavior ingroups. Journal of Applied Psychology, 75: 107–116.

Gonzalez-Roma, V., Peiro, J., & Tordera, N. 2002. Anexamination of the antecedents and moderator influ-ences of climate strength. Journal of Applied Psy-chology, 87: 465–473.

Goodman, P., Ravlin, E., & Schminke, M. 1987. Under-standing groups in organizations. In L. L. Cummings& B. M. Staw (Eds.), Research in organizationalbehavior, vol. 9: 121–173. Greenwich, CT: JAI.

Hackett, R. D. 1989. Work attitudes and employee absen-teeism: A synthesis of the literature. Journal of Oc-cupational Psychology, 62: 235–248.

Hackman, J. R. 1992. Team influences on individuals inorganizations. In M. D. Dunnette & L. M. Hough(Eds.), Handbook of industrial and organizationalpsychology (2nd ed.), vol. 3: 199–267. Palo Alto, CA:Consulting Psychologists Press.

Harrison, D. A., & Klein, K. J. In press. What’s the differ-ence? Diversity constructs as separation, variety, ordisparity in organizations. Academy of Manage-ment Review.

Harrison, D. A., & Martocchio, J. J. 1998. Time for absen-teeism: A 20-year review of origins, offshoots, andoutcomes. Journal of Management, 34: 305–350.

Harrison, D. A., Newman, D. A., & Roth, P. L. 2006. Howimportant are job attitudes? Meta-analytic compari-sons of integrative behavioral outcomes and timesequences. Academy of Management Journal, 49:305–326.

Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyondrelational demography: Time and the effects of sur-face- and deep-level diversity on work group cohe-sion. Academy of Management Journal, 41: 96–107.

Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A. T.2002. Time, teams, and task performance: Changingeffects of surface- and deep-level diversity on groupfunctioning. Academy of Management Journal, 45:1029–1045.

Harrison, D., & Sin, H. 2005. What is diversity and howshould it be measured? In A. Konrad, P. Prasad, & J.Pringle (Eds.), Handbook of workplace diversity:191–216. Thousand Oaks, CA: Sage.

Jackson, S. E., May, K. E., & Whitney, K. 1995. Under-standing the dynamics of diversity in decision mak-ing teams. In R. A. Guzzo & E. Salas (Eds.), Teameffectiveness and decision making in organiza-tions: 204–261. San Francisco: Jossey-Bass.

Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999. Whydifferences make a difference: A field study of diver-sity, conflict, and performance in workgroups. Ad-ministrative Science Quarterly, 44: 741–763.


Johns, G. 2006. The essential impact of context on organ-izational behavior. Academy of Management Re-view, 31: 386–408.

Judge, T., Thoresen, C., Bono, J., & Patton, G. 2001. Thejob satisfaction–job performance relationship: Aqualitative and quantitative review. PsychologicalBulletin, 127: 376–407.

Judge, T. A., & Bono, J. E. 2001. Relationship of coreself-evaluation traits—self-esteem, generalized self-efficacy, locus of control, and emotional stability—with job satisfaction and job performance: A meta-analysis. Journal of Applied Psychology, 86: 80–92.

Kirkman, B. L., & Rosen, B. 1999. Beyond self-manage-ment: Antecedents and consequences of team em-powerment. Academy of Management Journal, 42:58–74.

Kirkman, B. L., & Shapiro, D. L. 2005. The impact ofcultural value diversity on multicultural team per-formance. In J. L. Cheng & D. L. Shapiro (Eds.),Advances in international management: Manag-ing multinational teams, vol. 18: 33–67. London:Elsevier.

Kirkman, B. L., Tesluk, P. E., & Rosen, B. 2004. Theimpact of demographic heterogeneity and team lead-er–team member demographic fit on team empower-ment and effectiveness. Group & OrganizationManagement, 29: 334–368.

Klein, K., Conn, A., Smith, D. B., & Sorra, J. 2001. Iseveryone in agreement? An exploration of within-group agreement in employee perceptions of thework environment. Journal of Applied Psychology,86: 3–16.

Kozlowski, S., & Klein, K. 2000. A multilevel approach totheory and research in organizations: Contextual,temporal, and emergent processes. In K. Klein & S.Kozlowski (Eds.), Multilevel theory, research, andmethods in organizations: 3–90. San Francisco: Jos-sey-Bass.

Kristof-Brown, A. L., Barrick, M. R., & Stevens, C. K.2005. When opposites attract: A multi-sample dem-onstration of complementary person-team fit on ex-traversion. Journal of Personality, 73: 936–957.

Kristof-Brown, A. L., & Stevens, C. 2001. Goal congru-ence in project teams: Does the fit between members’personal mastery and performance goals matter?Journal of Applied Psychology, 86: 1083–1095.

Lindell, M., & Brandt, C. 2000. Climate quality and cli-mate consensus as mediators of the relationship be-tween organizational antecedents and outcomes.Journal of Applied Psychology, 85: 331–348.

Lindsley, D. H., Brass, D. J., & Thomas, J. B. 1995. Effi-cacy-performance spirals: A multilevel perspective.Academy of Management Review, 20: 645–678.

Lubinski, D., & Humphreys, L. G. 1990. Assessing spuri-ous “moderator effects”: Illustrated substantivelywith the hypothesized (“synergistic”) relation be-

tween spatial and mathematical ability. Psychologi-cal Bulletin, 107: 385–393.

Luhtanen, R., & Crocker, J. 1992. A collective self-esteemscale: Self-evaluation of one’s social identity. Per-sonality and Social Psychology Bulletin, 18: 302–318.

Markham, S., Dansereau, F., & Alutto, J. 1982. Group sizeand absenteeism rates: A longitudinal analysis.Academy of Management Journal, 25: 921–927.

Mason, C., & Griffin, M. 2002. Group task satisfaction:Applying the construct of job satisfaction to groups.Small Group Research, 33: 271–312.

Mathieu, J. E., & Kohler, S. S. 1990. A cross-level exam-ination of group absence influences on individualabsence. Journal of Applied Psychology, 75: 217–220.

McDonald, T., & Siegall, M. 1993. The effects of techno-logical self-efficacy and job focus on job perfor-mance, attitudes, and withdrawal behaviors. Journalof Psychology, 126: 465–475.

Milliken, F., & Martins, L. 1996. Searching for commonthreads: Understanding multiple effects of diversityin organizational groups. Academy of ManagementReview, 21: 402–433.

Morgeson, F. P., & Hofmann, D. A. 1999. The structureand function of collective constructs: Implicationsfor multilevel research and theory development.Academy of Management Review, 24: 249–265.

Ostroff, C. 1992. The relationship between satisfaction,attitudes, and performance: An organizational levelanalysis. Journal of Applied Psychology, 77: 963–974.

Ostroff, C., & Bowen, D. 2000. Moving HR to a higherlevel: HR practices and organizational effectiveness.In K. Klein & S. Kozlowski (Eds.), Multilevel theory,research, and methods in organizations: 211–266.San Francisco: Jossey-Bass.

Ostroff, C., & Harrison, D. 1999. Meta-analysis, level ofanalysis, and best estimates of population correla-tions: Cautions for interpreting meta-analytic resultsin organizational behavior. Journal of Applied Psy-chology, 84: 260–270.

Phillips, J. 2002. Antecedents and consequences of pro-cedural justice perceptions in hierarchical decision-making teams. Small Group Research, 33: 32–64.

Phillips, J., Douthitt, E., & Hyland, M. 2001. The role ofjustice in team member satisfaction with the leaderand attachment to the team. Journal of Applied Psy-chology, 86: 316–325.

Poole, M. S., & Van De Ven, A. H. 1989. Using paradox tobuild management and organizational theories.Academy of Management Review, 14: 562–578.

Porter, L. W., & Lawler, E. E. 1965. Properties of organi-zational structure in relation to job attitudes and jobbehavior. Psychological Bulletin, 64: 23–51.


Prussia, G. E., & Kinicki, A. J. 1996. A motivational in-vestigation of group effectiveness using social-cogni-tive theory. Journal of Applied Psychology, 81:187–198.

Rentsch, J. R., & Steel, R. P. 2003. What does unit-levelabsence mean? Issues for future unit-level absenceresearch. Human Resource Management Review,13: 185–202.

Ryan, A. M., Schmit, M., & Johnson, R. 1996. Attitudesand effectiveness: Examining relations at an organi-zational level. Personnel Psychology, 49: 853–882.

Salancik, G., & Pfeffer, J. 1978. A social information pro-cessing approach to job attitudes and task design.Administrative Science Quarterly, 23: 224–253.

Schneider, B., Salvaggio, A., & Subirats, M. 2002. Climatestrength: A new direction for climate research. Jour-nal of Applied Psychology, 87: 220–229.

Shaw, J., Duffy, M., & Stark, E. 2000. Interdependenceand preference for group work: Main and congru-ence effects on the satisfaction and performance ofgroup members. Journal of Management, 26: 259–279.

Siders, M., George, G., & Dharwadkar, R. 2001. The rela-tionship of internal and external commitment foci toobjective job performance measures. Academy ofManagement Journal, 44: 570–579.

Somers, M. J. 1995. Organizational commitment, turn-over and absenteeism: An examination of direct andinteraction effects. Journal of Organizational Be-havior, 16: 49–58.

Staw, B. M., Bell, N. E., & Clausen, J. A. 1986. Thedispositional approach to job attitudes: A lifetimelongitudinal test. Administrative Science Quar-terly, 31: 56–77.

Stodgill, R. M. 1972. Group productivity, drive, and co-hesiveness. Organizational Behavior and HumanPerformance, 8: 26–43.

Van Der Vegt, G., & Bunderson, S. 2005. Learning andperformance in multidisciplinary teams: The impor-tance of collective team identification. Academy ofManagement Journal, 48: 532–547.

Van Dick, R., & Wagner, U. 2002. Social identificationamong school teachers: Dimensions, foci, and corre-lates. European Journal of Work & OrganizationalPsychology, 11: 129–149.

Van Vianen, A., De Parter, I., Kristof-Brown, A., & John-son, E. 2004. Fitting in: Surface- and deep-level cul-tural differences and expatriates’ adjustment. Acad-emy of Management Journal, 47: 697–709.

Wanous, J., Reichers, A., & Hudy, M. 1997. Overall jobsatisfaction: How good are single-item measures?Journal of Applied Psychology, 82: 247–252.

Weeks, J. 2004. Whining away the hours: Employees’complaints are often good for morale, particularlywhen nothing’s done about them. Harvard BusinessReview, 82(5): 20–21.

Weiss, D., Dawis, R., England, G., & Lofquist, L. 1967.Manual for the Minnesota Satisfaction Question-naire (Minnesota studies for vocational rehabilita-tion, no. 22). Industrial Relations Center, Universityof Minnesota, Minneapolis.

Brian R. Dineen ([email protected]) is an assistantprofessor of management in the Gatton College of Busi-ness and Economics at the University of Kentucky. Hereceived his Ph.D. from The Ohio State University. Hisresearch interests include recruitment, person-environ-ment congruence, and counterproductive behavioramong members of collectives.

Raymond A. Noe ([email protected]) is the Robertand Anne Hoyt Designated Professor of Management atThe Ohio State University. He earned his Ph.D. in psy-chology from Michigan State University. His researchinterests include training and development, knowledgemanagement, recruiting, team processes, and work andfamily.

Jason D. Shaw ([email protected]) is an associateprofessor in the Carlson School of Management at theUniversity of Minnesota. He received his Ph.D. in man-agement from the University of Arkansas. His researchinterests include individual and organizational conse-quences of compensation decisions, voluntary and invol-untary turnover, and person-environment congruenceissues.

Michelle K. Duffy ([email protected]) is an associ-ate professor in the Carlson School of Management at theUniversity of Minnesota. She received her Ph.D. in man-agement from the University of Arkansas. Her researchinterests include social undermining, moral disengage-ment, and team processes.

Carolyn Wiethoff ([email protected]) is a clinicalassistant professor of management and the director of theKelley Women in Business Institute at the Kelley Schoolof Business at Indiana University–Bloomington. She re-ceived her Ph.D. from The Ohio State University. Herresearch explores diversity in the workplace and trust.


Documents

LEVEL AND DISPERSION OF SATISFACTION IN TEAMS: USING