Withholding Inputs in Team Contexts Member Composition, Interaction Processes, Evaluation Structure, and Social Loafing

Journal of Applied Psychology Copyright 2006 by the American Psychological Association 2006, Vol. 91, No. 6, 1375–1384 0021-9010/06/$12.00 DOI: 10.1037/0021-9010.91.6.1375 Withholding Inputs in Team Contexts: Member Composition, Interaction Processes, Evaluation Structure, and Social Loafing Kenneth H. Price David A. Harrison The University of Texas at Arlington The Pennsylvania State University This document is copyrighted by the American Psychological Association or one of its allied publishers. Joanne H. Gavin This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Marist College Social loafing was observed as a naturally occurring process in project teams of students working together for 3– 4 months. The authors assessed the contributions that member composition (i.e., relational dissimilarity and knowledge, skills, and abilities; KSAs), perceptions of the team’s interaction processes (i.e., dispensability and the fairness of the decision-making procedures), and the team’s evaluation structure (i.e., identifiability) make toward understanding loafing behavior. Identifiability moderated the impact of dispensability on loafing but not the impact of fairness on loafing. Perceptions of fairness were negatively related to the extent that participants loafed within their team. Specific aspects of relational dissimilarity were positively associated with perceptions of dispensability and negatively associated with perceptions of fairness, whereas KSAs were negatively associated with perceptions of dispensability. Keywords: loafing, teams, diversity, fairness, member composition Increasingly, organizations are attempting to maximize perfor- variety of other forces may be operating to influence loafing mance by using teams to accomplish work-related activities (e.g., behavior. Guzzo & Shea, 1992; Kirkman, 2000). For teams to achieve their maximal level of performance, motivational losses that can occur In the current article, we first propose and test relationships when participants work collectively on a task must be minimized between facets of the team’s interaction processes and social (Steiner, 1972). Why do individuals reduce their efforts or with- loafing. Then, working backward toward the more proximal inde- hold inputs when in team contexts? Social loafing theory (or pendent variables in our model, we attempt to link these process collective effort theory; Karau & Williams, 1993) provides one set variables with compositional variables of the team members (Fig- of answers about motivation losses. ure 1 brings together and organizes these constructs). To increase our understanding of why team members withhold their input as a Reviews of the literature on social loafing indicate a consistency naturally occurring process, we discuss the observed loafing in in findings across studies (Shepperd, 1993) with an effect size in project teams. To expand our theoretical understanding of loafing the middle range of comparable social behavior (Karau & Wil- in this setting, we integrate explanatory constructs from organiza- liams, 1993). Loafing behavior, however, has been most frequently tional justice theory and the literature on work-group diversity demonstrated in laboratory settings comparing collective group with traditional loafing constructs. performance with coactive group performance on simple maximiz- ing and additive tasks (Karau & Williams, 1993). The social As members of a project team, on the basis of perceptions of the loafing paradigm also tends to concentrate on structural reasons team’s interaction processes, individuals can make judgments (e.g., identifiability, the potential for external evaluation of indi- about their dispensability and the fairness of the decision-making vidual contributions to the team) for withheld inputs. Less evi- procedures. Whereas dispensability is a previously identified an- dence has been forthcoming from the highly interdependent and tecedent of loafing (Karau & Williams, 1993), this is the first time, conjunctive contexts of naturally occurring teams (such as task at least of which we are aware, that the relationship between forces; see George, 1992, for a notable exception) in which a perceptions of the fairness of the decision-making procedures and loafing behavior is proposed and assessed. A third potential ante- Kenneth H. Price, Department of Management, The University of Texas cedent of loafing, a contextual variable in our model, refers to at Arlington; David A. Harrison, Department of Management and Orga- whether individual contributions to the team’s output are identifi- nization, Smeal College of Business, The Pennsylvania State University; able in the evaluation system for the team. Given the critical role Joanne H. Gavin, School of Management, Marist College. that identifiability has played in loafing (Latane´, Williams, & Harkins, 1979), both its main and moderating effects are exam- We thank Anna Florey and Meghna Virick for their assistance in the ined. In terms of the latter effects, we study how identifiability data collection phases and Nancy Rowe for her assistance in the data might temper or neutralize the outflow of dispensability and fair- analysis phase of this study. ness on the restriction of team member effort (e.g., George, 1992). Correspondence concerning this article should be addressed to Kenneth In examining how member composition contributes to under- H. Price, Department of Management, The University of Texas at Arling- standing loafing behavior, we drew from the literature on team ton, Box 19467, Arlington, TX 76019. E-mail: [email protected] diversity (e.g., Riordan, 2000) and tested the association between 1375

1376 RESEARCH REPORTS This document is copyrighted by the American Psychological Association or one of its allied publishers. Figure 1. Framework for extending the determinants of social loafing. H ϭ hypothesis. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. team member relational dissimilarity and perceptions of fairness task more difficult or challenging, or by providing team members and dispensability. Based on arguments from the social loafing with unique aspects of the task to work on. Under such conditions, literature (e.g., Kerr & Bruun, 1983), and also examined for the the performance of individuals working collectively and whose first time, was the proposed link between the unique knowledge, efforts were not identifiable did not differ from individuals work- skills, and abilities (KSAs) that team members bring with them to ing alone and whose efforts were identifiable. Later research by the group and dispensability. All potential relationships in our Kerr and Bruun (1983) reported that low-ability team members felt model were studied using a multiwave, multisource design. This more dispensable and loafed when task performance was deter- methodology offers protection against cross-sectional confounds mined by the high-ability team member; likewise, high-ability and reverse causality. In the following sections, we review theory members felt more dispensable and loafed when task performance and research supporting the model’s links. was determined by the low-ability team member. In a field study, George (1992) examined potential interaction effects between Loafing, Evaluation Structure, and Interaction Processes identifiability and intrinsic task involvement (akin to indispens- ability) on social loafing and reported that when individual con- Early research by Latane´ et al. (1979) found a reduction in task tributions are not identifiable, a strong external force to maintain performance when participants working collectively were com- high levels of effort is absent and the internal perception of pared with participants working coactively or individually on a indispensability matters most. Having an evaluation structure that task. These differences in performance were attributed to the makes member inputs identifiable can overcome the lack of mo- inability, when task responsibility is shared, to identify the contri- tivation that may occur when team members perceive that their butions of individual team members to the final group product. inputs are more dispensable. As a result, it is possible to rephrase Later research suggested it was the potential for individual eval- the relationship between dispensability, identifiability, and social uation that resulted in increased effort when social loafing was loafing by stating the following: reduced by making individual efforts identifiable (e.g., Williams, Harkins, & Latane´, 1981). This distinctive feature of identifiability Hypothesis 2: Team evaluation structures allowing identifi- (evaluation potential; Karau & Williams, 1993) combines input cation of members’ individual contributions will be associ- identifiability with the ability to evaluate that input against some ated with a weaker positive relationship between perceived intrapersonal, interpersonal, or objective standard (e.g., Szymanski dispensability and social loafing. That is, when member con- & Harkins, 1987). Consequently, consistent with previous research tributions are not identifiable, there will be a stronger positive and theorizing, we hypothesized the following: relationship between dispensability and loafing. Hypothesis 1: Team evaluation structures allowing identifi- Through their interactions, most teams process information and cation of members’ individual contributions will be nega- make a number of decisions to successfully finish their task tively associated with social loafing. assignment (Arrow, McGrath, & Berdahl, 2000). For example, decisions must be made about how to allocate work loads among Research by Harkins and Petty (1982) indicated that a second team members, what type of information to collect, and what factor contributing to loafing was dispensability. Dispensability course of action to follow or recommendations to make. Proce- could be reduced by making individuals feel that their efforts were dural justice theorists posit and report that a critical element of the more instrumental to team performance, either by making the team

RESEARCH REPORTS 1377 This document is copyrighted by the American Psychological Association or one of its allied publishers. processes used to make decisions is the extent to which the ment and higher absences in men (Tsui et al., 1992). In addition, This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. procedures are seen as fair (e.g., Lind & Tyler, 1988). Procedures Gruenfeld, Mannix, Williams, and Neale (1996) reported that perceived as fair contribute to positive affective reactions toward unfamiliarity with other team members is associated with lower one’s membership in the group (e.g., Greenberg, 1990), higher trust. As a result, information is more likely to be dismissed or levels of team member satisfaction and commitment to the team’s ignored if proposed by individuals who are relationally dissimilar. decision (e.g., Colquitt, Noe, & Jackson, 2002), and higher levels of team member performance (e.g., Hunton, Price, & Hall, 1996). Taking these ideas a step further, we expected that team mem- Given that perceived unfairness is thought to hamper individual bers who are more relationally dissimilar would also perceive their performance by reducing levels of motivation, we expected that it efforts as more dispensable. Increasing feelings of dispensability would also generate the reduced inputs of loafing behavior in may result from expectations of differential treatment or actual teams. Hence, our next hypothesis was the following: differences in treatment by other team members on the basis of predictions from the similarity–attraction paradigm (Bryne, 1971), Hypothesis 3: Perceived fairness of a team’s decision pro- social identity theory (Tajfel, 1978), and self-characterization the- cesses will be negatively associated with social loafing. ory (Turner, 1982). Dissimilar team members may react to such treatment in ways that parallel the predictions about out-group As with dispensability, we anticipated that fairness consider- members in leader–member exchange theory: a belief that others ations would have stronger influences on social loafing when have a greater responsibility for task completion (e.g., Graen & evaluation structures make team member inputs less identifiable. Uhl-Bien, 1991). We therefore expected the following: When a strong external incentive for maintaining high levels of effort is removed, members’ perceptions or internal motivation Hypothesis 5: Team member relational dissimilarity will be that reflect their views of the fairness of team processes may have positively associated with perceived dispensability. more potent influences on behavior. Hence, we predicted: In addition to demographic characteristics, research reviewed Hypothesis 4: Team evaluation structures allowing identifi- earlier also suggests that perceptions of dispensability increase cation of members’ individual contributions will be associ- when team members believe they do not have the ability to make ated with a weaker negative relationship between perceived unique contributions to the group (Kerr & Bruun, 1983). This has fairness and social loafing. That is, when member contribu- been counteracted in laboratory settings by providing team mem- tions are not identifiable, there will be a stronger negative bers with nonredundant caches of information or portions of the relationship between fairness and loafing. task (Harkins & Petty, 1982). Carrying this notion to naturally occurring teams and borrowing long-standing ideas from job anal- Member Composition and Interaction Processes ysis and selection suggest that the uniqueness of the KSAs that the members bring with them to the group can increase perceptions of Evidence indicates that member compositional differences— indispensability. As a result, we predicted that: standing features such as age, gender, and ethnicity that make some members distinct from others in their teams— can have an Hypothesis 6: Task-relevant team member skills and (scho- impact on team processes (e.g., Pelled, Eisenhardt, & Xin, 1999). lastic) ability (i.e., KSAs) will be negatively associated with Theory and research suggest that these visible differences are often perceived dispensability. used for social characterization of team members (e.g., Riordan, 2000), forming the basis for faultlines that divide members of the The final link in our model asserts a negative connection be- larger team into in-groups and out-groups (Abrams, Wetherell, tween relational dissimilarity and the perceived fairness of a Cochrane, Hogg, & Turner, 1990; Lau & Murnighan, 1998; Tajfel team’s processes. The tendency of a team to marginalize or reduce & Turner, 1986). Consequences of being relationally dissimilar in communications with dissimilar members also means that the ethnicity, although at times nonsymmetrical across different eth- “voice” of such members is less likely to be heard during team nicities, can result in being a nascent out-group member (i.e., “at deliberations (Milliken & Martins, 1996). Not only does suppres- the margins” of the team) and include reduced influence, lower sion of voice have a direct effect on the focal person’s (i.e., group commitment, increased absences, and lowered perceptions dissimilar member’s) perceptions of unfairness (Greenberg, 1990), of performance (e.g., Mehra, Kiduff, & Brass, 1998; Riordan & but also the reduced participation of such members leads to less Shore, 1997; Tsui, Eagan, & O’Reilly, 1992). Relational dissimi- representative, and therefore less just, decisions (Colquitt et al., larity in age among team members can lead to lower levels of team 2002). Additionally, Azzi (1992) has argued that dissimilar team member conscientiousness, sportsmanship, and altruism (Chatto- members may perceive themselves as a minority with less process padhyay, 1999; Riordan & Weatherly, 1999); increased turnover; control resulting from being numerically underrepresented in the and importantly, fewer and more conflict-laden communications group. If procedures such as majority vote are more likely to be about technical and other issues with dissimilar team members used in more diverse groups (Gruenfeld et al., 1996), this would (Milliken & Martins, 1996; O’Reilly, Caldwell, & Barnett, 1989; tend to reinforce the unfairness of the decision procedures for the Zenger & Lawrence, 1989). Dissimilarity in sex has been associ- minority members of the team (Hunton et al., 1996). Conse- ated with feelings of isolation, dissatisfaction, and lack of attach- quently, relational dissimilarity may be both a cause and an attri- ment in women (Konrad, Winter, & Gutek, 1992; Pelled & Xin, bution for the behavior that unfolds in the team. Thus, our last 1997) and, in some situations, with lower organizational attach- hypothesis was as follows:

1378 RESEARCH REPORTS This document is copyrighted by the American Psychological Association or one of its allied publishers. Hypothesis 7: Team member relational dissimilarity will be project. This final questionnaire contained a three-item measure of dis- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. negatively associated with perceived fairness of team deci- pensability that was adopted from the work of Harkins and Petty (1982) sion processes. and Kerr and Bruun (1983). Responses were summed across items asking team members the extent they, for example, “had unique skills or abilities Method to add” to the team on a scale ranging from 1 (none at all) to 6 (extraor- dinary amount). In coding, items were reflected so that higher scores Sample and Team Projects indicated higher levels of dispensability. Estimated reliability was ␣ ϭ .87. Participants were 515 students enrolled in 13 different undergraduate Perceived fairness of the team decision processes was measured with and graduate-level courses. Respondents varied in age (M ϭ 28 years, three items adopted from other justice studies (e.g., Lind & Tyler, 1988) SD ϭ 7), ethnicity (66%, Caucasian; 18%, Asian; 9%, African American; and work on group decision rules (e.g., Miller, Jackson, Mueller, & 6%, Hispanic; 1%, other), year in program (62%, seniors; 19%, graduate Schersching, 1987). Respondents were asked the extent to which their students; 16%, juniors; 2%, sophomores; 1%, freshmen), gender (50%, team’s decision processes were “fair” and “just” on a response scale female; 50%, male), and marital status (58%, single; 37%, married; 5%, ranging from 1 (very untrue) to 7 (very true). Another question asked divorced). participants which alternative best described their team’s decision pro- cesses: “most important team decisions were made by . . . : (a) one person Participants worked in project teams (N ϭ 144) for 9 to 14 weeks of a making the decision for the team, (b) a few people making the decision for 16-week semester, with 85% of the teams having 3– 6 members. Projects the team, (c) a majority making the decision for the team, or (d) all accounted for at least 10%, and up to 75% (M ϭ 0.36, SD ϭ 0.19), of members agreeing on a decision.” As one proceeds from the first to the last members’ final course grades. The projects had multiple parts, required the of the response options, the decision rule was regarded as more just. To use of different skills and abilities, contained both conjunctive and dis- equalize each item’s contribution to the total perceived fairness score, we junctive elements, and were designed, in part, to develop participant skills. standardized responses to each item first (to put them on the same metric, given they were initially measured with different response scales), and then Administration of Questionnaires and Measures the standardized items were summed. The estimated reliability for the measure of fairness was ␣ ϭ .76. Data were collected via questionnaires with return envelopes (addressed to the researchers) that were distributed in each class by the instructor. Confirmatory factor analysis was conducted to provide evidence of the None of these classes was taught by a member of the research team. Both underlying structure of the dispensability and fairness items (convergent questionnaires were completed using a numerical code, and participants and discriminant validity). A two-factor model showed a very close fit to were told via an informed consent that their responses were confidential. the data, exceeding recommended levels of all indices (comparative fit Only members of the research staff had access to the individual index and goodness of fit index ϭ .99; non-normed fit index ϭ .98; questionnaires. standardized root-mean-square residual ϭ .03; root-mean-square error of approximation ϭ .05 or close fit; Browne & Cudeck, 1993; Jo¨reskog & Before project teams were formed, participants completed the first So¨rbom, 1993). The factor correlation was Ϫ.22. This solution also fit questionnaire requesting demographic information (e.g., age, gender), significantly better than a one-factor model, ⌬␹2(1) ϭ 337.07, p Ͻ .01. grade point average, and self-categorization of team member technical skills and abilities. To clearly differentiate KSAs from relational dissimi- To measure loafing, our dependent measure, we asked each individual to larity, we studied only non-task-related demographic differences (Pelled, rate the extent to which each team member: “loafed by not doing his or her 1996) in ethnicity, gender, age, year in the program (e.g., junior, senior), share of the tasks, by leaving work for others to do, by goofing off, and by and marital status. Although marital status has not typically been examined having other things to do when asked to help out.” Ratings were made on in previous studies of relational dissimilarity, it was included because it a 7-point scale ranging from 1 (extremely likely to loaf) to 7 (extremely (along with being a senior member of the team) might convey differences unlikely to loaf). In coding, items were reflected so that higher scores in social status (Tsui & Gutek, 1999) and expectations of work relation- indicated higher levels of loafing. This question includes elements of ships (e.g., availability for weekend meetings). loafing consistent with Veiga’s (1991) definition of self-limiting behavior and with the operational definition of loafing in a field study by George For each of the demographic variables (i.e., age, gender, ethnicity, year (1992). in the program, and marital status), the equation D ϭ [(1/n) ⌺ [(Si – Sj)2]1/2 was used to calculate measures of relational dissimilarity (Tsui et al., We computed James, Demaree, and Wolf’s (1993) interrater agreement 1992). For categorical variables, a match (0) versus mismatch (1) coding index (rwg) for teammates’ ratings of each focal team member. The median was used.1 To assess task-relevant KSAs, we provided participants with 13 rwg across the sample of over 500 persons was .87. Seventy-seven percent categories of experience and skills on the basis of categories using Stan- of the rwg indices were above .70. Note that rwg is being used in this context dard Industrial Classification codes, along with categories emphasizing to evaluate judgments of an individual-level stimulus (e.g., member with- programs of study (e.g., engineering, information technology), and asked holding inputs) rather than a group-level stimulus (e.g., cohesiveness). them to check all that applied. After meeting with instructors of the Hence, we consider these values to be acceptable. They do not appear to sampled courses who picked what they considered the most important KSA suffer in comparison to the average interrater reliabilities for performance groupings, we selected 4 categories from the original 13: technical, quan- reported by Viswesvaran, Schmidt, and Ones (1996), which did not exceed titative, management/administration, and marketing/advertising. Two .52. points were awarded for each unique skill or ability a team member possessed, one point if the team member and any other team member(s) In 16 of the 23 classes, the instructor required team members to rate each possessed that skill or ability, and zero points if a team member did not teammate’s performance, which was then used in computing a project have that ability or skill. For each member, scores were then summed grade for each student. A categorical variable was created (0 ϭ not across categories to create a measure of the technical skills of the partic- identifiable, 1 ϭ identifiable) based on whether individuals were held ipant, part of the task-relevant skill component of KSAs. Grade point average was used to assess (scholastic) ability. 1 Consider a team with 4 members, 30, 29, 27, and 26 years old, respectively. The calculation of the D statistic for the 30-year-old individ- Measures of perceived dispensability, perceived fairness, and social ual would be as follows: D ϭ 1/3 [(30 – 29)2 ϩ (30 – 27) 2 ϩ (30 – 26)2]1/2, loafing were obtained when participants had or had almost completed their D ϭ 1/3 [12 ϩ 32 ϩ 42]1/2, D ϭ 1/3 [5.099], D ϭ 1.6997.

RESEARCH REPORTS 1379 This document is copyrighted by the American Psychological Association or one of its allied publishers. accountable for their contributions to the team’s project. Team members Member Composition and Interaction Processes This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. were aware of this component. This “is to be evaluated” aspect is the distinctive feature of the conceptual definition of identifiability (Karau & Tests of the anticipated positive relationship between the rela- Williams, 1993). tional dissimilarity measures and dispensability (Hypothesis 5) and the negative relationship between the relational dissimilarity mea- Methods of Analysis and Control Variables sures and fairness (Hypothesis 7) received partial support. Rela- tional differences in marital status contributed to members’ feel- Hierarchical linear regression was used to test our hypotheses. Control ings that their inputs were more dispensable (␤ ϭ .19, p Ͻ .01; variables that were entered in Step 1 included the percentage of the class Column 2, Step 3). Being different in marital status and ethnicity grade that the project was worth (i.e., project percentage); group size; were each negatively associated with perceptions that the team whether the teams were selected by the instructors, the students (i.e., used fair decisions processes (␤ ϭ Ϫ.17, p Ͻ .01 and ␤ ϭ Ϫ.16, self-selected), or a mixed strategy was used (i.e., mixed selection); and p Ͻ .01, respectively; Column 3, Step 3). Supporting Hypothesis whether extra credit was offered for research participation (i.e., credit). 6, technical skills and scholastic ability (i.e., KSAs) were each Entered in Step 2, in addition to the control variables, were individual significantly and negatively related to perceived dispensability demographic variables (e.g., age; mean levels of these variables can con- (␤ ϭ Ϫ.09, p ϭ .05 and ␤ ϭ Ϫ.15, p Ͻ .01, respectively; Column found relational dissimilarity effects). For demographic variables with 2, Step 4). multiple categories, we created dummy variables to capture the range of attributes (e.g., sophomore, junior). Relational dissimilarity measures were Discussion added in Step 3, and technical skills and scholastic ability (task-relevant KSAs) were added in Step 4 to the variables considered in previous steps. Loafing in Teams We also entered main effects for identifiability, dispensability, and per- ceived fairness in Step 5, prior to testing the moderating effects of iden- Although observed in the literature and predicted by Hypothesis tifiability in Step 6. The large number of control variables in dissimilarity 1, we did not find a detectable main effect of identifiability on our research (Riordan, 2000) as well as the nonnormal (dichotomous) nature of measure of loafing. One such field study, using a perceptual identifiability and some control variables obviate the use of structural measure of identifiability and supervisory ratings of loafing, re- equations modeling. ported a main effect of identifiability on loafing behavior among sales personnel in a retail store (George, 1992). Perhaps identifi- Results ability was less potent in our study than in other research because team members anticipated more lenient ratings from coworkers, Loafing, Evaluation Structure, and Interaction Processes the rewards resulting from different levels of effort may have been less significant, hard measures of performance assessing loafing Means, correlations, and standard deviations for all measured were not available as in some laboratory studies (e.g., Harkins & variables are reported in Table 1. Results testing all hypotheses can Petty, 1982), or because we used a dichotomized environmental be found in Table 2. The betas reported in this table are the values variable of identifiability rather than a perceptual measure, which for variables considered at each step in the analysis. allows a greater range of the construct to be assessed. Hypothesis 1 was not supported. Team evaluation structures that However, consistent with the literature and Hypothesis 2, the allowed individual contributions to be identified were not signif- relationship between dispensability—a feature of how members icantly associated with social loafing (␤ ϭ Ϫ.01, p ϭ .82; Column perceive a team’s interaction process—and loafing was strength- 1, Step 5). Supporting Hypothesis 3, perceptions of fairness were ened in the absence of an individual evaluation structure imposed negatively associated with loafing behavior (␤ ϭ Ϫ.12, p Ͻ .01; by the environment. As reported in previous research (e.g., Har- Column 1, Step 5). Hypotheses 2 and 4 examined the moderating kins & Petty, 1982), the form of this interaction suggests that the impact of identifiability on the relationship between dispensability motivational potential created by teams treating their members’ and loafing, and the linkage between fairness and loafing. Consis- inputs as unique and indispensable is especially important when tent with Hypothesis 2, there was a significant Dispensability ϫ individual contributions are not measured. Our findings also indi- Identifiability interaction (␤ ϭ Ϫ.40, p Ͻ .01; Column 1, Step 6). cated a main effect of dispensability on loafing. Other research has The sign of this interaction indicated, as predicted, a significant indicated that identifiability, in some cases, may overwhelm task weakening of the positive relationship between dispensability and considerations such as intrinsic task involvement (e.g., George, loafing (␤ ϭ .24, p Ͻ .01; Column 1, Step 5). Procedures de- 1992). The association between dispensability and loafing in our scribed by Aiken and West (1991) were used to illustrate this study could reflect the reduced potency of identifiability. It could interaction. Separate regression lines were plotted linking dispens- also reflect the increased salience of dispensability, given its link ability with loafing for individuals high and low in identifiability with both KSAs and relational differences in our teams— elements (see Figure 2). The plotted values for the dispensability measure that are not present in other studies of loafing (e.g., Harkins & ranged from ϩ1SD to Ϫ1SD around the mean, whereas identifi- Petty, 1982). ability was a categorical (0, 1) variable. Results indicated that when there was no mechanism for identifying and evaluating each Perceptions of fairness, consistent with Hypothesis 3, were team member’s contribution to the team, the positive relationship negatively related to loafing, but there was no support for Hypoth- between dispensability and loafing was strengthened. There was esis 4. Fairness did not matter more when participants were not no dependable evidence for Hypothesis 4, as the Fairness ϫ identifiable. Perhaps perceptions of being treated unfairly and the Identifiability interaction term was not significant (␤ ϭ .04, p ϭ resentfulness that it can create may simply result in stronger .54; Column 1, Step 6). affective reactions that overwhelm identifiability considerations







RESEARCH REPORTS 1383 This document is copyrighted by the American Psychological Association or one of its allied publishers. References review and theoretical integration. Journal of Personality and Social This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Psychology, 65, 681–706. Abrams, D., Wetherell, M., Cochrane, S., Hogg, M. A., & Turner, J. C. Kerr, N., & Bruun, S. (1983). Dispensability of effort and group motiva- (1990). Knowing what to think by knowing who you are: Self- tional losses: Tree-rider effects. Journal of Personality and Social Psy- categorization and the nature of norm formation, conformity, and group chology, 44, 78 –94. polarization. British Journal of Social Psychology, 29, 97–119. Kirkman, B. L. (2000). Why do employees resist teams? Examining the “resistance barrier” to work team effectiveness. International Journal of Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and Conflict Management, 11, 74 –93. interpreting interactions. Newbury Park, CA: Sage. Konrad, A., Winter, S., & Gutek, B. (1992). Diversity in work group sex composition. Research in the Sociology of Organizations, 10, 115–140. Arrow, H., McGrath, J. E., & Berdahl, J. L. (2000). Small groups as Latane´, B., Williams, K., & Harkins, S. (1979). Many hands make light the complex systems. Thousand Oaks, CA: Sage. work: The causes and consequences of social loafing. Journal of Per- sonality and Social Psychology, 47, 822– 832. Azzi, A. E. (1992). Implicit and category-based allocations of decision- Lau, D. C., & Murnighan, J. K. (1998). Demographic diversity and fault making power in majority–minority relations. Journal of Experimental lines: The compositional dynamic of organizational groups. Academy of Social Psychology, 29, 203–228. Management Review, 23, 325–340. Lawrence, B. S. (1997). The black box of organizational demography. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model Organization Science, 8, 1–22. fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural models (pp. 132–162). Thousand Oaks, CA: Sage. justice. New York: Plenum Press. MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Bryne, D. (1971). The attraction paradigm. New York: Academic Press. Sheets, V. (2002). A comparison of methods to test mediation and other Chattopadhyay, P. (1999). Beyond direct and symmetrical effects: The intervening variable effects. Psychological Methods, 7, 83–104. McGrath, J. E. (1984). Groups: Interaction and performance. Englewood influence of demographic dissimilarity on organizational citizenship Cliffs, NJ: Prentice Hall. behaviors. Academy of Management Journal, 42, 273–287. Mehra, A., Kiduff, M., & Brass, D. J. (1998). At the margins: A distinc- Colquitt, J. A., Noe, R. A., & Jackson, C. L. (2002). Justice in teams: tiveness approach to the social identity and social networks of under- Antecedents and consequences of procedural justice climate. Personnel represented groups. Academy of Management Journal, 41, 441– 452. Psychology, 55, 83–109. Miller, C. E., Jackson, P., Mueller, J., & Schersching, C. (1987). Some Crampton, S. M., & Wagner, J. A. (1994). Percept–percept inflation in social psychological effects of group decision rules. Journal of Person- micro-organizational research: An investigation of prevalence and ef- ality and Social Psychology, 52, 325–332. fect. Journal of Applied Psychology, 79, 67–76. Milliken, F. J., & Martins, L. L. (1996). Searching for common threads: Folger, R., & Cropanzano, R. (1998). Organizational justice and human Understanding the multiple effects of diversity in organizational groups. resource management. Thousand Oaks, CA: Sage. Academy of Management Review, 21, 402– 433. George, J. M. (1992). Extrinsic and intrinsic origins of perceived social O’Reilly, C. A., III, Caldwell, D. F., & Barnett, W. P. (1989). Work group loafing in organizations. Academy of Management Journal, 35, 191– demography, social integration, and turnover. Administrative Science 202. Quarterly, 34, 21–37. Graen, G. B., & Uhl-Bien, M. (1991). The transformation of work group Pelled, L. H. (1996). Demographic diversity, conflict, and work group professionals into self-managing and partially self-designing contribu- outcomes: An intervening process theory. Organization Science, 7, tors: Toward a theory of leadership-making. Journal of Management 615– 631. Systems, 3(3), 33– 48. Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. (1999). Exploring the black Greenberg, J. (1990). Organizational justice: Yesterday, today, and tomor- box: An analysis of work group diversity, conflict, and performance. row. Journal of Management, 16, 399 – 432. Administrative Science Quarterly, 44, 1–28. Gruenfeld, D. H., Mannix, E. A., Williams, K. Y., & Neale, M. A. (1996). Pelled, L. H., & Xin, K. R. (1997). Birds of a feather: Leader member Group composition and decision making: How member familiarity and demographic dissimilarity and organizational attachment in Mexico. information distribution affect process and performance. Organizational Leadership Quarterly, 8, 433– 450. Behavior and Human Decision Processes, 67, 1–15. Riordan, C. M. (2000). Relational demography within groups: Past devel- Guzzo, R. A., & Shea, G. P. (1992). Group performance and intergroup opments, contradictions, and new directions. Research in Personnel and relations in organizations. In M. D. Dunnette & L. M. Hough (Eds.), Human Resources Management, 19, 131–173. Handbook of industrial and organizational psychology (Vol. 3, 2nd ed., Riordan, C. M., & Shore, L. M. (1997). Demographic diversity and pp. 269 –313). Palo Alto, CA: Consulting Psychologists Press. employee attitudes: An empirical examination of relational demography Hackman, J. R. (1987). The design of work teams. In J. W. Lorsch (Ed.), within work units. Journal of Applied Psychology, 82, 342–358. Handbook of organizational behavior (pp. 315–342). Englewood Cliffs, Riordan, C. M., & Weatherly, E. W. (1999, August). Relational demog- NJ: Prentice Hall. raphy within groups: An empirical test of a theoretical model. Paper Harkins, S., & Petty, R. (1982). Effects of task difficulty and task unique- presented at the meeting of the Academy of Management, Chicago. ness on social loafing. Journal of Personality and Social Psychology, 43, Shepperd, J. A. (1993). Productivity loss in performance groups: A moti- 1214 –1229. vation analysis. Psychological Bulletin, 113, 67– 81. Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A. T. (2002). Time, Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural teams, and task performance: Changing effects of diversity on group equations models. In S. Leinhart (Ed.), Sociological methodology (pp. functioning. Academy of Management Journal, 45, 129 –145. 290 –312). San Francisco: Jossey-Bass. Hunton, J. E., Price, K. H., & Hall, T. W. (1996). A field experiment Steiner, I. D. (1972). Group processes and productivity. Orlando, FL: examining the effects of membership in voting majority/minority sub- Academic Press. groups and the ameliorative effects of postdecisional voice. Journal of Szymanski, K., & Harkins, S. G. (1987). Social loafing and self-evaluation Applied Psychology, 81, 806 – 812. James, L. R., Demaree, R. G., & Wolf, G. (1993). RWG: An assessment of within-group interrater agreement. Journal of Applied Psychology, 78, 306 –309. Jo¨reskog, K. G., & So¨rbom, D. (1993). LISERL 8: Structural equations modeling with SIMPLIS command language. Chicago: Scientific Soft- ware International. Karau, S. J., & Williams, K. D. (1993). Social loafing: A meta-analytic

1384 RESEARCH REPORTS This document is copyrighted by the American Psychological Association or one of its allied publishers. with a social standard. Journal of Personality and Social Psychology, 53, Viswesvaran, C., Schmidt, F. L., & Ones, D. S. (1996). Comparative This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 891– 897. analysis of the reliability of job performance ratings. Journal of Applied Tajfel, H. (1978). Differentiation between social groups: Studies in the Psychology, 81, 557–575. social psychology of intergroup relations. New York: Academic Press. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup Webber, S. S., & Donahue, L. M. (2001). Impact of highly and less behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of inter- job-related diversity on work group cohesion and performance: A meta- group relations (pp. 7–24). Chicago: Nelson-Hall Publishers. analysis. Journal of Management, 27, 141–162. Tsui, A. S., Eagan, T. D., & O’Reilly, C. A. (1992). Being different: Relational demography and organizational attachment. Administrative Williams, K., Harkins, S., & Latane´, B. (1981). Identifiability as a deterrent Science Quarterly, 37, 549 –579. to social loafing: Two cheering experiments. Journal of Personality and Tsui, A. S., & Gutek, B. A. (1999). Demographic differences in organi- Social Psychology, 40, 303–311. zations: Current research and future directions. Lanham, MD: Lexing- ton Books. Zenger, T. R., & Lawrence, B. S. (1989). Organizational demography: The Turner, J. C. (1982). Toward a cognitive redefinition of the social group. differential effects of age and tenure distributions on technical commu- In Henri Tajfel (Ed.), Social identity and intergroup relations (pp. nication. Academy of Management Journal, 32, 353–376. 15– 40). Cambridge, MA: Cambridge University Press. Veiga, J. F. (1991). The frequency of self-limiting behavior in groups: A Received January 30, 2003 measure and an exploration. Human Relations, 44, 877– 895. Revision received August 26, 2005 Accepted August 29, 2005 Ⅲ