the impact of social context on the relationship …

31
THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP BETWEEN INDIVIDUAL JOB SATISFACTION AND ABSENTEEISM: THE ROLES OF DIFFERENT FOCI OF JOB SATISFACTION AND WORK-UNIT ABSENTEEISM STEFAN DIESTEL Leibniz Research Centre for Working Environment and Human Factors JÜRGEN WEGGE Institute for Work, Organizational and Social Psychology, TU Dresden KLAUS-HELMUT SCHMIDT Leibniz Research Centre for Working Environment and Human Factors Building upon recent conceptualizations of different foci of job satisfaction and theo- ries of social-contextual influence, we develop and test an integrative cross-level model of the individual relationships between both “externally focused satisfaction” (refer- ring to job conditions) and “internally focused satisfaction” (referring to the work unit) and absenteeism. For both of these foci, we hypothesize differential three-way inter- active effects of work-unit absenteeism patterns as characterized by their mean and dispersion levels, as well as individual satisfaction levels on subsequent individual absenteeism. Based on two German multi-level samples, our analyses demonstrate that the negative relationship between externally focused satisfaction and individual ab- senteeism is strongest in the presence of high mean and dispersion levels of work-unit absenteeism, whereas this relationship is weaker when either the mean or dispersion levels of work-unit absenteeism, or both, are low. In contrast, the negative relationship between internally focused satisfaction and individual absenteeism is strongest under conditions of low mean and dispersion levels of work-unit absenteeism, whereas this relationship is weaker when either the mean or dispersion levels of work-unit absen- teeism, or both, are high. The present findings suggest that simultaneously improving individual internally focused satisfaction and reducing work-unit absenteeism is the most promising approach to reducing individual absenteeism. Given the high costs of losses in productivity resulting from absenteeism and the theoretical complexity of attitudes toward work and with- drawal from work, scholars and managers continue to devote their attention to the effects of work- related attitudes, especially the effect of job satis- faction on absenteeism (Dineen, Noe, Shaw, Duffy, & Wiethoff, 2007; Hausknecht, Hiller, & Vance, 2008; Wegge, Schmidt, Parkes, & van Dick, 2007). According to findings from meta-analyses, how- ever, the moderate correlation between job satisfac- tion and absenteeism, which ranges from .21 to .23 (Hackett, 1989; see also Harrison, Newman, & Roth, 2006), may tempt scholars to conclude that job satisfaction is not a trustworthy predictor of absenteeism. Nevertheless, the prevailing scholarly interest in this relationship results from the empir- ical observation that job attitude–absenteeism rela- tions are notably susceptible to the moderating ef- fects of variables that are embedded in the job-related context (Biron & Bamberger, 2012; Hausknecht et al., 2008; Xie & Johns, 2000). Al- though a focus on such combined effects seems promising when explaining the satisfaction–absen- teeism relation, the limited predictive power may also be due to the mono-level view that has been predominantly applied in past research. In extend- ing this view, recent theoretical developments and empirical findings regarding the role of the social context call for a broader perspective that explicitly accounts for contextual cues and stimuli from We thank the action editor Jason D. Shaw and three anonymous reviewers for helpful comments and suggestions. 353 Academy of Management Journal 2014, Vol. 57, No. 2, 353–382. http://dx.doi.org/10.5465/amj.2010.1087 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 express written permission. Users may print, download, or email articles for individual use only.

Upload: others

Post on 24-Jun-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIPBETWEEN INDIVIDUAL JOB SATISFACTION AND

ABSENTEEISM: THE ROLES OF DIFFERENT FOCI OF JOBSATISFACTION AND WORK-UNIT ABSENTEEISM

STEFAN DIESTELLeibniz Research Centre for Working Environment and Human Factors

JÜRGEN WEGGEInstitute for Work, Organizational and Social Psychology, TU Dresden

KLAUS-HELMUT SCHMIDTLeibniz Research Centre for Working Environment and Human Factors

Building upon recent conceptualizations of different foci of job satisfaction and theo-ries of social-contextual influence, we develop and test an integrative cross-level modelof the individual relationships between both “externally focused satisfaction” (refer-ring to job conditions) and “internally focused satisfaction” (referring to the work unit)and absenteeism. For both of these foci, we hypothesize differential three-way inter-active effects of work-unit absenteeism patterns as characterized by their mean anddispersion levels, as well as individual satisfaction levels on subsequent individualabsenteeism. Based on two German multi-level samples, our analyses demonstrate thatthe negative relationship between externally focused satisfaction and individual ab-senteeism is strongest in the presence of high mean and dispersion levels of work-unitabsenteeism, whereas this relationship is weaker when either the mean or dispersionlevels of work-unit absenteeism, or both, are low. In contrast, the negative relationshipbetween internally focused satisfaction and individual absenteeism is strongest underconditions of low mean and dispersion levels of work-unit absenteeism, whereas thisrelationship is weaker when either the mean or dispersion levels of work-unit absen-teeism, or both, are high. The present findings suggest that simultaneously improvingindividual internally focused satisfaction and reducing work-unit absenteeism is themost promising approach to reducing individual absenteeism.

Given the high costs of losses in productivityresulting from absenteeism and the theoreticalcomplexity of attitudes toward work and with-drawal from work, scholars and managers continueto devote their attention to the effects of work-related attitudes, especially the effect of job satis-faction on absenteeism (Dineen, Noe, Shaw, Duffy,& Wiethoff, 2007; Hausknecht, Hiller, & Vance,2008; Wegge, Schmidt, Parkes, & van Dick, 2007).According to findings from meta-analyses, how-ever, the moderate correlation between job satisfac-tion and absenteeism, which ranges from �.21 to.23 (Hackett, 1989; see also Harrison, Newman, &

Roth, 2006), may tempt scholars to conclude thatjob satisfaction is not a trustworthy predictor ofabsenteeism. Nevertheless, the prevailing scholarlyinterest in this relationship results from the empir-ical observation that job attitude–absenteeism rela-tions are notably susceptible to the moderating ef-fects of variables that are embedded in thejob-related context (Biron & Bamberger, 2012;Hausknecht et al., 2008; Xie & Johns, 2000). Al-though a focus on such combined effects seemspromising when explaining the satisfaction–absen-teeism relation, the limited predictive power mayalso be due to the mono-level view that has beenpredominantly applied in past research. In extend-ing this view, recent theoretical developments andempirical findings regarding the role of the socialcontext call for a broader perspective that explicitlyaccounts for contextual cues and stimuli from

We thank the action editor Jason D. Shaw and threeanonymous reviewers for helpful comments andsuggestions.

353

� Academy of Management Journal2014, Vol. 57, No. 2, 353–382.http://dx.doi.org/10.5465/amj.2010.1087

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.

Page 2: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

higher organizational levels when predicting with-drawal behavior on the basis of individual job atti-tudes (Johns, 2006; Liu, Mitchell, Lee, Holtom, &Hinkin, 2012).

In addition to the empirical evidence of the in-fluential role of the social context on withdrawalbehavior (Bamberger & Biron, 2007; Felps, Mitch-ell, Hekman, Lee, Holtom, & Harman, 2009), thereare at least two more interrelated reasons for usinga multi-level framework to disentangle the satisfac-tion–absenteeism relationship. First, for the pur-poses of high effectiveness and productivity, agrowing number of organizations have adoptedwork-unit structures that persist over time (Mes-mer-Magnus & DeChurch, 2009). Therefore, to pro-vide a comprehensive understanding of organiza-tional behaviors such as absenteeism and to deriveeffective managerial recommendations, theories onjob satisfaction and absenteeism must take thecharacteristics of work units that represent distinctsocial-contextual features into account. Second,and more importantly, scholars have characterizedabsenteeism as a meaningful organizational eventthat is perceived and evaluated by members of awork unit and, thus, may be “closely tied to socialand normative expectations particular to workgroups” (Hausknecht et al., 2008: 1223). Consistentwith this conceptualization, past studies have re-peatedly reported considerable amounts of vari-ance in absenteeism between (even highly homoge-neous) work units (Chadwick-Jones, Nicholson, &Brown, 1982; Harrison & Martocchio, 1998; Ma-thieu & Kohler, 1990). Moreover, shared under-standings or normative expectations in terms ofabsence cultures (Bamberger & Biron, 2007; Xie &Johns, 2000) and mean levels of work-unit absen-teeism (Mathieu & Kohler, 1990; Schmidt, 2002)have been found to predict individual absenteeism(for a review, see Rentsch & Steel, 2003); that is, theoccurrence of absenteeism is highly influenced bysocial-contextual features of work units that mayalso determine the degree to which job satisfactionlevels relate to patterns of absence behavior. Thisproposition relies on the empirically well-foundednotion that attitude–behavior relations are influ-enced by the social context (Mischel, 1977; Wal-lace, Paulson, Lord, & Bond, 2005).

However, extant theories have only consideredthe implications of social-contextual influences onthe satisfaction–absenteeism relationship at higherorganizational levels. For example, Dineen et al.(2007) developed a framework that explains howdifferent foci of job satisfaction interact with social-

contextual factors to predict absence rates in teams.Their framework relied on the conceptual dichot-omy of “internal” (referring to the team or workunit) and “external” (referring to the job or organi-zation) foci of work-related attitudes (Siders,George, & Dharwadkar, 2001). The differential in-teractions of both foci with social-contextual fac-tors found in the study by Dineen et al. (2007)demonstrate that the two foci of satisfaction playdifferent roles in predicting absenteeism. However,because different psychological mechanisms mate-rialize at different organizational levels, especiallywith respect to withdrawal behavior (Johns, 2006;Kozlowski & Klein, 2000; Rentsch & Steel, 2003),further research regarding the specific conditionsunder which employees decide to “take a sickie”when they are dissatisfied with certain aspects oftheir job is warranted.

To further contribute to the scholarly under-standing of the satisfaction–absenteeism relation-ship at the individual level, we develop and test amulti-level model that accounts for both foci of jobsatisfaction and social context (see Figure 1). Weconceptualize social contexts via combinations ofthe level and the consistency of contextual featuresthat have been found to jointly influence individ-ual behavior (Liao, Liu, & Loi, 2010; Liu et al.,2012). In particular, on the basis of Chan’s (1998)typology of composition models, the strength of asocial context is defined by the mean levels ofwork-unit absenteeism, which are normative refer-ence points, and by the dispersion levels of work-unit absenteeism, which indicate the uniformity ofabsence behavior within a work unit. Social infor-mation processing theory (Salancik & Pfeffer, 1978)and attitude representation theory (Lord & Lepper,1999) suggest different impacts of the absence con-text (mean and dispersion levels of absence behav-ior) depending on the satisfaction focus. Therefore,we derive distinct predictions of how each of thesatisfaction foci interacts with work-unit absentee-ism to influence individual absenteeism.

In explaining our model, we provide a concep-tual definition of the different job satisfaction foci.Next, we elaborate the theoretical and empiricalarguments for conceptualizing the absence contextsof work units through both their mean and theirdispersion levels of absence. Finally, we presentour hypotheses regarding the moderating effects ofwork-unit absenteeism on the negative relationshipbetween individual externally and internally fo-cused satisfaction on the one hand and individualabsenteeism on the other.

354 AprilAcademy of Management Journal

Page 3: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

THEORETICAL BACKGROUND ANDHYPOTHESES DEVELOPMENT

Job Satisfaction Foci and Their Relationshipswith Absenteeism

According to widely accepted definitions, indi-vidual job satisfaction is a work-related attitudethat reflects the extent to which an employee eval-uates certain aspects of his or her job—such asco-workers, the supervisor, career opportunities,the organization, and working conditions—as ben-eficial to him or her (Hausknecht et al., 2008;Locke, 1976; Schleicher, Watt, & Greguras, 2004;Weiss, 2002). In light of these different aspects, agrowing body of research points to the necessity ofdistinguishing between different targets or foci ofjob-related attitudes (Becker, 1992; Dineen et al.,2007; Siders et al., 2001). This necessity is height-ened by the empirical finding that different foci ofjob attitudes exhibit different relationships withorganizational outcomes, such as performance andabsenteeism. Based on Siders et al.’s (2001) conclu-sion that the foci distinction is useful in explainingthe typically weak relationship between broadermeasures of attitudes and organizational outcomes(see also Schleicher et al., 2004), Dineen et al.(2007) differentiated between externally and inter-

nally focused job satisfaction. According to theirdefinition, “externally focused satisfaction” re-flects employees’ attitudes concerning their jobs ingeneral, whereas “internally focused satisfaction”is defined as the extent to which an employee de-rives enjoyment from working with others in his orher work unit. The attributes “external” and “inter-nal” indicate whether the attitudinal targets gener-alize across work units and characterize the job ingeneral or are specific to the work unit, respec-tively. To disentangle the role of social context inthe relationship between job satisfaction and ab-senteeism, Dineen et al. (2007) noted that researchon this issue should take different foci or targets ofjob satisfaction into account because they have dif-ferent influences on the way that employees re-spond to the social context. In the present study,we predict that specific patterns of interactions be-tween social contexts and individual satisfactionlevels on absenteeism are contingent upon the atti-tudinal focus.

Our prediction draws on attitude representationtheory (Lord & Lepper, 1999; Watt, Maio, Rees, &Hewstone, 2007), which proposes that an attitudi-nal response is mainly influenced by how the atti-tudinal target is mentally represented. According tothis theory, specific features or characteristics of an

FIGURE 1Theoretical Framework

2014 355Diestel, Wegge, and Schmidt

Page 4: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

attitudinal target determine its representation,which shapes the function of the attitude. Thefunction of an attitude characterizes its meaningand regulates how people respond to both the atti-tudinal object and the social context (Snyder &DeBono, 1989). Abelson and Prentice (1989) distin-guished between “instrumental functions,” whichprovide information about the usefulness of theattitudinal object from an individual’s point ofview, and “social-adjustive functions,” which ex-press group identity needs and indicate the extentto which an individual perceives herself or himselfas integrated into a group (see also Smith, Bruner, &White, 1956). On a conceptual level, with careeropportunities, working conditions, and manage-ment included in the focus, we advance “externallyfocused job satisfaction” as an attitude that fulfillsan instrumental function. In defining the attitudi-nal targets of externally focused job satisfaction, weconcentrate on features that are more task-related(resources, equipment, and facilities) or instrumen-tally based (opportunities for advancement andqualification, rules and regulations within the or-ganization), rather than features that are value- oraffect-based (Weiss, 2002). In contrast, includingentities of the social context or work unit (col-leagues and supervisor) in the focus, we conceptu-alize “internally focused job satisfaction” as an at-titude that fulfills social-adjustive functions.Because employees derive group identity fromtheir work unit that is defined by an interdepen-dent collection of individuals and provides a basisfor familiarity, cohesion, and reciprocal social sup-port (Moreland & Levine, 2001; van Knippenberg &van Schie, 2000), satisfaction with the colleaguesand supervisor in the work unit should indicate thedegree of social integration or identification that anemployee experiences as a result of interactionswithin his or her work unit (Dineen et al., 2007:626).

Research indicates that instrumental and social-adjustive attitudes differ in how they relate to be-havioral outcomes and interact with social-contex-tual influences (Ledgerwood & Trope, 2010; Watt etal., 2007). In the case of an instrumental attitude, abehavioral response primarily serves instrumentalor calculative purposes, and, thus, derives from amore global evaluation, which involves the integra-tion of information from different sources, e.g., thesocial context. In line with this argument, Harrisonet al. (2006) posited that absenteeism is a behav-ioral strategy or a controllable form of input reduc-tion that aims to maintain a balanced exchange of

resources between the employee and the organiza-tion (reducing the level of investment in an unbal-anced relationship with the employer or using thetime to find alternative jobs). One implication ofthis conceptualization is that, when more instru-mental aspects of the job are in the satisfactionfocus, employees consider or interpret social-con-textual information within the environment (workunit) as potential constraints or opportunities thataffect their response to their evaluation of the ex-ternal target (Biron & Bamberger, 2012; Salancik &Pfeffer, 1978). Thus, we predict that the social con-text moderates the effects of individual externallyfocused satisfaction on absenteeism. Some empiri-cal support for our prediction has been provided byHausknecht et al. (2008), who reported that thenegative relationship between both job satisfactionand organizational commitment and absenteeism atthe work-unit level is attenuated as a function ofthe local unemployment rate. According to theirline of reasoning, dissatisfied employees or thosewith a low level of commitment interpret absentee-ism as a risky strategy because a high unemploy-ment rate as a contextual cue indicates a scarcity ofpotential job alternatives (see also Markham, 1985).

In contrast, social-adjustive attitudes function tocoordinate social actions and interactions withinan interdependent collection of individuals and theresulting behavioral response is mainly directed atsocial alignment with or avoidance of the attitudi-nal objects (e.g., work-unit colleagues or the super-visor). Consistent with this view, Gellatly and Al-len (2012) argued that high or low levels ofindividual absence behavior can also serve as aform of social alignment whereby employees dis-play compliance with a group norm. Conversely,low levels of work-unit identification result inalienation and demotivation that manifest in avoid-ance or withdrawal behaviors, such as absenteeism(Dineen et al., 2007; van Dick, van Knippenberg,Kerschreiter, Hertel, & Wieseke, 2008). Thus, whenemployees evaluate their satisfaction with the en-tities of the social context (work-unit colleaguesand the supervisor), their attitudinal responses (ab-senteeism) reflect their level of enjoyment resultingfrom social interaction within the work unit. Inother words, the social context directly guides afocal employee’s behavior if internally focused sat-isfaction, which determines behavioral alignmentwith the context, is high. Bamberger and Biron(2007) have provided some support for this line ofreasoning by showing that co-workers’ absencenorms were most strongly related to employees’

356 AprilAcademy of Management Journal

Page 5: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

absence behavior when the focal employees re-ported that they were prone to being influenced bypeers in their group. This finding suggests that,even in the case of high work-unit satisfaction,employees are likely to engage in withdrawal be-havior if the social context in which they are em-bedded provides strong cues for withdrawing (Liuet al., 2012).

Conceptualization of Work-Unit Absenteeism asa Social-Context Factor

In light of the hypothesized role of the socialcontext in the relationships between both satisfac-tion foci and absenteeism, research must define therelevant social-contextual facets or manifestationsthat are most likely to affect these relationships.Mowday and Sutton (1993: 198) defined context as“stimuli and phenomena [such as behavior of oth-ers] that surround and thus exist in the environ-ment external to the individual, most often at adifferent level of analyses.” On the basis of thisdefinition, Johns (2006: 386) argued that the socialcontext is a strong influential factor that “affect[s]the occurrence and meaning of organizational be-havior as well as functional relationships betweenvariables.” In support, previous studies have re-peatedly demonstrated that the social context notonly influences behavior but also shapes the con-ditions under which the functional relationshipsbetween individual job attitudes and behavioraloutcomes materialize (Bamberger & Biron, 2007;Duffy, Scott, Shaw, Tepper, & Aquino, 2012; Liu etal., 2012).

For two interrelated theoretical reasons, we as-sert that work-unit absenteeism is the most salientand influential social-contextual cue or stimulusthat influences the satisfaction–absenteeism rela-tionship (Mathieu & Kohler, 1990; Schmidt, 2002).First, Ajzen and Fishbein’s (1977) compatibilityprinciple predicts that attitudes are only related tocorresponding behavior if the attitudes, context,and behavior match each other; that is, the degreeof correspondence between the context and the tar-get behavior determines whether and to what ex-tent attitudes influence behavior. Specifically, peo-ple embedded in groups are particularly sensitiveto contextual stimuli with respect to their focalbehaviors, and their responses to their own atti-tudes are mostly likely to be influenced by thecorresponding behavioral patterns of their group.Because employees recognize whether their ownabsence behavior matches the patterns of absentee-

ism in their social environment (e.g., work unit),the degree to which their absence behaviors corre-spond to their satisfaction levels should be influ-enced by the absence patterns within their workunit (Johns & Nicholson, 1982; Mason & Griffin,2003). Several studies have demonstrated that in-dividual patterns of a specific behavior, such as jobsearch behavior and tardiness, are often highly re-lated to corresponding patterns of that behavior atthe work-unit level (Eder & Eisenberger, 2008;Felps et al., 2009).

Second, social conformity theory (Asch, 1956,1966; Sherif, 1936) asserts that groups develop im-plicit or explicit expectations and norms concern-ing meaningful behavior (such as absenteeism) due,for example, to convergence effects, social confor-mity pressures, or the intention to stabilize or en-hance a self-image that is consistent with the per-ceived norms of the group (Bamberger & Biron,2007). In contrast to absence cultures, which areexpected to be explicitly perceived and adopted bythe members of work units in which such culturesexists (Xie & Johns, 2000), we argue that work-unitabsenteeism exerts its effects through behavioralconvergence, implicit norms, and alignment pro-cesses that are not manifest in a salient culture. Inline with this argument, Postmes, Spears, Sakhel,and De Groot (2001) reported that, even in anony-mous situations, individuals within groups de-velop implicit norms through the observation of thebehavioral patterns of others and that these normslead to behavioral convergence.

A growing body of multi-level research demon-strates that the extent to which the social context(work-unit absenteeism) influences an individual’sbehavior depends not only on the level of a partic-ular cue but also on the consistency of the contex-tual information surrounding it (see also Dineen etal., 2007; González-Roma, Peiró, & Tordera, 2002;Liu et al., 2012: 1364). This finding is in line withthe notion of situational strength (Cooper & Withey,2009; Mischel, 1977), according to which the com-bination of the level and consistency of contextualinformation determines the strength of a particularcontext. “Social context strength” is commonly de-fined as the degree to which an in-group situationis characterized by consistently high or low levelsof observable behavior across the members of aparticular group due to consensual notions regard-ing the meaning of the focal behavior, the situa-tional constraints, or the social agreements (e.g.,Schneider, Salvaggio, & Subirats, 2002). Strong sit-uations or strong contexts “lead everyone to con-

2014 357Diestel, Wegge, and Schmidt

Page 6: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

strue the particular events the same way, [and]induce uniform expectancies regarding the mostappropriate response pattern” (Mischel, 1977: 347).In other words, uniform patterns of behavior amonggroup members provide “clear and consistent cues”(Liu et al., 2012: 1365) through which employeescan gain a sense of their social environment andadapt their behavioral responses accordingly. Insupport, Liu et al. (2012) recently found that meanlevels of job satisfaction in work units were onlynegatively related to individual and work-unitturnover when the dispersion levels of job satisfac-tion were low. Therefore, the level of work-unitabsenteeism should influence individual absentee-ism only when work-unit members show uniformor consistent absence patterns.

Consequently, we conceptualize absence contextthrough both the mean and the dispersion levels ofwork-unit absenteeism. The mean levels of work-unit absenteeism are defined as normative refer-ence points (see also Hausknecht et al., 2008: 1224;Mathieu & Kohler, 1990). Our definition is in linewith Chan’s (1998) additive composition model,according to which the mean level of work-unitabsenteeism is a contextual cue that reflects thelevel of absence that is allowed within a work unit.To capture the contextual consistency that moder-ates the effect of the mean level, we advance dis-persion levels of work-unit absenteeism as the de-gree of uniformity in the absence patterns within awork unit. Consistent with Chan’s (1998) disper-sion composition model, high within-unit variancein individual absences reveals a low level of con-sistency in behavioral patterns and a weaker con-textual strength, whereas low within-unit varianceindicates a high degree of contextual strength thatis driven by agreements within a work unit, cohe-siveness, norms, or supervisors’ instructions (seealso Cole, Bedeian, Hirschfeld, & Vogel, 2011; Har-rison & Klein, 2007).

Adopting the concept of situational strength, wedelineate a context-based framework of how themean and dispersion levels of work-unit absencejointly moderate (cross-level, three-way interac-tions) the individual negative relationships be-tween both of the satisfaction foci and absenteeism(see Figure 2). Accordingly, externally focused sat-isfaction should only negatively relate to individ-ual absenteeism when both the mean and the dis-persion levels are high, indicating a weak context.Conversely, regardless of externally focused satis-faction, mean levels of work-unit absenteeismshould predict individual absenteeism when dis-

persion levels are low, indicating a strong context.In contrast, internally focused satisfaction shouldonly be negatively related to individual absentee-ism when both the mean and the dispersion levelsof work-unit absence are low, whereas individualabsenteeism should be higher at all levels of inter-nal satisfaction when a strong context favors highlevels of absenteeism (high mean and low disper-sion). In other words, for those with high levels ofinternally focused satisfaction, the mean levels ofwork-unit absenteeism predict individual absencewhen the dispersion is low, whereas those withlow levels of internal satisfaction should often beabsent regardless of their work unit’s absencepatterns.

Externally Focused Job Satisfaction, Work-UnitAbsenteeism and Patterns of IndividualAbsenteeism

One of the core premises of social informationprocessing theory is that employees within workunits gather information about appropriate behav-ioral patterns from their social environment anduse this information to adapt their attitudes andbehaviors accordingly (Salancik & Pfeffer, 1978:226). On the basis of this premise, Thomas andGriffin (1989) theorized that such information in-fluences the relationships between job attitudesand behavioral outcomes through evaluation andchoice processes. In other words, implicit normswithin groups, communication processes, socialconstraints, behavioral observations, and modelinginfluence the extent to which job attitudes manifestin specific behavioral patterns. In support, Wellen,Hogg, and Terry (1998) reported that the attitude–behavior relationship varies as a function of groupnorms. When salient group norms strongly suggestspecific behavioral patterns, participants are lesslikely to be influenced by their own attitudes andmost likely to follow those norms. According toMischel (1977), such conditions indicate a strongcontext that inhibits the effects of attitudes on be-havior. Because externally focused job satisfactionhas an instrumental function, we predict that astrong absence context attenuates the negative re-lationship between individual external satisfactionand individual absenteeism, whereas a weak ab-sence context strengthens this relationship. In thefollowing section, we elaborate on two theoreticalreasons for our prediction.

First, in the case of instrumental attitudes, behav-ioral responses draw on calculative or instrumental

358 AprilAcademy of Management Journal

Page 7: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

reasoning, global evaluation, and choice. Thus,when externally focused job satisfaction is low,employees should tend to gather relevant informa-tion from the social context to determine whetherabsenteeism is an appropriate reaction in their so-cial environment. An absence context that is char-acterized by low mean and dispersion levels ofwork-unit absenteeism reflects a low tolerancelevel and suggests that higher individual absentee-ism would be highly visible to other work-unitmembers and could lead to unpleasant conse-quences. Thus, low mean and dispersion levels associal-contextual cues (highly visible to employ-ees) should be evaluated as social constraints (e.g.,due to clear instructions from the supervisor)within the work unit that interdict absence behav-ior in cases of low externally focused satisfaction(see the lower left quadrant of Figure 2). Likewise,when absence behavior is consistently high withina work unit (see the upper left quadrant of Fig-

ure 2), employees may infer that high absenteeismis a part of the behavioral norms of their work unitand is thus somewhat expected (Gellatly & Allen,2012). Under such conditions, employees may un-deremphasize their own evaluation of the job ingeneral; that is, for instrumental purposes (e.g.,avoiding negative consequences or maintaining so-cial support), employees adapt or align their ab-sence behavior to the mean level of work-unitabsenteeism when dispersion levels are low, re-gardless of their levels of externally focused jobsatisfaction. Conversely, with increasing mean anddispersion levels of work-unit absenteeism, em-ployees should perceive their social environmentas less assertive because inconsistent social-contex-tual cues and a generally high level of absenteeismreveal a high tolerance for absence within a workunit. In such a weak context, a focal employee mayinterpret absenteeism as a possible alternative forreducing his or her investment in an unbalanced

FIGURE 2Expected Patterns of Relationships between Both of the Satisfaction Foci and Individual Absenteeism,

Contingent upon Mean and Dispersion Levels of Work-Unit Absenteeism

2014 359Diestel, Wegge, and Schmidt

Page 8: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

relationship with the employer or organization (seethe upper right quadrant of Figure 2).

Second, work units are social environments thatshape the meanings of organizational behaviors byproviding cues regarding the interpretation ofmeaningful events such as absenteeism (Johns,2006; Salancik & Pfeffer, 1978). Strong contextswith uniform absence levels may suggest that ab-senteeism is an inadequate or disadvantageous re-sponse to evaluations of the job because of thesocial interdependence among work-unit members(Dineen et al., 2007) or the cohesiveness within awork unit (Xie & Johns, 2000). Therefore, in thepresence of uniformly low or high absence patternswithin a work unit, a focal employee should per-ceive a highly restricted scope of interpretation ofabsenteeism and, thus, should adapt his or herbehavior in accordance with that of the work unit.Otherwise, a weak context characterized by bothhigh levels of and high variability in unit members’absence behavior may generate conflicting socialcues regarding the meaning of absenteeism. Thus,as social influence declines with increasing ambi-guity and tolerance of the context (Asch, 1966;Wallace et al., 2005), the interpretation of absencebehavior is strongly influenced by the focal em-ployee’s own evaluation, and actual individual ab-sence patterns should correspond to levels of exter-nally focused satisfaction when mean anddispersion levels are high. On the basis of thesearguments, we propose the following hypothesis:

Hypothesis 1. The mean and dispersion levelsof work-unit absenteeism jointly moderate thenegative relationship between externally fo-cused job satisfaction and individual absentee-ism (three-way interaction): the relationship isstrongest when the mean and dispersion levelsof work-unit absenteeism are high, whereas therelationship is weaker when either the mean orthe dispersion levels or both are low.

Internally Focused Job Satisfaction, Work-UnitAbsenteeism and Patterns of IndividualAbsenteeism

As we elaborated above, when entities of thesocial context (colleagues and the supervisor in thework unit) are in the attitudinal focus, internallyfocused satisfaction fulfills a social-adjustive func-tion and can be considered an indicator of the levelof social integration or identification that a focalemployee experiences in his or her work unit

(Dineen et al., 2007). An employee who is dissatis-fied with his or her supervisor and colleagues isunlikely to socially identify with the work unit.Research shows that with decreasing integration,people perceive themselves as disconnected fromothers in their social environment, resulting in so-cial differentiation (Duffy et al., 2012: 647; Opotow,1995) and withdrawal behavior (Schneider, 1987).In contrast, when employees are satisfied with theirwork-unit colleagues and supervisor, strong con-texts (e.g., uniformly high or low absenteeism)should have stronger impacts on their behavior be-cause social integration causes employees to alignwith the social environment from which they de-rive their identification (Tajfel, 1978).

Therefore, we predict that in the case of a strongcontext with a low level of absenteeism (low meanand low dispersion of work-unit absenteeism; seethe lower left quadrant of Figure 2), the negativerelationship between individual internal satisfac-tion and individual absenteeism should be strongerthan it is in other contextual conditions. With de-creasing levels of internal satisfaction, employeestend to be absent often because a low perceived fitor level of integration causes employees to be resis-tant to social-contextual influences and to avoidthe dissatisfying context. Thus, as the experience ofnot fitting in is likely to disconnect employees fromsocial constraints or implicit normative expecta-tions within their work unit, uniformly low ab-sence patterns as strong contextual cues shouldexert no impact on individual absenteeism. Re-search has repeatedly demonstrated that weakbonds between group members are associated withhigher levels of absenteeism (Keller, 1983; Xie &Johns, 2000). Group members do not blindly con-form to group norms; instead, conformity dependson the level of members’ in-group identification(Falomir-Pichastor, Gabarrot, & Mugny, 2009):Hekman, Steensma, Bigley, and Hereford (2009)showed that social influence on behavior declineswith decreasing identification. Conversely, as astrong context with low absence levels providesclear cues for consistent attendance at work, em-ployees who are satisfied with their work-unit col-leagues and supervisors should rarely be absentbecause they construe that their work unit expectslow absenteeism. Thus, in line with theories onattitudes and social influences in groups (Asch,1966; Wallace et al., 2005), patterns of uniformlylow levels of absence will most likely be adoptedby the focal employee if he or she strongly identi-

360 AprilAcademy of Management Journal

Page 9: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

fies with the work unit, as manifested in high levelsof internally focused satisfaction.

In contrast, a strong context that favors high lev-els of absenteeism (high mean and low dispersionof work-unit absenteeism; see the upper left quad-rant of Figure 2) is predicted to attenuate the neg-ative effects of individual internally focused satis-faction on individual absenteeism. In the presenceof such a context, individual absenteeism shouldbe high across all levels of internally focused satis-faction. Employees with high levels of internallyfocused satisfaction should adapt their absence be-havior to the level of absenteeism within their workunit, whereas those with low levels of internallyfocused satisfaction should often be absent becauseof their behavioral tendency to avoid the dissatis-fying targets (their supervisor and colleagues). Be-havioral alignment of those with higher levels ofsatisfaction can be explained by behavioral conta-gion and the spillover effects of withdrawal behav-ior. For example, Felps et al. (2009: 547) argued thatsocial comparisons among co-workers result in be-havioral adaptations (even toward risky and uncer-tain behaviors) because employees focus on the be-havioral patterns of others to obtain a clear sense ofthe meanings of those behaviors. Liu et al. (2012)posited that, when a focal employee observes consis-tent withdrawal-related patterns within the socialcontext, he or she may readily infer that withdrawalbehavior is a reasonable response, increasing the like-lihood of withdrawal. Thus, because behavioralalignment to the social context mainly results from apositive social-adjustive attitude toward the context,employees who are satisfied with their work-unit col-leagues and supervisor adapt their absence behavioraccordingly. Conversely, as noted above, low levels ofinternally focused satisfaction prompt employees towithdraw from their social environment. Thus, underconditions of uniformly high absenteeism patternswithin a work unit, higher individual absenteeismreflects avoidance behavior of those with low levelsof satisfaction regarding their work-unit colleaguesand supervisor.

Finally, with decreasing consistency of behavioralpatterns or diminishing strength of the social context(in terms of higher variability), employees are lessable to develop a concept of absenteeism that con-forms to that of the work unit, thus social influencedeclines. In other words, in the presence of a highdegree of absence dispersion within a work unit (seethe two right-hand quadrants of Figure 2), the indi-vidual relationship between internally focused satis-faction and absenteeism should be weaker compared

to a case in which there is a low mean and dispersionof work-unit absenteeism. Integrating all of the rele-vant contextual conditions, we propose the followinghypothesis:

Hypothesis 2. The mean and dispersion levelsof work-unit absenteeism jointly moderate thenegative relationship between internally fo-cused satisfaction and individual absenteeism(three-way interaction): the relationship isstrongest when the mean level and dispersionof work-unit absenteeism are low, whereas therelationship is weaker when either the mean orthe dispersion levels or both are high.

OVERVIEW OF THE PRESENT RESEARCH

We tested our predictions in two studies thatwere conducted in different German organizations.In the first study, we surveyed health-care workerswho provided care for elderly people. Past researchhas repeatedly found that nursing organizations areoften faced with high absence rates (Schmidt,2010). To test whether the hypothesized interac-tions are generalizable across different occupa-tional and organizational settings, we conducted asecond study in a tax and revenue organization inwhich tasks, procedures, and processes are highlyformalized. Both datasets had nested structuresthat allowed for multi-level modeling.

STUDY 1—METHODS

Participants and Procedures

Participants in this study were health-care work-ers at a residential elderly care organization locatedin a federal state in Germany. The organization is alimited company operating as a public utility and,thus, belongs to the services sector. All participantswere involved in the daily care of elderly people,which included physical care, medical support,and social activities. Approximately 80% of theparticipants’ daily working time was invested ininteractions with elderly people. Participants wererecruited through announcements at staff meetingsand memos sent by the work-unit managers. Partic-ipants were assured that completing the question-naire was voluntary and that their data would re-main confidential. A response rate of 75% yieldeda sample of 432 participants who completed thequestionnaire and consented to the use of theirabsenteeism data, which were assessed before andafter the survey was administered. After the survey,

2014 361Diestel, Wegge, and Schmidt

Page 10: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

20 participants left the organization (final n � 412).According to a one-way analysis of variance(ANOVA), the mean scores for the study variables(absenteeism and satisfaction levels) of those wholeft the organization did not differ from those whoremained in the study (p � .15). Participants wereorganized into 63 work units (MSIZE � 6.54, range �3–14) that remained intact throughout the study.One supervisor led each work unit. The averageage of the participants was 39.5 years (SD � 9.99,range � 19 – 64). The organizational tenure of theparticipants ranged from 0 to 35 years (M � 6.67,SD � 6.1), and 88% were women. All participantshad completed junior high school and had beentrained as geriatric nurses.

Measures

Individual externally and internally focusedjob satisfaction. Using Kunin’s (1955) faces, jobsatisfaction was assessed on a five-item scale thatwas developed by Neuberger and Allerbeck (1978).Consistent with Dineen et al.’s (2007; Study 1) opera-tionalization of both satisfaction foci, the itemwording was similar to that of the items developedby Cammann, Fichman, Jenkins, and Klesh (1983).We assessed externally focused satisfaction usingthree items (� � .83) that relate to working condi-tions (“All in all, are you satisfied with your work-ing conditions?”), organization and management(“All in all, are you satisfied with the organizationand management?”), and career opportunities (“Allin all, are you satisfied with your career opportu-nities?”). For each item, the scale provided a clearexplanation of the reference object:—working con-ditions: “This question relates to the general con-ditions under which you are working (e.g., re-sources, equipment, facilities, etc.)”; organizationand management: “This question relates to howyou view the organization as a whole, how differentwork units cooperate, and how you evaluate proce-dures, rules, and regulations, as well as the uppermanagement”; career opportunities: “This questionrelates to your personal development (your pastand current opportunities for advancement, quali-fication, and job enlargement).” Internally focusedsatisfaction was measured with two items (� � .77)that referred to work-unit colleagues and the work-unit supervisor (“All in all, are you satisfied withyour colleagues?” and “All in all, are you satisfiedwith your supervisor?”). To ensure that both itemswere answered in reference to the work unit, weagain provided additional explanations for each

item—colleagues: “This question relates to the col-leagues with whom you directly work and inter-act”; supervisor: “This question relates to your di-rect supervisor, who is one hierarchical positionabove you and gives you directions.” Confirmatoryfactor analyses supported the distinctness of bothof the satisfaction foci: a two-factor model had abetter fit (�2 � 3.76, df � 4, p � .44, RMSEA � .00,95% CI [.00–.07], CFI � 1.00, SRMR � .01) thandid a one-factor model (�2 � 170.18, df � 5, p � .01,RMSEA � .28, 95% CI [.25–.32], CFI � .72,SRMR � .11). The analysis of the intraclass corre-lations (ICC(1)) of both dimensions lent furthersupport to the supposed foci distinction. Internallyfocused satisfaction should indicate the level ofsocial integration within a work unit (Dineen et al.,2007) and, thus, may vary substantially betweenwork units, whereas externally focused satisfactionreflects individual evaluations of targets that do notrefer to the work unit and, thus, should be lessconsistent across work unit members. Supportingour conceptualization, 18% of the variation ininternal satisfaction resided between the workunits. In comparison, an ICC(1) of .08 showedthat the between-group variance in external sat-isfaction was substantially lower. Because thesurvey items associated with colleagues and su-pervisors were presented before the other items,it seems unlikely that the participants interpretedthe terms “conditions” or “management” to in-clude colleagues or supervisors.

Individual and work-unit absenteeism. Weused “absence frequency” (the number of absenceevents) to assess voluntary absenteeism (Sagie,1998). Absence frequency referred to short-term ab-sences that included absences of fewer than fourconsecutive workdays. Absences due to vacationdays or holidays, maternity leave, military service,participation in training courses, or long-term ab-sences that encompassed more than three consec-utive workdays were not included in the data. Ac-cording to German Civil Code, employees have toprovide documentation from a physician if theirabsence encompasses more than three days (long-term absence). Consistent with Hausknecht et al.(2008), long-term absences were therefore assumedto be medically, rather than motivationally, based.The management of the nursing homes did notapply official regulations or procedures to reduceor to sanction short-term absenteeism (such as set-ting goals or providing incentives to avoid absen-teeism). In addition, extra days off or personal dayswere not used as rewards, and work-unit managers

362 AprilAcademy of Management Journal

Page 11: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

did not receive any reports on the absences of theiremployees. Provided that no official procedureswere applied by the management, our measuredid not distinguish between unexcused absencesand excused absences that may have been sanc-tioned by the work-unit manager (see also Sagie,1998: 162). The absenteeism data were obtainedfrom archival records and were restricted to a pe-riod of 6 months before (referred to hereafter as“pre 6 months”) and 6 months after (hereafter “post6 months”) the administration of the question-naires. Because the distribution of absence fre-quency deviated from the thresholds that are com-monly seen as critical for unbiased parameterestimations (Hammer & Landau, 1981), all individ-ual raw scores were subjected to a square root trans-formation (Clegg, 1983). After this transformation,the skewness and kurtosis of the pre- and post-questionnaire absence frequencies were eitherwithin or below the ranges deemed acceptable byHammer and Landau (1981; see also Steel (2003).

The transformed archival data were also used tomeasure the mean and dispersion levels of work-unit absenteeism. Consistent with previous re-search on work-unit absenteeism (Hausknecht etal., 2008; Schmidt, 2002), we averaged the absencefrequency of employees within each work unit. AnICC(1) of .23 indicated a variation of 23% in ab-sence frequency (pre 6 months) between all 63work-units. An ICC(2) of .66 also justified the meanaggregation of absence frequency (pre 6 months) tothe work-unit level.

In line with Chan’s (1998) dispersion model,which utilizes within-group variation and has beenput forward to conceptualize consistency, the dis-persion levels of work-unit absenteeism were opera-tionalized as the standard deviations between thework-unit employee absence frequency levels.1 Our

operationalization draws on Harrison and Klein’s(2007) separation concept; that is, differences inposition along a continuum represent dissimilarityin behavior. Meeting Cooper and Withey’s (2009)requirement to confirm that the situational strengthsystematically varied across the conditions of thestudy, Bartlett’s test (Snedecor & Cochran, 1983)showed significant differences in the variances ofabsence frequency (pre 6 months) between thework units (p � .01).

Consistent with recommendations from priorlongitudinal research on absenteeism (Mathieu &Kohler, 1990), we controlled for individual absen-teeism in the 6 months prior to the questionnairesurvey when predicting absence frequency in the6 months after the questionnaire survey in themulti-level analyses.

Control variables. Because previous researchhas revealed that age, gender, and organizationaltenure are significant predictors of individual ab-senteeism (e.g., George, 1989; Wegge et al., 2007),we controlled for these variables in predicting ab-senteeism. In addition, Hausknecht et al. (2008)have reported that the local unemployment rate isrelated to absenteeism. Because the perception ofhaving job alternatives is a significant precursor towithdrawal behavior (Trevor, 2001), and becausethe work units in this study were located in differ-ent regions, we introduced the local unemploy-ment rate as a control variable at the work unitlevel. The unemployment data were obtained fromthe Federal Statistical Office of Germany (2012) foreach district in which a work unit operated. Fi-nally, because medical reasons cannot be com-pletely ruled out in predicting individual short-term absenteeism, we included a measure of healthdisorders comprising a broad spectrum of physicaland psychological symptoms (24 items; 4-level re-sponse format ranging from 1 � “not at all” to 4 �“a great deal”; � � .91), such as headaches, insom-nia, poor concentration, neck pain, rheumatic1 Our conceptualization of work-unit absenteeism sug-

gests that all work-unit members constitute the socialcontext, and the absence aggregates should capture theabsence patterns within the whole work unit. Conse-quently, a low response rate might affect the validity ofour measure because non-responders in our study mayhave higher absence rates. To provide further evidencefor the validity of the absence aggregates in our study, wetested whether the presence of study dropouts withinwork units would considerably change the values of ourabsence aggregates. Specifically, we chose work unitswith response rates of at least 90% (max. one membermissing: Nlevel 2 � 29; 23 work units had a response rateof 100%) and randomly dropped members to create re-

sponse rates of approximately 80 and 70%. After reduc-ing the sample sizes within the work units, the mean anddispersion levels of work-unit absence frequency wererecalculated and correlated with the mean and disper-sion levels of the 90% response rate values. The high andsignificant correlations (approx. 80% response rate:rmean levels � .98; p � .01/rdispersion levels � .96; p � .01;approx. 70% response rate: rmean levels � .96; p � .01/rdispersion levels � .86; p � .01) indicated that study drop-outs would not considerably bias our assessment ofwork-unit absenteeism in the nursing sample.

2014 363Diestel, Wegge, and Schmidt

Page 12: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

pains, nausea, and shortness of breath (von Zers-sen, 1976). Past research has repeatedly docu-mented significant relationships between thesesymptoms and individual absenteeism (Darr &Johns, 2008; Schmidt, 2010).

STUDY 1—RESULTS

Table 1 shows the descriptive statistics and cor-relations for all variables. We used hierarchical lin-ear modeling (HLM; Raudenbush & Bryk, 2002) totest our hypotheses because our model (Figure 1)predicts cross-level effects, and our data had anested structure.

Table 2 presents the HLM results. First, we spec-ified and tested a null model without independentvariables. Second, we entered all of the individual-and work-unit-level variables (Model 1: controlvariables at level 1 and level 2, previous individualabsenteeism, both satisfaction foci at level 1, andmean and dispersion levels of work-unit absentee-ism at level 2). Next, we estimated the cross-level,two-way interactions in a slope-as-outcomes model(Model 2). To avoid the biasing effects of multicol-linearity (Aiken & West, 1991), the mean and dis-persion levels of work-unit absence frequency (pre6 months) were standardized prior to calculatingthe two-way interaction of these variables. Finally,we simultaneously analyzed both of the hypothe-sized cross-level, three-way interactions betweenboth of the satisfaction foci and both work-unitabsenteeism variables (Model 3). To provide anunbiased estimate of the cross-level interactions,we group-mean centered both satisfaction vari-ables (Raudenbush & Bryk, 2002). To separatelevel 1 and level 2 effects and reduce possibleproblems with multicollinearity at both levels

(Hofmann & Gavin, 1998), we centered all otherlevel 1 and level 2 variables (except for gender)around their grand means (see also Erdogan &Enders, 2007: 325).

The null model allowed us to determine the be-tween-work unit variance in individual absencefrequency (post 6 months). An ICC(1) of .19 indi-cated that 19% of the variance to be explained bylevel 2 variables resided between work units. Aprecondition for analyzing cross-level interactionsis that the slopes of the relation between both sat-isfaction foci and absence frequency (post 6months) vary across the work units. HLM estimatesrevealed that the slope variances for both of thesatisfaction foci were significant (externally fo-cused satisfaction: U1 � .02, p � .05; internallyfocused satisfaction: U1 � .03, p � .05).

Main effects at level 1 and 2. As Table 2 (Model1) shows, health disorders (� � .17, p � .05) and themean levels of absence frequency (pre 6 months)(� � .23, p � .001) were significantly related toabsence frequency (post 6 months) after controllingfor previous levels of absence frequency (pre 6months) at level 1. Both of the individual satisfac-tion foci exhibited significant and negative rela-tionships to absence frequency (externally focusedsatisfaction: � � �.16, p � .01; internally focusedsatisfaction: � � �.12, p � .05).

Cross-level, three-way interactions (Hypothe-ses 1 and 2). We predicted two cross-level, three-way interaction effects of the mean and dispersionlevels of work-unit absenteeism and both job satis-faction foci on subsequent individual absenteeism.As seen in Model 3 of Table 2, HLM estimationsrevealed significant three-way interactions for ex-ternally (� � �.08, p � .01) and internally (� ��.09, p � .01) focused satisfaction. Both interac-

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

1.1: Individual-Level Variables

Variable M SD 1 2 3 4 5 6 7 8

1. Age 39.54 9.99 —2. Gender (1 � female; 2 � male) 1.12 0.32 �.10* —3. Tenure 6.67 6.13 .30** �.07 —4. Health disorders 0.90 0.50 .11* �.09 .05 —5. Absence frequency (pre 6 months) 0.95 1.40 �.02 �.04 .08 .12* —6. Externally focused satisfaction 4.90 1.28 .11* �.03 .02 �.11* �.13** —7. Internally focused satisfaction 5.51 1.19 .03 �.04 .03 �.14** �.14** .33** —8. Absence frequency (post 6 months) 0.93 1.35 �.08 .01 �.05 .20** .15** �.40** �.34** —

Notes. (n � 412). Descriptive statistics of absence frequency represent non-transformed scores.

* p � .05; ** p � .01

364 AprilAcademy of Management Journal

Page 13: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

tions explained an additional 3% of the variance(and, in combination with conditional two-wayinteractions, an additional 23% of the variance)in absence frequency (post 6 months). To facili-tate the interpretation of the different interactionpatterns, we conducted simple slope analyses(Preacher, Curran, & Bauer, 2006; Shacham,2009) and visualized the interactions using theprocedure described by Aiken and West (1991)(Figure 3 and Figure 4).

Hypothesis 1 predicted that low externally fo-cused satisfaction would translate into higher lev-els of individual absenteeism only when both themean and the dispersion levels of work-unit absen-teeism were high. As Figure 3 reveals, the pattern ofresults lends strong support to this hypothesis. Inpredicting individual absence frequency (post 6months) based on externally focused job satisfac-tion, the slope for high mean and dispersion levelsof work-unit absence frequency (pre 6 months) wasnegative and significant (� � �.34, p � .01). Incontrast, the slope (� � .05, n.s.) was not significantin cases of low mean and dispersion levels of work-unit absence frequency (pre 6 months). In both ofthe other conditions (low mean and high disper-sion/high mean and low dispersion), the slopeswere also not significant (� � .07, n.s./� � �.04,n.s., respectively). Finally, the mean and disper-sion of work-unit absence frequency (pre 6 months)jointly influenced the individual levels (intercepts)of absence frequency (post 6 months). Specifically,when the dispersion levels were low, the meanlevels of work-unit absence frequency were posi-tively related to individual absence frequency. Thisresult indicates that, regardless of external satisfac-tion, employees adapted their absence behavior tothe absence level of their work-unit in the presenceof uniformly low or high absenteeism (strong con-text: low absence dispersion).

Hypothesis 2 posited that internally focused jobsatisfaction would exhibit a strong negative rela-tionship with individual absenteeism only when

both the mean and dispersion levels of work-unitabsenteeism were low. Figure 4 shows that thepattern of the significant cross-level three-way in-teraction supports Hypothesis 2. In cases of lowmean and dispersion levels of work-unit absencefrequency (pre 6 months), the slope was negativeand significant (� � �.42, p � .01), whereas therelationship between internal satisfaction and indi-vidual absence frequency (post 6 months) failed toreach significance when either or both of the meanand dispersion levels of work-unit absence fre-quency (pre 6 months) were high (high mean andlow dispersion: � � .04, n.s.; low mean and highdispersion: � � �.04, n.s.; high mean and highdispersion: � � .06, n.s.) Again, individual absencefrequency (intercepts) was positively related tomean levels of work-unit absence frequency underconditions of low dispersion levels; that is, em-ployees tended to show absence levels similar tothe mean levels of their work unit when the workunit was characterized by uniformly high or lowabsence frequency. Only individuals who reportedlow internal satisfaction were frequently absent re-gardless of work-unit absence patterns. In conclu-sion, the data from Study 1 provide strong supportfor both of our hypotheses.

Supplementary analyses. Although our opera-tionalization of the mean and dispersion levels ofwork-unit absenteeism as indicators of the socialenvironment is consistent with conceptual devel-opments in previous studies on the cross-level ef-fects of social contexts (Hirst, van Knippenberg,Chen, & Sacramento, 2011; Liu et al., 2012; Mathieu& Kohler, 1990), scholars have noted that the sim-ple aggregation of individual variables to higher-level constructs necessarily includes the data of thefocal individual and, thus, may result in a biasedassessment of the context, which should only cap-ture the patterns of the other individuals in a par-ticular work unit (Eder & Eisenberger, 2008; Gel-latly & Allen, 2012). To cross-validate our findingsusing a measure of work-unit absenteeism that is

1.2: Work-Unit Level Variables

Variable M SD 1 2 3 4

1. Team size 6.54 2.44 —2. Unemployment rate 0.10 0.02 .10 —3. Mean levels of absence frequency (pre 6 months) 0.86 0.68 .27* �.11 —4. Dispersion levels of absence frequency (pre 6 months) 0.96 0.72 .04 .02 .24 —

Notes. (n � 63). Descriptive statistics of absence frequency represent non-transformed scores.

*p � .05; ** p � .01

2014 365Diestel, Wegge, and Schmidt

Page 14: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

not affected by the absence behavior of target em-ployees, we adopted the split-sample procedure,which allows for separation of focal employee datafrom the data of the other employees (i.e., contex-tual data) (Dollard, Tuckey, & Dormann, 2012).Specifically, to maintain statistical power at theindividual level (Mathieu, Aguinis, Culpepper, &Chen, 2012), we selected only work units with 5 ormore employees and then randomly defined cases

that were used to calculate the mean and disper-sion levels of absence frequency for each work unit,which were subsequently excluded from the focalindividual’s dataset. On the basis of a reduced sam-ple size (Nlevel 1 � 244; Nlevel 2 � 50), we reanalyzedboth of the cross-level, three-way interactions usingthe same specifications as described above. HLMestimations revealed a significant cross-level,three-way interaction effect between external satis-

TABLE 2Study 1 HLM Analyses Predicting Individual Absence Frequency—Post 6 Monthsa

Variable

Model 1 Model 2 Model 3

Coefficient SE Coefficient SE Coefficient SE

Intercept 0.64** 0.14 0.68** 0.11 0.67** 0.11Level 1 variablesAge �0.00 0.00 �0.00 0.00 �0.00 0.00Gender 0.01 0.10 �0.03 0.08 �0.02 0.08Organizational tenure �0.00 0.01 �0.00 0.00 �0.00 0.00Absence frequency (pre 6 months) �0.06 0.05 �0.06 0.04 �0.04 0.04Health disorders 0.17* 0.08 0.22** 0.07 0.23** 0.07Externally focused satisfaction �0.16** 0.05 �0.09** 0.02 �0.07* 0.02Internally focused satisfaction �0.12* 0.05 �0.12** 0.03 �0.09** 0.03Level 2 variablesTeam size �0.02 0.02 0.01 0.03 0.01 0.03Unemployment rate �1.61 2.06 �1.34 1.92 �1.43 1.93Absence frequency mean (pre 6 months) 0.23** 0.04 0.14* 0.06 0.15* 0.06Absence frequency dispersion (pre 6 months) �0.05 0.04 0.01 0.06 �0.00 0.06Absence frequency mean � Absence frequency

dispersion (pre 6 months)�0.16* 0.06 �0.12* 0.06

Cross-level interaction termsExternally focused satisfaction � Absence frequency

mean (pre 6 months)�0.11** 0.02 �0.12** 0.02

Externally focused satisfaction � Absence frequencydispersion (pre 6 months)

�0.11** 0.03 �0.07* 0.03

Externally focused satisfaction � Absence frequencymean � Absence frequency dispersion(pre 6 months)

�0.08** 0.02

Internally focused satisfaction � Absence frequencymean (pre 6 months)

0.17** 0.04 0.14** 0.04

Internally focused satisfaction � Absence frequency dispersion(pre 6 months)

0.07 0.03 0.10* 0.04

Internally focused satisfaction � Absence frequencymean � Absence frequency dispersion (pre 6 months)

�0.09** 0.03

R2Level 1

b 0.18 0.41 0.44R2

Level 2 0.35 0.41 0.42R2

Slope: externally focused satisfaction 0.00 0.74 0.88R2

Slope: internally focused satisfaction 0.00 0.58 0.78R2

Totalc 0.21 0.41 0.44

�2d 126.10** 172.66** 169.31**Model Deviance 820.66 716.72 710.70

Notes.a Employee n � 412; work-unit n � 63.b Level 1 variance includes both slope variances, when calculating R2

Level 1 (Mathieu et al., 2012: 957; Shin, Lee, Kim, & Bian, 2012).c R2

Total � R2Level 1 � (1 � ICC(1)) � R2

Level 2 � ICC(1) (Raudenbush & Bryk, 2002).d Chi-square values refer to level 2 variance (Raudenbush & Bryk, 2002).* p � .05; ** p � .01 two-tailed tests.

366 AprilAcademy of Management Journal

Page 15: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

faction and work-unit absence frequency on subse-quent individual absenteeism. The pattern of thisinteraction was similar to that observed in Figure 3.In addition, both mean absence frequency and itsinteraction with dispersion levels were signifi-cantly related to subsequent individual absentee-ism. Finally, mean levels of work-unit absencefrequency moderated the negative relationships be-tween both the satisfaction foci and individual ab-sence frequency, with signs corresponding to ourexpectations (Hypothesis 1 and Hypothesis 2). Al-though the three-way interaction with internal sat-isfaction was not significant, the signs of the pa-rameters suggested that the interaction was similarto that observed in Figure 4 and that the lack of

significance was due to reduced statistical power.Thus, the split-sample procedure indicates that thefocal employees’ absence data did not bias thecross-level effects of work-unit absenteeism whenthe measure included the absence levels of thetarget individuals.

Furthermore, for at least three additional reasons,we believe that our design is conceptually appro-priate for testing our hypotheses. First, in accor-dance with gestalt characteristics theory (Ariely &Carmon, 2000) and social information processingtheory (Salancik & Pfeffer, 1978), evidence fromresearch on social-contextual influences demon-strates that individuals in groups consider theirown attributes (behavior and attitudes) when inter-

FIGURE 3Study 1: Three-Way Interaction Effect of Mean and Dispersion Levels of Work-Unit Absence Frequency

and Externally Focused Job Satisfaction on Individual Absence Frequency

2014 367Diestel, Wegge, and Schmidt

Page 16: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

preting socially relevant and meaningful aspects oftheir environment (Johns, 2001; Xie & Johns, 2000).In particular, in developing a clear sense of thenormative expectations within their work unit, fo-cal employees recognize their own past behavioralpattern as an important gestalt (informational cue)that is viewed in conjunction with the behavioralpatterns of others (Liu et al., 2012). Thus, the ab-sence behavior of focal employees constitutes arelevant part of the social context as represented bythe mean and dispersion levels of work-unit absen-teeism. Second, both hypotheses assert that underconditions of low dispersion (i.e., uniform patternsof absence behavior) individual absence levels willeither increase with decreasing satisfaction (Hy-

pothesis 2) or will correspond to mean levels ofwork-unit absence (Hypothesis 1 and Hypothesis2). Thus, large differences in absenteeism betweenfocal employees and other work-unit memberswould create a condition that is not relevant to ourcore predictions. The same is true for conditionsof high mean and dispersion levels of work-unitabsenteeism (Hypothesis 1); specifically, a targetemployee’s own contribution to a high mean anddispersion of absenteeism within his or her workunit is relatively marginal given the absence dis-tribution that is required to create such a condi-tion. Finally, in predicting individual absentee-ism (post 6 months), we controlled for previousindividual absenteeism and thus partialled out

FIGURE 4Study 1: Three-Way Interaction Effect of Mean and Dispersion Levels of Work-Unit Absence Frequency

and Internally Focused Job Satisfaction on Individual Absence Frequency

368 AprilAcademy of Management Journal

Page 17: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

the potential influences of focal employees’ ab-sences on the moderating effect of work-unit ab-senteeism on the individual relationship betweensatisfaction levels and absenteeism (Mathieu &Kohler, 1990).

STUDY 2—METHODS

Participants and Procedures

The second study was carried out in a tax andrevenue organization of a federal state in Germanywith the purpose of analyzing the stress levels andwell-being of employees in the realm of adminis-trative work. After gaining approval from themanagement, all employees in the organizationwere asked to participate. A total of 441 employ-ees (response rate of 88%) voluntarily partici-pated in the survey. Questionnaires were distrib-uted at the workplace during working hours. Allparticipants were assured that their responseswould remain confidential and that their answerswould be matched with individual absence datathrough an individual code number that wasgiven solely to the researchers and remained intheir hands only.

Of the 441 respondents, we identified 350employees who were allocated to 53 work units(MSIZE � 6.23, range � 3–17) that remained intactthroughout the study. In identifying work units, wemade sure that the tasks performed (evaluating taxdeclarations) were similar across individuals andunits, and that each work unit was led by a singlesupervisor. After the questionnaire was adminis-tered, 20 participants left either their work units orthe entire organization. ANOVAs revealed no sig-nificant differences in study variables betweenthose who left and those who remained (p � .20).The remaining sample of 330 participants was69% female. The average age of the participantswas 34.6 years (SD � 8.6, range � 19 – 60), andthe mean organizational tenure was 14 years(SD � 9.2, range � 0 – 42).

Measures

Control variables. As in Study 1, we includedage, gender, organizational tenure, and health dis-orders (von Zerssen, 1976; � � .89) as control vari-ables at level 1 and unemployment rate (FederalStatistical Office of Germany, 2012) as a controlvariable at level 2 in our analyses.

Individual externally and internally focusedjob satisfaction. The assessment of both satisfac-tion foci was based on the same instrument as thatused in Study 1. Again, confirmatory factor analy-ses provided support for the distinctness of exter-nally (� � .83) and internally focused satisfaction(� � .76), and a two-factor model had a superior fit(�2 � 5.86, df � 4, p � .21, RMSEA � .04, 95% CI[.00�.10], CFI � 1.00, SRMR � .02) compared withthat of a one-factor model (�2 � 103.66, df � 5, p �.01, RMSEA � .25, 95% CI [.21�.29], CFI � .78,SRMR � .11). An ICC(1) of .16 revealed substantialvariation in internal satisfaction between the 53work units, whereas external satisfaction did notvary across the work units (ICC(1) � .01). Thus, theanalysis of between-group variance provided fur-ther support for our distinction between thetwo foci.

Individual and work-unit absenteeism. InStudy 2, we took absence frequency to be an indi-cator of voluntary absence (Sagie, 1998). Again, thisindex referred to short-term absences, defined asabsences of fewer than four consecutive workdays,and did not involve absences due to holidays orvacation days, maternity leave, military service,participation in training courses, or long-term ab-sences. As in Study 1, the absenteeism data wereobtained from archival records and included datafrom a period of 6 months before and 6 months afterthe administration of the questionnaires. Again, themanagement did not apply any official proceduresto control or to sanction short-term absenteeism,and the organization did not inform work-unitmanagers about absence rates. Due to deviationsfrom the normal distribution, absence frequencywas subjected to a square root transformation. Afterthe transformation procedure, skewness and kurto-sis did not exceed the threshold values summa-rized by Steel (2003).

An ICC(1) value of .18 suggests that there wasconsiderable variance in absence frequency (pre 6months) between the 53 work units. Comparedwith Study 1, the ICC(2) value (.57) was somewhatlower. Although our values are slightly below therecommended levels (Glick, 1985), they indicate ahigh enough level of reliability of the work-unitmean to allow the mean aggregation of the indices(see also Liao & Rupp, 2005; Schneider, White, &Paul, 1998). Again, the dispersion levels of work-unit absenteeism were operationalized as the stan-dard deviation of both of the absence variables (pre

2014 369Diestel, Wegge, and Schmidt

Page 18: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

6 months) within the work units.2 Bartlett’s testconfirmed that the work units significantly dif-fered with respect to variations in absence fre-quency (p � .01).

STUDY 2—RESULTS

Descriptive statistics for all the variables are pre-sented in Table 3. Age and tenure were highlycorrelated (r � .82; p � .01). To avoid the biasingeffects of multicollinearity, we excluded age fromour design because tenure contains more occupa-tionally relevant information (Ng & Feldman,2010). As in Study 1, the Study 2 sample allowedus to use HLM. We applied the same procedures formodel specification as in Study 1. Table 4 depictsthe results.

The estimation of the null model revealed anICC(1) of .15 for absence frequency (post 6 months).The slopes of the relationships between both of thesatisfaction foci and absence frequency (post 6months) significantly varied across the work units

(externally focused satisfaction: U1 � .08, p � .01;internally focused satisfaction: U1 � .04, p � .01).

Main effects at level 1 and 2. Table 4 (Model 1)depicts the estimations of the linear relationshipsof level 1 and level 2 variables with individualabsence frequency (post 6 months). As in Study 1,only health disorders explained variance in indi-vidual absenteeism (� � .22, p � .01). Both satis-faction foci were negatively and significantly re-lated to absence frequency (external: � � �.16, p �.01; internal: � � �.13, p � .01). Again, the meanlevels of work-unit absence frequency (pre 6months) predicted subsequent individual absencefrequency (post 6 months) (� � .24, p � .01).

Cross-level, three-way interactions (Hypothe-ses 1 and 2). As Table 4 reveals, the mean anddispersion levels of absence frequency (pre 6months) jointly moderated the negative individualrelationship between externally focused satisfac-tion and absence frequency (post 6 months) (� ��.10, p � .01). However, the cross-level, three-wayinteraction between work-unit absenteeism vari-ables and internally focused satisfaction was non-significant. In sum, the cross-level interactions inModel 3 explained an additional 1% of the vari-ance (and, in combination with the conditionaltwo-way interactions, an additional 21% of thevariance) in individual absence frequency (post 6months). To interpret the significant three-way in-teraction, we applied simple slope analyses andplotted the interaction pattern (Figure 5).

According to Hypothesis 1, we expected thatexternally focused satisfaction would be moststrongly related to individual absenteeism whenboth the mean and the dispersion levels of work-unit absenteeism were high. The regression lines in

2 As in Study 1 (see footnote 2 above), we tested thevalidity of the absence aggregates. Again, we chose workunits with response rates of at least 90% (max. one mem-ber missing: Nlevel 2 � 28; 26 work units had a responserate of 100%) and created response rates of approxi-mately 80% and 70% for these work units. Correlationsbetween the lower and higher response rate values (meanand dispersion levels of work-unit absence frequency)demonstrated that study dropouts would not consider-ably change the values of the absence aggregates in thesample of tax and revenue offices (approx. 80% re-sponse rate: rmean levels � .98; p � .01/rdispersion levels �.96; p � .01; approx. 70% response rate: rmean levels � .96;p � .001/rdispersion levels � .94; p � .01).

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

3.1: Individual Level Variables

Variable M SD 1 2 3 4 5 6 7 8

1. Age 34.55 8.58 —2. Gender (1 � female; and 2 � male) 1.31 0.46 .21** —3. Tenure 14.04 9.19 .82** .17** —4. Health disorders 0.69 0.43 �.04 �.07 �.05 —5. Absence frequency (pre 6 months) 1.64 1.58 �.11* �.12* �.09 .18** —6. Externally focused satisfaction 3.01 1.05 .25** �.09 .25** �.17** �.09 —7. Internally focused satisfaction 5.17 1.12 .01 �.02 .03 �.20** �.14* .32** —8. Absence frequency (post 6 months) 1.35 1.21 .00 �.05 �.07 .26** .20** �.27** �.23** —

Note. (n � 330). Descriptive statistics of absence frequency represent non-transformed scores.* p � .05; **p � .01.

370 AprilAcademy of Management Journal

Page 19: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

Figure 5 correspond to the hypothesized interac-tion pattern. In the cases of high mean levels anddispersions of absence frequency (pre 6 months)within work units, the slope was negative and sig-nificant (� � �.43, p � .01). In contrast, the slopefor low mean levels and dispersions of absencefrequency was not significant (� � .06, n.s.) Wheneither the mean or the dispersion was high, theestimated relationships between external satisfac-tion and individual absence frequency (post 6months) were also not significant (� � �.02, n.s./� � .04, n.s., respectively). As in Study 1, regard-less of external satisfaction, the mean levels of ab-sence frequency were related to individual absencefrequency when the dispersion level was lower.Thus, the results suggest that employees aligned theirabsence behavior with the mean work-unit absentee-ism in the presence of uniform absence patterns. Insum, Hypothesis 1 was strongly supported by thedata of the tax and revenue organization.

In Models 2 and 3, the two-way interaction be-tween the mean absence frequency and internallyfocused satisfaction was significant (� � .10, p �.01). This result indicates that the relationship be-tween internal satisfaction and individual absen-teeism was moderated by the social context interms of work-unit absenteeism. We visualized theinteraction and analyzed the significance of theslopes. Figure 6 shows that the slope was negativeand significant when the mean levels of work-unitabsence frequency (pre 6 months) were low (� ��.20, p � .01), whereas the slope was non-signifi-cant when the mean absence frequency was high(� � .01, n.s.). In cases of high absence meanswithin work units, the intercept for the relationshipwas high. Thus, internally satisfied employeesshowed high levels of absenteeism when the ab-sence mean of their work unit was also high. Al-though the dispersion levels of work-unit absencefrequency did not exert moderating effects (neitheralone nor in combination with mean levels), andthus Hypothesis 2 was not fully supported by the

data from Study 2, the present results provide somesupport for our assertion that internally focusedsatisfaction is negatively related to individual ab-sence when work-unit absence is low.

Supplementary analyses. As in Study 1, em-ploying the split-sample procedure, we reanalyzedthe data from the tax and revenue offices to testwhether the absence behavior of focal employeesmay bias their assessment of the absenteeism con-text within their work unit. After reducing the da-taset due to the small sizes of several work units,the sample comprised 167 individuals and 39 workunits. HLM estimations revealed a significantcross-level relationship between the mean levels ofwork-unit absence frequency (pre 6 months) andindividual absence frequency (post 6 months).Moreover, we found that the negative individualrelationships between satisfaction foci and absen-teeism were moderated by the mean and dispersionlevels (dispersion: only for external satisfaction) ofwork-unit absenteeism and that the signs corre-sponded to our expectations. While we did not finda three-way interaction (most likely due to reducedstatistical power), the pattern of results suggeststhat the operationalization of work-unit absentee-ism and the analysis of its interactive effects are notbiased by the focal employees’ absence data.

DISCUSSION

The present study drew on recent research ondifferent job satisfaction foci and work-unit absen-teeism (Dineen et al., 2007; Hausknecht et al., 2008)and was designed to disentangle the individualrelationships of externally and internally focusedjob satisfaction with absenteeism. To this end, wedelineated and tested a cross-level model that ex-patiates upon the social-contextual boundary con-ditions of these relationships. Inspired by severaltheories on social influences within groups, weconceptualized the absence context through thecombination of the mean and dispersion levels of

3.2: Work-Unit Level Variables

Variable M SD 1 2 3 4

1. Team size 6.23 2.78 —2. Unemployment rate 0.10 0.03 �.18 —3. Mean levels of absence frequency (pre 6 months) 1.54 0.90 .14 �.09 —4. Dispersion levels of absence frequency (pre 6 months) 1.09 0.67 .25 .00 .06 —

Note. (n � 53). Descriptive statistics of absence frequency represent non-transformed scores.* p � .05; **p � .01

2014 371Diestel, Wegge, and Schmidt

Page 20: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

work-unit absenteeism and analyzed differentcross-level, three-way interaction effects betweenboth of these facets of work-unit absenteeism andboth of the job satisfaction foci on individual ab-senteeism. On the basis of two multi-level datasetsfrom different occupational contexts, our resultsindicated that an individual’s level of externallyfocused job satisfaction predicted subsequent ab-senteeism only when both the mean and the dis-persion of work-unit absenteeism were high (Hy-pothesis 1). Conversely, in the presence of a lowlevel of dispersion of work-unit absence, individ-

ual absenteeism was predicted by the mean levelsof work-unit absence regardless of external satisfac-tion. Moreover, the mean and dispersion levels ofwork-unit absenteeism jointly moderated the nega-tive relationship between internally focused satisfac-tion and individual absenteeism among participantsin Study 1. Specifically, in cases of low absencemeans and dispersions within work units, the rela-tionship between internal satisfaction and individualabsenteeism was strongest (Hypothesis 2). Con-versely, individual absenteeism was high for all lev-els of internal satisfaction when the absence mean

TABLE 4Study 2 HLM Analyses Predicting Individual Absence Frequency—Post 6 Monthsa

Variable

Model 1 Model 2 Model 3

Coefficient SE Coefficient SE Coefficient SE

Intercept 1.02** 0.08 0.99** 0.08 0.99** 0.08Level 1 variablesGender �0.05 0.06 �0.03 0.05 �0.03 0.05Organizational tenure �0.00 0.00 0.00 0.00 0.00 0.00Absence frequency (pre 6 months) �0.08 0.05 �0.05 0.04 �0.04 0.04Health disorders 0.22** 0.08 0.18* 0.07 0.16* 0.07Externally focused satisfaction �0.16** 0.06 �0.09* 0.03 �0.09* 0.03Internally focused satisfaction �0.13** 0.04 �0.11** 0.03 �0.10** 0.03Level 2 variablesTeam size 0.00 0.01 0.01 0.01 0.01 0.01Unemployment rate �1.55 1.61 �1.80 1.59 �1.96 1.56Absence frequency mean (pre 6 months) 0.24** 0.04 0.23** 0.04 0.22** 0.04Absence frequency dispersion (pre 6 months) �0.04 0.03 �0.01 0.04 0.00 0.04Absence frequency mean � Absence

frequency dispersion (pre 6 months)�0.06 0.04 �0.08 0.04

Cross-level interaction termsExternally focused satisfaction � Absence

frequency mean (pre 6 months)�0.14** 0.03 �0.14** 0.03

Externally focused satisfaction � Absencefrequency dispersion (pre 6 months)

�0.18** 0.03 �0.11** 0.04

Externally focused satisfaction � Absence frequencymean � Absence frequency dispersion (pre 6 months)

�0.10** 0.03

Internally focused satisfaction � Absence frequencymean (pre 6 months)

0.10** 0.04 0.10** 0.03

Internally focused satisfaction � Absence frequency dispersion(pre 6 months)

�0.01 0.03 0.03 0.04

Internally focused satisfaction � Absence frequencymean � Absence frequency dispersion (pre 6 months)

�0.06 0.04

R2Level 1

b 0.19 0.42 0.43R2

Level 2 0.51 0.57 0.57R2

Slope: externally focused satisfaction 0.00 0.84 0.91R2

Slope: internally focused satisfaction 0.00 0.67 0.74R2

Totalc 0.24 0.44 0.45

�2d 71.57 108.46** 107.18**Model Deviance 578.80 504.57 506.20

Notes.a Employee n � 330; work unit n � 63.b Level 1 variance includes both slope variances, when calculating R2

Level 1 (Mathieu et al., 2012: 957; Shin et al., 2012).c R2

Total � R2Level 1 � (1 � ICC(1)) � R2

Level 2 � ICC(1) (Raudenbush & Bryk, 2002).d Chi-square values refer to level 2 variance (Raudenbush & Bryk, 2002).* p � .05; ** p � .01 two-tailed tests.

372 AprilAcademy of Management Journal

Page 21: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

was high and the absence dispersion was low. InStudy 2, we found a similar pattern: specifically, thenegative relationship between internal satisfactionand individual absenteeism was attenuated as a func-tion of the mean level of work-unit absenteeism.

Theoretical Implications

Since Hackett’s (1989) meta-analytic examina-tion of the individual satisfaction–absenteeism re-lationship, scholars have repeatedly searched forboundary conditions, primarily from a mono-levelperspective, that would explain the broad range ofcorrelations that have been found between satisfac-tion and absenteeism. In light of both the inconsis-tent pattern of results and the low predictive power

that have been reported across studies (Harrison etal., 2006; Sagie, 1998; Somers, 1995; Wegge et al.,2007), this enterprise has been of little theoreticalvalue. Provided that absenteeism is a meaningfulevent that draws the attention of other group orwork-unit members (Hausknecht et al., 2008; Johns& Nicholson, 1982), the social and normative ex-pectations within the immediate environment con-stitute relevant boundary conditions that may havea higher explanatory power. To capture such so-cial-contextual influences, we adopted the conceptof situational strength (Cooper & Withey, 2009;Mischel, 1977), which encompasses two contextualcues: the degree and the consistency of meaningfulorganizational behavior (namely, absenteeism). Inother words, the combination of both pieces of

FIGURE 5Study 2: Three-Way Interaction Effect of Mean and Dispersion Levels of Work-Unit Absence Frequency

and Externally Focused Job Satisfaction on Individual Absence Frequency

2014 373Diestel, Wegge, and Schmidt

Page 22: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

information (mean and dispersion) reflects thestrength of the absence context within a work unitand provides behavioral guidance for employees inaligning their behavior with the social and norma-tive expectations of their work unit. This line ofreasoning calls for a multi-level perspective on theindividual satisfaction–absenteeism relationship,and is further supported by our empirical findingthat only under specific contextual conditions dosatisfaction levels predict individual absenteeism.Compared with the rather moderate correlationsranging from �.21 to �.23 found in a previousmeta-analysis (Hackett, 1989), our simple slope re-sults indicate a more convincing degree of predic-tive power. As a result, we were able to observe theinterplay of contextual factors that drive the effect ofindividual job satisfaction on absenteeism. In the fol-lowing sections, we will further elaborate on the im-plications of our findings for theories on individualsatisfaction–absence relationships, absenteeism in or-ganizations, and social information processing.

Individual satisfaction–absenteeism relation-ships. Both of the different interaction patterns thatwe found enrich our knowledge about the nature of

the satisfaction–absenteeism relationship. Whenthe context (the work unit) is ambiguous andunclear with regard to the importance and theoccurrence of absenteeism, dissatisfaction with theexternal entity (e.g., job conditions, career oppor-tunities, and the organization itself) results in indi-vidual absenteeism. This finding is consistent withequity theory (Adams, 1965) and social exchangetheory (Thibaut & Kelley, 1959), which contendthat dissatisfied employees reduce their invest-ments when they perceive an unbalanced relation-ship with their organization (Harrison et al., 2006).Specifically, in the case of a weak context, themeaning of absenteeism is highly influenced by anemployee’s evaluation of his or her job conditions.In contrast, as a close inspection of the interceptsshowed, employees adapt their absence pattern tothe work-unit absence level (irrespective of theirown evaluation) when the work unit demonstratesuniformly high or low absence levels. In otherwords, in the case of a strong context with consis-tent absence patterns, the importance and occur-rence of individual absenteeism are determined by

FIGURE 6Study 2: Two-way Interaction Effect of Mean Levels of Work-Unit Absence Frequency and Internally

Focused Job Satisfaction on Individual Absence Frequency

374 AprilAcademy of Management Journal

Page 23: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

the social environment rather than by externallyfocused job satisfaction.

However, the cross-level interactive effect of in-ternally focused satisfaction and work-unit absen-teeism on subsequent individual absenteeism indi-cates that satisfaction with the work unit can beassociated with high individual absenteeism, par-ticularly when work-unit colleagues are also fre-quently absent. This inverse moderator effect ofwork-unit absenteeism is consistent with the theo-retical prediction that a high level of perceivedsocial integration motivates employees to alignwith their social environment (Fredrickson, 2001;Schneider, 1987). Although it is seemingly para-doxical that employees who are satisfied with theirwork unit should “avoid” their pleasant environ-ment, it should be noted that our operationalizationof absenteeism only included short-term absencesand that this effect may reflect a shared under-standing of work-related behavior (as indicated byhigh work-unit absenteeism) that is not beneficialto the organization (Goodman, Ravlin, & Schminke,1987). Similar effects have been reported in studieson cohesion, which, in combination with low per-formance norms, can lead to low performance (e.g.,Gammage, Carron, & Estabrooks, 2001). In this case,absenteeism may not be a form of avoidance orwithdrawal behavior; instead, we suggest that ab-senteeism is driven by a high degree of attendanceflexibility within a work unit or by a consensualagreement that justifies short-term absences. More-over, on the basis of social information processingtheory, Gellatly and Allen (2012: 107) recently ar-gued that in cases of high group absence, employ-ees may infer that “such behavior [is] both expectedand accepted.” From a social exchange perspective(Blau, 1964), this argument implies that behavioralalignment is a form of compliance with a groupnorm that is exchanged for social approval fromothers. As our results indicate, such processes aremost likely to occur when employees have a posi-tive attitude toward their work unit, which is re-flected by high internally focused satisfaction.

Absenteeism in organizations. From a more gen-eral perspective, the conclusions drawn from bothinteraction patterns have encouraged us to developa context-based theory of absence (see Figure 2)that elaborates the conditions under which individ-ual absence patterns reflect behavioral manifesta-tions of attitudinal differences or shared under-standings of absenteeism within organizationalstructures, such as work units, groups, or wholeorganizations (Gellatly & Allen, 2012; Rentsch &

Steel, 2003). According to this theory, our resultssuggest that the mean level of work-unit absentee-ism is only a meaningful contextual cue and istherefore predictive of individual absenteeismwhen the dispersion level of work-unit absentee-ism is low. In other words, consistent with thenotion of situational strength (Cooper & Withey,2009), social influences on individual absenteeismand their relationships with attitudes can only beunderstood by conceptualizing and examining ab-sence levels in combination with the consistency ofabsence patterns within social structures. Cooperand Withey (2009: 70) explicitly called for empiri-cal studies that include variations on situationalstrength and are devoted to testing its moderatingeffects on the relationships between individual at-tributes and “theoretically relevant behavior.” Inthis study, we have demonstrated that the conceptof situational strength applies to absenteeism inorganizations. Going beyond absenteeism, our re-search invites scholars to thoroughly consider sit-uational strength when building theories aboutother forms of meaningful organizational behavior(such as organizational citizenship behavior) thatare often embedded in social contexts and arethus observed, evaluated, and adopted by others(Ehrhart & Naumann, 2004).

Social information processing. The present re-sults also enhance our understanding of how em-ployees in organizations translate relevant andmeaningful information from their social environ-ment into behavior (Salancik & Pfeffer, 1978). Al-though Thomas and Griffin (1989: 71) have sug-gested moderating effects of social cues andcontextual information on attitude–behavior rela-tionships, the relevant factors and mechanisms in-fluencing the interactions between contexts andattitudes have remained largely unclear to date. Asrevealed by both of the different cross-level inter-actions we found, one relevant factor is the func-tion of an attitude, which influences the ways inwhich employees construe social-contextual infor-mation and respond to attitudinal targets. Specifi-cally, when evaluating instrumental aspects of a job(externally focused satisfaction), the immediate en-vironment conditions the attitudinal response (in-dividual absenteeism) in cases of clear and consis-tent contextual cues (low mean and/or lowdispersion levels of work-unit absenteeism). Incontrast, when attitudes (e.g., toward the workunit) have a social-adjustive function and are pos-itive (internally focused satisfaction), employeesadapt their attitudinal responses (individual absen-

2014 375Diestel, Wegge, and Schmidt

Page 24: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

teeism) to the behavioral patterns (work-unit absen-teeism) they observe within their social environ-ment. Thus, the distinction between differentattitudinal functions, as represented by differentattitudinal foci, has turned out to be useful in ex-plaining the interaction effect of social-contextualfactors and individual attitudes on organizationalbehavior. Given that previous studies have docu-mented different effects of internal and externalfoci of job attitudes on absenteeism and job perfor-mance (Becker, Billings, Eveleth, & Gilbert, 1996;Dineen et al., 2007; Siders et al., 2001), social infor-mation processing theory may benefit from distin-guishing between the different functions of job atti-tudes when deriving predictions regarding social-contextual influences on organizational behaviors,such as turnover (Liu et al., 2012), creativity (Hirst etal., 2011), and job performance (Liao & Rupp, 2005).

Limitations and Avenues for Future Research

Our research is subject to several limitations thatsuggest avenues for future research. First, althoughour multi-level analyses drew on longitudinal data,our research design was, nonetheless, correlationaland did not permit strong causal inferences. Giventhe dynamic nature of job satisfaction (Chen, Ploy-hart, Thomas, Anderson, & Bliese, 2011), our predic-tions should be further tested in designs in whichchanges in absenteeism and satisfaction over time aremodeled (e.g., Liu et al., 2012). However, in view ofour control of temporal changes in individual absen-teeism and that work-unit absenteeism is relativelystable even after controlling for dynamic effects(Hausknecht et al., 2008), we assert that our designprovides valuable insights into the functional role ofwork-unit absenteeism in the processes determiningthe effect of job satisfaction on individual absence.

Second, our conceptualization and operation-alization of job satisfaction only refer to five aspectsof jobs, whereas other relevant aspects, such as tasktypes or payment levels, were not included. Forexample, given that the evaluation of tasks is sub-ject to specific requirements (e.g., overcoming in-ner resistance in cases of unattractive tasks) thatrelate to absenteeism (Diestel & Schmidt, 2011),social-contextual factors can be thought to play apivotal role in task satisfaction. In addition, al-though our definition of attitudinal targets, whichare specific to instrumental and social-adjustiveattitudinal functions, draws on past research (e.g.,Dineen et al., 2007), a more explicit distinctionbetween different relevant attitudinal components

(such as beliefs, global evaluations, and affectiveexperiences; Weiss, 2002), which reflect attitudinalfunctions, would enhance scholarly understandingof how different foci of satisfaction (or job attitudesin general) relate to organizational behavior. Inlight of the different ICC(1) values and the differentmoderating effects of the absence context for bothsatisfaction foci found in our study, a theoreticallygrounded definition and a measure of satisfactioncomponents are imperative for future research(Schleicher et al., 2004).

Third, the finding that both of the different inter-actions are contingent upon satisfaction focistrongly suggests that the relationships betweenboth of the satisfaction foci and absenteeism aredriven by different psychological mechanisms. Inother words, mediating processes (e.g., calculativejudgment or social identification) may explain thedifferent interaction effects that were found in ourstudies. However, while we did not measure or testsuch mediating effects, we still argue that most of thetheoretically relevant information (e.g., different as-pects of a job) relies on the assessment and concep-tualization of both satisfaction foci, and that our pre-dictions do not require assumptions about mediatingprocesses. Because our predictions are grounded inthe concepts of attitude function and situationalstrength, analyses of mediating effects would provideonly a marginal improvement to our understanding ofthe interaction effects of the social context and thedifferent satisfaction foci on absenteeism. Neverthe-less, theories of contextual effects on organizationalbehavior (e.g., absenteeism) would still benefit fromintegrating and testing mediating effects.

Fourth, given the broad spectrum of variablesthat influence satisfaction and absenteeism, wecannot fully rule out the possibility of unmeasuredconfounding effects. Although we controlled forrelevant demographic variables (Wegge et al.,2007), local unemployment rate as an indicator ofeconomic welfare (Hausknecht et al., 2008), andsomatic complaints as predictors of involuntary ab-sences due to illness (Darr & Johns, 2008), othervariables may also play a pivotal role in absentee-ism, such as job performance, sanctioned absence,or specific working conditions. For example, thereasons for colleagues’ voluntarily or involuntarilyabsences (“taking a sickie” or taking care of a familymember) may influence the perceived meaning ofor justification for absence. In addition, our ab-sence measure does not explicitly distinguish be-tween excused and unexcused or voluntary andinvoluntary cases; specifically, although our mea-

376 AprilAcademy of Management Journal

Page 25: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

sure only included absences that were short term(Hausknecht et al., 2008), based on the absencepolicies of German organizations, absences due tosickness or personal issues (e.g., work–life balance)in addition to attitudinal reasons may have beencounted in our absence data. However, given thatboth of the satisfaction foci were substantially re-lated to our absence measure, that our findingsgeneralized across different organizational and oc-cupational contexts, and that we controlled forother relevant factors, we assert that the presentresults are somewhat protected against confound-ing effects. Nevertheless, future research shoulddifferentiate between voluntary/involuntary andexcused/unexcused absences (Sagie, 1998) andconsider further relevant control variables (Steel,Rentsch, & van Scotter, 2007).

Finally, in light of the variety of unit-level ante-cedents that prior research has found to predictwork-unit absenteeism (Dineen et al., 2007;Hausknecht et al., 2008; Xie & Johns, 2000), thespecific manifestation of the social (absence) con-text (configurations of high/low mean and disper-sion of work-unit absenteeism) can be consideredas a symptom of contextual influences (such asleadership, labor market conditions, or group cli-mate or cohesion). Indeed, we did not explicitlymeasure the unit-level mechanism that driveswork-unit absenteeism. Provided that prior re-search has already documented considerable evi-dence on main and interaction effects of unit-levelvariables on work-unit absenteeism (e.g., Dineen etal., 2007), our main purpose was to examine thefunctional role of work-unit absenteeism as a mean-ingful contextual construct (Mason & Griffin, 2003)in the relationship between individual job satisfac-tion and individual absenteeism. Thus, our find-ings broaden the theoretical understanding of con-text and shed light on the psychological function ofpatterns of meaningful behavior at higher organi-zational levels. Research on situational strengthsuggests that behavioral patterns within the socialenvironment typically have stronger impacts thanshared attitudes, communication, or other forms ofcontextual stimuli on individual behavior (see Coo-per & Withey, 2009; Johns, 2006; Mischel, 1977).The level and dispersion of meaningful behaviors(such as absenteeism) within social structures consti-tute relevant forms or manifestations of discrete con-texts that moderate individual satisfaction–absentee-ism relationships, even after controlling for economicand individual variables. In conclusion, our researchcalls for conceptualizations of contexts that integrate

behavioral patterns at higher structural levels (otherthan climate or team composition), especially whenexplaining the functional relationships between jobattitudes and individual behavior.

Managerial Implications

The present findings point to two factors that re-duce employee absenteeism either directly orthrough moderating effects: individual internally fo-cused satisfaction and work-unit absenteeism. On theone hand, irrespective of the level of externally fo-cused satisfaction, the lowest levels of employee ab-senteeism are most likely to occur when the meanand dispersion levels of work-unit absenteeism areboth low. On the other hand, the combination of highinternal satisfaction and low work-unit absenteeismis also associated with low individual absenteeism.Thus, HR practices should aim at simultaneously in-creasing individual internally focused satisfactionwhile reducing work-unit absenteeism.

The findings described by Dineen et al. (2007)have already highlighted the influence of cohesionand social integration on absenteeism. Based ontheir study, these researchers recommended thatmanagers enhance team satisfaction by focusing onensuring appropriate team composition. Likewise,our results indicate that managers should fostersocial integration within teams and social relation-ships among teammates and supervisors who pro-vide support—because, irrespective of the contex-tual conditions of work-unit absenteeism, lowindividual levels of internally focused job satisfac-tion result in higher levels of individual absentee-ism. However, as our results clearly show, highinternally focused satisfaction can also be associ-ated with high employee absenteeism when work-unit absenteeism is high. Thus, managers shoulddevelop and adopt practices that target the grouplevel and aim to control work-unit absenteeismrather than individual behavior directly. For exam-ple, clear supervisory instructions, formalized at-tendance procedures, feedback systems, or unit-level incentives could decrease the mean anddispersion levels of work-unit absenteeism. We ar-gue for controlling work-unit absenteeism, and weemphasize that such interventions or practices aredifferent from reducing absenteeism at the individuallevel. Reducing individual absenteeism often impliesfinancial incentives for each employee or individualappraisals with feedback (Hausknecht et al., 2008).These practices are often extensive, and their effects

2014 377Diestel, Wegge, and Schmidt

Page 26: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

on individual absenteeism are largely unknown atpresent (Camden, Price, & Ludwig, 2011).

Finally, the findings of Hausknecht et al. (2008)emphasize the influential interplay between satis-faction and commitment at the work-unit level.When both of these variables are high, low levels ofwork-unit absenteeism are likely. Thus, attempts tocontrol work-unit absenteeism that include in-creasing shared emotional attachment to the organ-ization (through, e.g., cultural development pro-grams and team-building programs that foster theshared sense of enjoyment that resides within workunits) appear to be promising.

REFERENCES

Abelson, R. P., & Prentice, D. A. 1989. Beliefs as posses-sions: A functional perspective. In A. R. Pratkanis, S. J.Breckler, & A. G. Greenwald (Eds.), Attitude structureand function: 361–381. Hillsdale, NJ: Lawrence.

Adams, S. J. 1965. Inequity in social exchange. In L.Derkowitz (Ed.), Advances in experimental socialpsychology, vol. 2: 267–299. New York: AcademicPress.

Aiken, L. S., & West, S. G. 1991. Multiple regression:Testing and interpreting interactions. ThousandOaks, CA: Sage.

Ajzen, I., & Fishbein, M. 1977. Attitude–behavior rela-tions: A theoretical analysis and review of empiricalresearch. Psychological Bulletin, 84: 888–918.

Ariely, D., & Carmon, Z. 2000. Gestalt characteristics ofexperiences: The defining features of summarizedevents. Journal of Behavioral Decision Making, 13:191–201.

Asch, S. E. 1956. Studies of independence and confor-mity: I. A minority of one against a unanimous ma-jority. Psychological Monographs, 70: 1–70.

Asch, S. E. 1966. Opinions and social pressure. In A. P.Hare, E. F. Borgatta & R. F. Bales (Eds.), Smallgroups: Studies in social interaction: 318–324.New York: Alfred A. Knopf.

Bamberger, P., & Biron, M. 2007. Group norms and ex-cessive absenteeism: The role of peer referent others.Organizational Behavior and Human DecisionProcesses, 103: 179–196.

Becker, T. 1992. Foci and bases of commitment: Are theydistinctions worth making? Academy of Manage-ment Journal, 35: 232–244.

Becker, T., Billings, R., Eveleth, D., & Gilbert, N. 1996.Foci and bases of employee commitment: Implica-tions for job performance. Academy of Manage-ment Journal, 39: 464–482.

Biron, M., & Bamberger, P. 2012. Aversive workplaceconditions and absenteeism: Taking referent groupnorms and supervisor support into account. Journalof Applied Psychology, 97: 901–912.

Blau, P. M. 1964. Exchange and power in social life.New York: Wiley.

Camden, M. C., Price, V. A., & Ludwig, T. D. 2011. Re-ducing absenteeism and rescheduling among grocerystore employees with point-contingent rewards.Journal of Organizational Behavior Management,31: 140–149.

Cammann, C., Fichman, M., Jenkins, G. D., & Klesh, J. R.1983. Assessing the attitudes and perceptions of organ-izational members. In S. E. Seashore, E. E. Lawler, P. H.Mirvis & C. Cammann (Eds.), Assessing organization-al change: A guide to methods, measures, andpractices: 71–138. New York: John Wiley.

Chadwick-Jones, J. K., Nicholson, N., & Brown, C. 1982. So-cial psychology of absenteeism. New York: Praeger.

Chan, D. 1998. Functional relations among constructs inthe same content domain at different levels of anal-ysis: A typology of compositional models. Journal ofApplied Psychology, 83: 234–236.

Chen, G., Ployhart, R. E., Thomas, H. C., Anderson, N., &Bliese, P. D. 2011. The power of momentum: A newmodel of dynamic relationships between job satis-faction change and turnover intentions. Academy ofManagement Journal, 54: 159–181.

Clegg, C. W. 1983. Psychology of employee lateness, ab-sence, and turnover: A methodological critique andan empirical study. Journal of Applied Psychology,68: 88–101.

Cole, M. S., Bedeian, A. G., Hirschfeld, R. R., & Vogel, B.2011. Dispersion-composition models in multilevelresearch: A data-analytic framework. Organization-al Research Methods, 14: 718–734.

Cooper, W. H., & Withey, M. J. 2009. The strong situationhypothesis. Personality and Social Psychology Re-view, 13: 62–72.

Darr, W., & Johns, G. 2008. Work strain, health, andabsenteeism: A meta-analysis. Journal of Occupa-tional Health Psychology, 13: 293–318.

Diestel, S., & Schmidt, K.-H. 2011. Costs of simultaneouscoping with emotional dissonance and self-controldemands at work: Results from two German samples.Journal of Applied Psychology, 96: 643–653.

Dineen, B. R., Noe, R. A., Shaw, J. D., Duffy, M. K., &Wiethoff, C. 2007. Level and dispersion of satisfac-tion in teams: Using foci and social context to ex-plain the satisfaction–absenteeism relationship.Academy of Management Journal, 50: 623–643.

Dollard, M. F., Tuckey, M. R., & Dormann, C. 2012.Psychosocial safety climate moderates the job de-

378 AprilAcademy of Management Journal

Page 27: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

mand–resource interaction in predicting workgroupdistress. Accident; Analysis and Prevention, 45:694–704.

Duffy, M. K., Scott, K. L., Shaw, J. D., Tepper, B. J., &Aquino, K. 2012. A social context model of envy andsocial undermining. Academy of ManagementJournal, 55: 643–666.

Eder, P., & Eisenberger, R. 2008. Perceived organizationalsupport: Reducing the negative influence of co-worker withdrawal behavior. Journal of Manage-ment, 34: 55–68.

Ehrhart, M. G., & Naumann, S. E. 2004. Organizationalcitizenship behavior in work groups: A group normsapproach. Journal of Applied Psychology, 89: 960–974.

Erdogan, B., & Enders, J. 2007. Support from the top:Supervisors’ perceived organizational support as amoderator of leader–member exchange to satisfac-tion and performance relationships. Journal of Ap-plied Psychology, 92: 321–330.

Falomir-Pichastor, J. M., Gabarrot, F., & Mugny, G. 2009.Conformity and identity threat: The role of ingroupidentification. Swiss Journal of Psychology, 68: 79–87.

Federal Statistical Office of Germany. 2012. LaborMarket Reports. https://www.destatis.de/EN/FactsFigures/NationalEconomyEnvironment/LabourMarket/Unemployment/Unemployment.html. Accessed onFebruary 05, 2013.

Felps, W., Mitchell, T. R., Hekman, D. R., Lee, T. W.,Holtom, B. C., & Harman, W. S. 2009. Turnover con-tagion: How coworkers’ job embeddedness and jobsearch behaviors influence quitting. Academy ofManagement Journal, 52: 545–561.

Fredrickson, B. L. 2001. The role of positive emotions inpositive psychology: The broaden-and-build theoryof positive emotions. American Psychologist, 56:218–226.

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

Gellatly, I. R., & Allen, N. J. 2012. Group mate absence, dis-similarity, and individual absence: Another look at“monkey see, monkey do”. European Journal of Workand Organizational Psychology, 21: 106–124.

George, J. M. 1989. Mood and absence. Journal of Ap-plied Psychology, 74: 317–324.

Glick, W. H. 1985. Conceptualizing and measuring organ-izational and psychological climate: Pitfalls in mul-tilevel research. Academy of Management Review,10: 601–616.

Goodman, P. S., Ravlin, E. C., & Schminke, M. 1987.Understanding groups in organizations. In B. M.Staw & L. L. Cummings (Eds.), Research in organi-zational behavior, 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.

Hammer, T. H., & Landau, J. 1981. Methodological issuesin the use of absence data. Journal of Applied Psy-chology, 66: 574–581.

Harrison, D. A., & Klein, K. J. 2007. What’s the differ-ence? Diversity constructs as separation, variety, ordisparity in organizations. Academy of Manage-ment Review, 32: 1199–1228.

Harrison, D. A., & Martocchio, J. J. 1998. Time for absen-teeism: A 20-year review of origins, offshoots, andoutcomes. Journal of Management, 24: 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, 49:305–325.

Hausknecht, J. P., Hiller, N. J., & Vance, R. J. 2008. Work-unit absenteeism: Effects of satisfaction, commit-ment, labor market conditions and time. Academyof Management Journal, 51: 1223–1245.

Hekman, D. R., Steensma, H. K., Bigley, G. A., & Here-ford, J. F. 2009. Effects of organizational and profes-sional identification on the relationship between ad-ministrators’ social influence and professionalemployees’ adoption of new work behavior. Journalof Applied Psychology, 94: 1325–1335.

Hirst, G., van Knippenberg, D., Chen, C. H., & Sacra-mento, C. A. 2011. How does bureaucracy impact onindividual creativity? A cross-level investigation ofteam contextual influences on goal orientation-cre-ativity relationships. Academy of ManagementJournal, 54: 624–641.

Hofmann, D. A., & Gavin, M. B. 1998. Centering decisionsin hierarchical linear models: Implications for re-search organizations. Journal of Management, 24:623–641.

Johns, G. 2001. In praise of context. Journal of Organi-zational Behavior, 22: 31–42.

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

Johns, G., & Nicholson, N. 1982. The meanings of ab-sence: New strategies for theory and research. InB. M. Staw & L. L. Cummings (Eds.), Research inorganizational behavior, 4: 127–172. Greenwich,CT: JAI Press.

Keller, R. T. 1983. Predicting absenteeism from prior

2014 379Diestel, Wegge, and Schmidt

Page 28: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

absenteeism, attitudinal factors, and nonattitudinalfactors. Journal of Applied Psychology, 68: 536–540.

Kozlowski, S. W. J., & Klein, K. J. 2000. A multilevelapproach to theory and research inorganizations:Contextual, temporal, and emergent processes. InK. J. Klein & S. W. J. Kozlowski (Eds.), Multileveltheory, research and methods in organizations:Foundations, extensions, and new directions: 3–90.San Francisco, CA: Jossey-Bass.

Kunin, T. 1955. The construction of a new type of atti-tude measure. Personnel Psychology, 8: 65–77.

Ledgerwood, A., & Trope, Y. 2010. Attitudes as globaland local action guides. In J. Forgas, J. Cooper & W.Crano (Eds.), The 12th annual Sydney symposiumof social psychology: The psychology of attitudesand attitude change. New York: Psychology Press.

Liao, H., Liu, D., & Loi, R. 2010. Looking at both sides ofthe social exchange coin: A social cognitive perspec-tive on the joint effects of relationship quality anddifferentiation on creativity. Academy of Manage-ment Journal, 53: 1090–1109.

Liao, H., & Rupp, D. E. 2005. The impact of justice cli-mate and justice orientation on work outcomes: Across-level multifoci framework. Journal of AppliedPsychology, 90: 242–256.

Liu, D., Mitchell, T. R., Lee, T. W., Holtom, B. C., &Hinkin, T. R. 2012. When employees out step withcoworkers: How job satisfaction trajectory and dis-persion influence individual- and unit-level volun-tary turnover. Academy of Management Journal,55: 1360–1380.

Locke, E. A. 1976. The nature and causes of job satisfac-tion. In M. D. Dunnette (Ed.), Handbook of indus-trial and organizational psychology: 1293–1349.Chicago: Rand McNally.

Lord, C. G., & Lepper, M. R. 1999. Attitude representationtheory. In M. P. Zanna (Ed.), Advances in experi-mental social psychology, vol. 31: 265–343. SanDiego, CA: Academic Press.

Markham, S. E. 1985. An investigation of the relationshipbetween unemployment and absenteeism: A multi-level approach. Academy of Management Journal,28: 228–234.

Mason, C. M., & Griffin, M. A. 2003. Group absenteeismand positive affective tone: A longitudinal study.Journal of Organizational Behavior, 24: 667–687.

Mathieu, J. E., Aguinis, H., Culpepper, S. A., & Chen, G.2012. Understanding and estimating the power todetect cross-level interaction effects in multilevelmodeling. Journal of Applied Psychology, 97: 951–966.

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.

Mesmer-Magnus, J. R., & DeChurch, L. A. 2009. Informa-tion sharing and team performance: A meta-analysis.Journal of Applied Psychology, 94: 535–546.

Mischel, W. 1977. The interaction of person and situa-tion. In D. Magnusson & N. S. Endler (Eds.), Person-ality at the crossroads: Current issues in interac-tional psychology: 333–352. Hillsdale, NJ: Erlbaum.

Moreland, R. L., & Levine, J. M. 2001. Socialization inorganizations and work groups. In M. E. Turner(Ed.), Groups at work: Theory and research: 69–112. Mahwah, NJ: Lawrence Erlbaum.

Mowday, R. T., & Sutton, R. I. 1993. Organizational be-havior: Linking individuals and groups to organiza-tional contexts. Annual Review of Psychology, 44:195–229.

Neuberger, O., & Allerbeck, M. 1978. Messung und Anal-yse von Arbeitszufriedenheit. Erfahrungen mitdem “Arbeitsbeschreibungsbogen (ABB)” [Mea-suring and analysis of job satisfaction]. Bern, Ger-many: Huber.

Ng, T. W. H., & Feldman, D. C. 2010. Occupational em-beddedness and job performance. Journal of Organ-izational Behavior, 30: 863–891.

Opotow, S. 1995. Drawing the line: Social categorization,moral exclusion, and the scope of justice. In B. B.Bunker & J. Z. Rubin (Eds.), Conflict, cooperation,and justice: Essays inspired by the work of MortonDeutsch: 347–369. San Francisco, CA: Jossey-Bass.

Postmes, T., Spears, R., Sakhel, K., & De Groot, D. 2001.Social influence in computer-mediated communica-tion: The effects of anonymity on group behavior.Personality and Social Psychology Bulletin, 27:1243–1254.

Preacher, K. J., Curran, P. J., & Bauer, D. J. 2006. Compu-tational tools for probing interactions in multiplelinear regression, multilevel modeling, and latentcurve analysis. Journal of Educational and Behav-ioral Statistics, 31: 437–448.

Raudenbush, S. W., & Bryk, A. S. 2002. Hierarchicallinear models: Applications and data analysismethods (2nd ed.). Thousand Oaks, CA: Sage Publi-cations.

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.

Sagie, A. 1998. Employee absenteeism, organizationalcommitment, and job satisfaction: Another look.Journal of Vocational Behavior, 52: 156–171.

Salancik, G. R., & Pfeffer, J. 1978. A social information

380 AprilAcademy of Management Journal

Page 29: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

processing approach to job attitudes and task design.Administrative Science Quarterly, 23: 224–253.

Schleicher, D. J., Watt, J. D., & Greguras, G. J. 2004.Reexamining the job satisfaction–performance rela-tionship: The complexity of attitudes. Journal ofApplied Psychology, 89: 165–177.

Schmidt, K.-H. 2002. Organisationales und individuellesAbwesenheitsverhalten:EineCross-Level-Studie [Or-ganizational and individual absence behavior: Across-level study]. Zeitschrift für Arbeits- und Or-ganisationspsychologie, 46: 69–77.

Schmidt, K.-H. 2010. The relation of goal incongruenceand self-control demands to indicators of job strainamong elderly care nursing staff: A cross-sectionalsurvey study combined with longitudinally assessedabsence measures. International Journal of NursingStudies, 47: 855–863.

Schneider, B. 1987. The people make the place. Person-nel Psychology, 40: 437–453.

Schneider, B., Salvaggio, A. N., & Subirats, M. 2002.Climate strength: A new direction for climate re-search. Journal of Applied Psychology, 87: 220–229.

Schneider, B., White, S. S., & Paul, M. C. 1998. Linkingservice climate and customer perceptions of servicequality: Test of a causal model. Journal of AppliedPsychology, 83: 150–163.

Shacham, R. 2009. A utility for exploring HLM, 2- and3-way interactions [Computer software]. Availablefrom http://www.shacham.biz/HLM_int/index.htm.

Sherif, M. 1936. The psychology of social norms. NewYork: Harper.

Shin, S. J., Lee, T.-Y., Kim, J.-Y., & Bian, L. 2012. Cogni-tive team diversity and individual team member cre-ativity: A cross-level investigation. Academy ofManagement Journal, 55: 197–212.

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.

Smith, M. B., Bruner, J., & White, R. 1956. . Opinions andpersonality. New York: Wiley.

Snedecor, G., & Cochran, W. 1983. Statistical methods(8th ed.). Ames, IA: Iowa State University Press.

Snyder, M., & DeBono, K. G. 1989. Understanding thefunctions of attitudes: Lessons from personality andsocial behavior. . In A. R. Pratkanis, S. J. Breckler, &A. G. Greenwald (Eds.) Attitude structure and func-tion: 339–359. Hillsdale, NJ: Lawrence.

Somers, M. J. 1995. Organizational commitment, turn-over and absenteeism: An examination of direct and

interaction effects. Journal of Organizational Be-havior, 16: 49–58.

Steel, R. P. 2003. Methodological and operational issuesin the construction of absence variables. Human Re-source Management Review, 13: 243–251.

Steel, R. P., Rentsch, J. R., & van Scotter, J. R. 2007.Timeframes and absence frameworks: A test ofSteers and Rhodes’ (1978) model of attendance. Jour-nal of Management, 33: 180–195.

Tajfel, H. 1978. Social categorization, social identity, andsocial comparison. In H. Tajfel (Ed.), Differentiationbetween social groups: Studies in the social psy-chology of intergroup relations: 61–76. London:Academic Press.

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

Thibaut, J. W., & Kelley, H. H. 1959. The social psychol-ogy of groups. New York: John Wiley & Sons.

Thomas, J., & Griffin, R. 1989. The power of social infor-mation in the workplace. Organizational Dynam-ics, 18: 63–75.

Trevor, C. O. 2001. Interactions among ease-of-movementdeterminants and job satisfaction in the prediction ofvoluntary turnover. Academy of Management Jour-nal, 44: 621–638.

van Dick, R., van Knippenberg, D., Kerschreiter, R., Her-tel, G., & Wieseke, J. 2008. Multiple organizationalidentities: The interactive effect of work group andorganizational identification on job satisfaction andextra-role behavior. Journal of Vocational Behav-ior, 72: 388–399.

van Knippenberg, D., & van Schie, E. C. M. 2000. Fociand correlates of organizational identification. Jour-nal of Occupational and Organizational Psychol-ogy, 73: 137–147.

von Zerssen, D. 1976. Die Beschwerden-Liste. Manual[List of complaints. Manual]. Weinheim, Germany:Beltz.

Wallace, D. S., Paulson, R. M., Lord, C. G., & Bond, C. F.,Jr.. 2005. Which behaviors do attitudes predict?Meta-analyzing the effects of social pressure and per-ceived difficulty. Review of General Psychology, 9:214–227.

Watt, S. E., Maio, G. R., Rees, K., & Hewstone, M. 2007.Functions of attitudes towards ethnic groups: Effectsof level of abstraction. Journal of Experimental So-cial Psychology, 43: 441–449.

Wegge, J., Schmidt, K.-H., Parkes, C., & van Dick, R. 2007.“Taking a sickie”: Job satisfaction and job involve-ment as interactive predictors of absenteeism in a

2014 381Diestel, Wegge, and Schmidt

Page 30: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

public organization. Journal of Occupational andOrganizational Psychology, 80: 77–89.

Weiss, H. M. 2002. Deconstructing job satisfaction: Sep-arating evaluations, beliefs and affective experi-ences. Human Resource Management Review, 12:173–194.

Wellen, J. M., Hogg, M. A., & Terry, D. J. 1998. Groupnorms and attitude-behavior consistency: The role ofgroup salience and mood. Group Dynamics: The-ory, Research, and Practice, 2: 48–56.

Xie, J. L., & Johns, G. 2000. Interactive effects of absenceculture salience and group cohesiveness: A multi-level and cross-level analysis of work absenteeism inthe Chinese context. Journal of Occupational andOrganizational Psychology, 73: 31–52.

Stefan Diestel ([email protected]) is a postdoctoral re-search fellow at the Leibniz Research Centre for WorkingEnvironment and Human Factors at the Technical Uni-versity of Dortmund. He received his PhD (2011) in work

and organizational psychology from the Ruhr Universityof Bochum in Germany. His research focuses on absen-teeism, emotional labor, job attitudes, and self-controlat work.

Jürgen Wegge ([email protected]) is cur-rently a full professor of work and organizational psy-chology at the Technical University of Dresden. Heearned his PhD (1994) in industrial and organizationalpsychology from the Technical University of Dortmundin Germany. His research interests are in the fields ofwork motivation, leadership, demographic change, andoccupational health.

Klaus-Helmut Schmidt ([email protected]) is a profes-sor of work and organizational psychology at the Leib-niz Research Centre for Working Environment and Hu-man Factors at the Technical University of Dortmund.He holds a PhD (1987) in psychology from the Univer-sity of Wuppertal in Germany. His research focuses onabsenteeism, occupational stress, and self-controlat work.

382 AprilAcademy of Management Journal

Page 31: THE IMPACT OF SOCIAL CONTEXT ON THE RELATIONSHIP …

Copyright of Academy of Management Journal is the property of Academy of Managementand its content may not be copied or emailed to multiple sites or posted to a listserv withoutthe copyright holder's express written permission. However, users may print, download, oremail articles for individual use.