collective efficacy beliefs in student work teams: relation to self-efficacy, cohesion, and...

Journal of Vocational Behavior 68 (2006) 73–84

www.elsevier.com/locate/jvb

Collective eYcacy beliefs in student work teams: Relation to self-eYcacy, cohesion,

and performance

Robert W. Lent ¤, Janet Schmidt, Linda Schmidt

Department of Counseling and Personnel Services, University of Maryland, College Park, MD 20742, USA

Received 15 January 2005Available online 23 May 2005

Abstract

A measure of collective eYcacy was developed and administered to undergraduates work-ing in project teams in engineering courses. Findings in each of two samples revealed that themeasure contained a single factor and was related to ratings of team cohesion and personaleYcacy. Collective eYcacy was also found to relate to indicators of team performance at bothindividual and group levels of analysis. Consistent with social cognitive theory, collectiveeYcacy was a stronger predictor of team performance than team members’ perceptions of theirself-eYcacy. We consider the implications of these Wndings for further research, theory, andpractice on team functioning within occupational and educational settings. 2005 Elsevier Inc. All rights reserved.

Keywords: Collective eYcacy; Self-eYcacy; Cohesion; Performance; Student work teams

1. Introduction

Social cognitive theory (Bandura, 1986, 1997) is an inXuential approach tounderstanding the psychological and social processes involved in human motivation,self-regulation, choice, and performance. A large body of research has accumulated

* Corresponding author. Fax: +1 301 405 9995.E-mail address: [email protected] (R.W. Lent).

0001-8791/$ - see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jvb.2005.04.001

mailto: [email protected]

mailto: [email protected]

74 R.W. Lent et al. / Journal of Vocational Behavior 68 (2006) 73–84

relating social cognitive variables, especially self-eYcacy, to various aspects of educa-tional and career behavior (e.g., Lent, Brown, & Hackett, 2002; Stajkovic & Luthans,1998). The focus of this literature has been on the relation of social cognitive variablesto outcomes achieved by students and workers as individuals. Such a focus is under-standable given that vocational and educational psychologists have traditionally beenconcerned with maximizing the development and remediating the problems of individ-uals, and that prevalent reward mechanisms in educational and work settings (e.g.,grades, salaries) tend to be linked to the quality of individuals’ performance andachievement. However, group processes have been garnering increasing attentionamong educational and organizational scholars in recent years, reXecting the growingpopularity of team approaches to learning and working (e.g., Stajkovic & Lee, 2001).

Although research on social cognitive theory has emphasized individual-levelmechanisms (e.g., self-eYcacy) and outcomes, the theory is also concerned with howpeople work together within teams and other social units. For instance, collectiveeYcacy, the group counterpart to self-eYcacy, is a key social cognitive element thatmay help to explain how groups function more or less well together. Bandura (1997)deWned collective eYcacy as a “group’s shared beliefs in its conjoint capabilities toorganize and execute the courses of action required to produce given levels of attain-ments” (p. 477). In contrast to self-eYcacy, which involves a person’s beliefs abouthis or her ability to perform particular behaviors individually, collective eYcacyrefers to group members’ aggregate beliefs about how they can perform as a unit. Theliterature on collective eYcacy has grown much more slowly than that of self-eYcacy,but its research base has expanded considerably in recent years and it has proven tobe a very Xexible group-level explanatory construct, Wnding application to groups ofdiverse size, function, and organizational context (Zaccaro, Blair, Peterson, &Zazanis, 1995).

While eVect sizes vary from study to study and not all studies have demonstratedimpressive collective eYcacy-criterion relations (e.g., Lee, Tinsley, & Bobko, 2002;Riggs & Knight, 1994), collective eYcacy has been reliably linked to (a) antecedentfactors (e.g., prior group achievement, Goddard, 2001; training, Gibson, 2001; self-eYcacy, Fernandez-Ballesteros, Diez-Nicolas, Caprara, Barbaranelli, & Bandura,2002); (b) group process and environment factors (e.g., team cohesion, Paskevich,Brawley, Dorsch, & Widmeyer, 1999; leadership climate, Chen & Bliese, 2002); (c)aVective outcomes (e.g., job satisfaction, psychological strain, organizational commit-ment, Jex & Bliese, 1999); and (d) group performance (Gibson, 2001; Goddard, 2001;Parker, 1994).

Stajkovic and Lee (2001) recently reported a meta-analysis of collective eYcacy-performance relations, involving data from 35 studies (including 67 correlation esti-mates and nearly 3000 groups with an average size of 3.8 members per group). Theyfound an average correlation between collective eYcacy and performance of .45.Thus, across the set of studies, collective eYcacy accounts for roughly 20% of the var-iance in group performance, representing a moderately strong eVect size. Consistentwith expectations, task interdependence was found to moderate collective eYcacy-group performance relations, with stronger relations under conditions where tasksrequire high versus low member coordination. Collective eYcacy-performance

R.W. Lent et al. / Journal of Vocational Behavior 68 (2006) 73–84 75

relations did not diVer substantially as a function of study design characteristics (e.g.,experimental vs. correlational studies) or type of sample (student vs. managerial/pro-fessional groups).

A separate meta-analysis by Gully, Incalcaterra, Joshi, and Beaubien (2002)reported similar relations of collective eYcacy to performance (corrected mean corre-lation of .41). They also conWrmed the Wnding that high-interdependence groups pro-duce larger collective eYcacy-performance relations than do low-interdependencegroups. In addition to interdependence, Gully et al.’s meta-analysis tested level ofanalysis as an eVect size moderator, Wnding that collective eYcacy relates more highlyto performance at the team than at the individual level of analysis. (In the individuallevel of analysis, individual team members’ ratings were correlated with performance;in the group level of analysis, team ratings were combined across team members toproduce aggregate percepts of team capabilities.)

Although collective eYcacy has not yet been applied to the context of engineeringeducation, there has been growing emphasis on student acquisition of team skills andexperience in engineering education within recent years (ABET, 2000). Mirroring theimportance of work teams in the engineering workplace, student project teams aredesigned to enhance the learning process by enabling students to develop skills atmanaging team interactions. Use of teams also allows students to work on more real-istic engineering problems (e.g., design of a bridge vs. the sizing of one beam). How-ever, team interpersonal dynamics often pose unique challenges for students andprofessors, such as how to handle inter-member conXicts and ensure that all studentsare contributing to, and proWting from, the team experience (Brannick, Roach, &Salas, 1993; Society of Manufacturing Engineers, 1997). Much is yet to be learnedabout what factors enhance team functioning and how such factors can be intention-ally fostered by professors and team members. It therefore seems important to studygroup-level variables, such as collective eYcacy, that may both shed light on projectteam functioning and suggest ways to assist teams to work together more eVectively.

In the present study, we sought to examine the factor structure, correlates, andpredictive validity of a novel measure of collective eYcacy within the context of stu-dent project teams in engineering. SpeciWcally, we Wrst developed a measure of collec-tive eYcacy beliefs linked to student team functioning and administered it, alongwith measures of self-eYcacy and group process (team cohesion) to students enrolledin a freshman engineering course involving student project teams. Since it has beensuggested that a team’s collective eYcacy is likely to derive from such sources as theself-eYcacy of its individual members (Bandura, 1997) and perceptions of team cohe-sion (Zaccaro et al., 1995), we expected individuals’ collective eYcacy percepts torelate to their personal eYcacy beliefs and team cohesion ratings.

We also explored the relations of team members’ aggregate collective eYcacy esti-mates to their ratings of self-eYcacy, team cohesion, and performance, as well as toexternal (instructor) ratings of team performance. Based on theory and prior Wnd-ings, we expected that collective eYcacy beliefs would (a) be predicted by the combi-nation of cohesion and self-eYcacy, (b) be predictive of team performance asassessed both by team members and course instructors, and (c) serve as stronger pre-dictors of team performance than do percepts of self-eYcacy.


2. Method

2.1. Participants

Two samples were recruited for the study. The Wrst sample, which participated inthe initial measure development phase of the study, consisted of 165 students (81%male) enrolled in an introductory engineering design course at a large Eastern univer-sity. In terms of class standing, 67% were Wrst-year students, 24% were sophomores,and 9% were juniors. The mean age was 19.52 years (SD D 3.12). Sixty-one percent ofthe participants were European American, 12% were African American, 5% were His-panic American, 16% were Asian American, and 6% reported other racial/ethnic iden-tiWcations. Students represented a variety of engineering majors/disciplines, with mostreporting mechanical (23%), electrical (21%), or computer (19%) engineering majors.

In the second, replication and extension, phase of the study, participants were 312students (74% men, 22% women, and 4% sex-unidentiWed) enrolled in a subsequentsemester of the same introductory engineering design course. They were divided into56 project teams. The students were primarily Wrst (78%) and second-year (16%) stu-dents. In terms of race/ethnicity, 8% self-identiWed as Black or African American, 4%as Hispanic, 18% as Asian or Asian American, 64% as White or European American,and 6% reported other (e.g., multiracial) racial/ethnic identiWcations. Most partici-pants were majoring in mechanical (26%), electrical (19%), aerospace (14%), or com-puter (13%) engineering specialties.

2.2. Procedure and instruments

Participants in both phases of the study completed measures of collective eYcacy,team cohesion, and self-eYcacy. They also provided demographic and academic sta-tus data. In addition, students in the second phase rated their team’s performance,and independent ratings of each team were also obtained from course instructors. Allmeasures were obtained during the last or next to last class meeting of the semester sothat students’ team ratings would be based on maximal exposure to their projectteams and their estimates of their self-eYcacy would be informed by at least onesemester of college-level experience.

As part of the engineering design course, students were divided into project teamsby course instructors. Teams were assigned a common project: to develop a workingwater pump by the end of the semester. They were expected to pool their talents andcoordinate all tasks needed to produce the Wnal product. In most cases, teams metweekly. They were allowed to select their own leaders and to distribute speciWc func-tions (e.g., technical report writing) as they saw Wt. Instructors served as team consul-tants and provided the teams with basic and technical background material related tothe design and construction of the project.

In the measurement development phase of the study, participants’ responses to themeasures were treated at the individual level of analysis whereas, in the second phaseof the study, both individual and group levels of analysis were performed. In the lat-ter phase, survey responses were obtained from 3 to 10 members per team


(average D 5.57 respondents per team). While all 56 teams provided collectiveeYcacy, team cohesion, and self-eYcacy ratings, student and instructor ratings ofteam performance were available only for 49 and 42 teams, respectively. For thepurpose of group-level analyses, team members’ ratings were averaged together toproduce group indices of collective eYcacy, self-eYcacy, cohesion, and performance.Course instructors (n D 10) also provided independent ratings of team performanceat the end of the semester.

2.2.1. Collective eYcacyThe collective eYcacy measure, which initially contained 18 items, was based on a

review of the collective eYcacy literature, conceptual analysis of the educational andinterpersonal tasks of the student project teams, and discussions with course instruc-tors, teaching assistants, and students who had previously taken the introductorydesign course. Participants were asked to indicate their conWdence in their team’sability to perform each of the 18 tasks successfully “as a unit, rather than how wellindividual group members perform.” Sample items included “work well togethereven in challenging situations” and “adapt to changes in group tasks or goals.” Stu-dents responded by rating their conWdence in their team’s capabilities on a 10-pointscale, from no conWdence (0) to complete conWdence (9). Psychometric data arepresented in the Results section.

2.2.2. Team cohesionTeam cohesion was assessed with an adapted version of the Cohesion subscale

from the Group Environment Scale (GES; Moos, 1986). The original subscale con-tained nine true/false items (e.g., “there is a feeling of unity and cohesion in thisgroup”) designed to reXect members’ “involvement in and commitment to the group,and the concern and friendship they show for one another” (Moos, 1986, p. 2). TheGES subscales have been used as indicators of group climate or process in diversegroup types (e.g., self-help, task-oriented groups). The Cohesion subscale, in particu-lar, has shown adequate internal consistency (�D .86) and one-month test-retest reli-ability (r D .79). In terms of validity, it has been found to be related to ratings ofgroup attraction and quality of group interaction (Moos, 1986). We substituted theword “team” for “group” in the items, reverse-scored negatively worded items, andhad participants indicate the degree to which they agreed with each of the statementsalong a 5-point Likert scale, from strongly disagree (1) to strongly agree (5). Internalconsistency reliability estimates were .82 and .92 in samples 1 and 2, respectively.

2.2.3. Self-eYcacyTo assess self-eYcacy, we used a 7-item measure asking participants to indicate

their conWdence in their ability to cope with a variety of barriers, or problems, thatengineering students could potentially experience (e.g., “cope with a lack of supportfrom professors or your advisor”). Participants responded to items on a 10-pointscale ranging from no conWdence (0) to complete conWdence (9). Scores were calculatedby dividing the summed item responses by 7, with higher scores reXecting strongercoping eYcacy beliefs. Prior studies have found that this measure produced adequate


internal consistency reliability estimates in general (�D .94; Lent et al., 2001) andengineering (�D .89; Lent et al., 2003) college student samples. The measure has alsobeen found to relate to measures of content-speciWc academic self-eYcacy, choice,barriers, and supports in theory-consistent ways (Lent et al., 2001, 2003). Weobtained alpha coeYcients in samples 1 and 2, respectively, of .88 and .90.

2.2.4. Team performanceParticipants in the second phase of the study were asked to assess their team’s per-

formance using a 10-item measure (e.g., “discussions are focused and useful;” “teammeetings are always productive”). Item responses were obtained along a 5-point scale(strongly disagree D 1; strongly agree D 5), summed, and divided by 10, producing a1–5 score range. Higher scores reXect more positive assessments of the team’s processin fulWlling its assignment. This measure had been designed by faculty associatedwith the design course to assist in evaluating team functioning.

Conceptually, whereas collective eYcacy involved members’ beliefs about theirteam’s prospective capabilities, the team performance measure tapped their percep-tions of how well the team had actually performed its tasks that semester. Despite thisconceptual distinction, correlations between collective eYcacy and performance arelikely to be inXated when both sets of ratings are made by team members (Gully et al.,2002). An eVort was made to reduce method bias by minimizing item overlap betweenthe two measures and by gathering collective eYcacy and student-rated performancemeasures in separate class meetings, one week apart. The student-rated performancemeasure produced a coeYcient alpha estimate of .90 in the second sample.

To acquire instructor ratings of team performance, at the end of the semester, justafter team projects were completed, instructors were asked to judge each team intheir course sections along three dimensions: amount of eVort put into the project,quality of the product, and how eVectively the team functioned overall. Each dimen-sion was rated on a 3-point scale, with higher ratings reXecting better team perfor-mance. The item ratings were summed and divided by three, producing a 1–3 scorerange. Such ratings are conceptually similar to group project grades, which have beenused as performance criteria in prior studies of collective eYcacy with student teams(Lee et al., 2002; Peterson, Mitchell, Thompson, & Burr, 2000).

3. Results

To examine the latent structure of the new collective eYcacy measure, we subjectedits items to an exploratory factor analysis in the Wrst sample, using principal axis fac-toring and oblimin oblique rotation. Good support was obtained for a single-factorsolution (e.g., only one eigenvalue >1 and a clear scree after one factor), whichaccounted for 69% of the total variance after rotation. All 18 items loaded substan-tially on the single factor. Given this simple factor structure and the desirability of cre-ating a brief collective eYcacy measure for use in future research, we decided to retainonly half of the items (those with the highest factor loadings), yielding a 9-iteminstrument with loadings ranging from .83 to .92. The correlation between the original


18-item version and the briefer scale was .98, suggesting that the two are measuring thesame construct. Scores on the brief scale were calculated by dividing the summed itemresponses by nine (the Wnal number of retained items), yielding a possible score rangeof 0–9, with higher scores reXecting stronger collective eYcacy beliefs.

To examine the stability of the factor structure, we replicated the factor analysiswith the 9-item collective eYcacy measure in the second sample, once again Wndingsupport for a single-factor solution (one eigenvalue >1, scree after one factor, 63% oftotal variance explained). The items, along with their loadings in both samples, areshown in Table 1. CoeYcient alpha estimates of the collective eYcacy measure in theWrst and second samples were, respectively, .96 and .94.

We next tested the hypothesis that collective eYcacy would relate to self-eYcacyand perceptions of team cohesion. Collective eYcacy was found to correlatemoderately with self-eYcacy (r D .35, p < .001) and strongly with team cohesion(r D .66, p < .001) in sample 1. Self-eYcacy and team cohesion were also interrelated toa small (r D .20), though signiWcant (p < .05), degree. The means of the collectiveeYcacy, self-eYcacy, and cohesion scales were, in order, 7.06 (SD D 1.65), 6.45(SD D 1.52), and 3.59 (SD D .68).

Table 2 contains means, standard deviations, and intercorrelations among thevariables in the second sample, presenting Wndings at both the individual and grouplevels of analysis. (Instructor ratings of team performance appear only as a group-level variable because teams, rather than individuals, constituted the sole ratingtarget on this variable.) The individual level correlations were consistent with the

Table 1Collective eYcacy items and factor loadings

Note. N D 165 for Sample 1; N D 312 for Sample 2. Kaiser–Meyer–Olkin index D .95 and .93 for Samples 1and 2, respectively. Copyright 2004 by R.W. Lent, J. Schmidt, and L. Schmidt.

a Collective eYcacy items were preceded by the stem, “How conWdent are you that your team couldƒ”

Collective eYcacy itemsa Factor 1

Sample 1 Sample 2

Reach agreement about what needs to get done at each meeting

.86 .82

Find ways to bridge individual diVerences (e.g., in age, major, or personality) between team members

.84 .76

Assist members who are having diYcultywith certain tasks

.83 .79

Develop a workable project design in a reasonable amount of time

.83 .76

Communicate well with one another despite diVerences in cultural background

.84 .76

Adapt to changes in group tasks or goals .89 .82Work well together even in challenging

situations.92 .85

Deal with feedback or criticism from thecourse instructor

.84 .78

Find ways to capitalize on the strengths of each member

.87 .79


Wndings in the Wrst sample (e.g., collective eYcacy correlated substantially with bothteam cohesion, r D .60, and self-eYcacy, r D .46).

We had also hypothesized that collective eYcacy beliefs would be predicted by thecombination of cohesion and self-eYcacy. To test this hypothesis, we Wrst performeda multiple regression analysis predicting collective eYcacy at the individual level ofanalysis in the Wrst sample, Wnding that cohesion and self-eYcacy together werestrongly predictive of collective eYcacy, accounting for 49% of the predictive vari-ance (the beta weights for cohesion and self-eYcacy were, respectively, .62 and .22,p < .001). This pattern of Wndings was replicated in the second sample, where cohe-sion (� D .53) and self-eYcacy (� D .36) jointly explained 48% of the variation incollective eYcacy.

Before testing our hypotheses at the group level of analysis, we calculated theintraclass correlation coeYcients (ICC) of each student-rated variable. ICC valuesreXect scale reliability at the group level, or the extent to which teams can be reliablydiVerentiated from one another using mean levels of each variable. ICC values forcollective eYcacy, cohesion, student-rated performance, and self-eYcacy were,respectively, .44, .49, .54, and .14. These values suggest that teams could be reliablydiVerentiated based on mean levels of all variables except self-eYcacy. Althoughmuch of the variance in each variable was at the individual level, team membershipdid seem to aVect students’ perceptions across the group-focused variables, justifyingtheir inclusion in group-level analyses. Self-eYcacy may best be considered an indi-vidual-level variable since the teams did not vary appreciably as a function of theirmembers’ self-eYcacy scores. However, we analyzed self-eYcacy at individual andgroup levels based on theoretical considerations—speciWcally, to test social cognitiveassumptions that collective eYcacy reXects more than just aggregation of team mem-bers’ individual self-eYcacy beliefs, and is a better predictor of team performancewhere outcomes depend on joint eVort (e.g., Moritz & Watson, 1998).

In a multiple regression analysis predicting collective eYcacy at the group level ofanalysis (sample 2 only), cohesion (� D .65), and self-eYcacy (�D .45) each explainedsigniWcant unique predictive variance, jointly accounting for 65% of the variance in

Table 2Scale correlations, means, and standard deviations: Sample 2

Note. Individual-level correlations appear below the diagonal; group-level correlations are above the diag-onal. Team Perform 1 D Ratings of team performance by students; Team Perform 2 D Ratings of teamperformance by course instructors.

¤ p < .05, one-tailed.

Scales 1 2 3 4 5 M SD n

1. Collective eYcacy — .48¤ .67¤ .70¤ .44¤ 7.57 .70 562. Self-eYcacy .46¤ — .06 .20 .08 6.55 .69 563. Cohesion .60¤ .19¤ — .62¤ .52¤ 3.95 .49 564. Team Perform1 .60¤ .17¤ .51¤ — .43¤ 4.32 .39 495. Team Perform2 — — — — — 2.57 .47 42M 7.56 6.54 3.97 4.31 —SD 1.19 1.47 .77 .63 —N 312 311 312 234 —


collective eYcacy. This pattern of Wndings was similar to that obtained at the individ-ual level of analysis, although analysis at the group level accounted for 16–17% moreof the variation in collective eYcacy. On balance, these Wndings support our hypothe-ses about the antecedents of collective eYcacy.

We had also predicted that collective eYcacy would be predictive of team perfor-mance as assessed both by team members and course instructors. Examination ofTable 2 conWrms that collective eYcacy was strongly related to the team’s assessmentof its performance at both the individual (r D .60) and group (r D .70) levels. Collec-tive eYcacy was also moderately related to instructor ratings of team performance(r D .44). Thus, the Wndings support the hypothesized relation of collective eYcacyrelative to team performance. Student and instructor ratings of team performancewere also found to be moderately interrelated (r D .43) and cohesion was stronglyrelated to both student and instructor ratings of team performance. However, corre-lations of self-eYcacy to team cohesion and performance criteria were consistentlysmall at both the individual and group levels of analysis.

Comparison of correlations of the performance criteria with collective eYcacy andself-eYcacy revealed that, in each case, team performance was more strongly relatedto collective eYcacy than to self-eYcacy. SpeciWcally, correlations with student rat-ings of team performance at the individual level of analysis were .60 for collectiveeYcacy and. 17 for self-eYcacy, t (234) D 7.88, p < .001; at the group level, the r’s were,respectively, .70 and .20, t (49) D 4.71, p < .001. Correlations with instructor ratings ofteam performance at the group level of analysis were .44 for collective eYcacy and. 08for self-eYcacy, t (42) D 2.49, p < .05.

4. Discussion

The present Wndings provide initial support for the reliability and validity of ournovel measure of project team collective eYcacy. A factor analysis indicated that thecollective eYcacy items comprised a single factor, and this latent structure was repli-cated in a second sample. The measure produced satisfactory estimates of internalconsistency reliability. Collective eYcacy was also found to be predicted by perceptsof self-eYcacy and team cohesion, both at individual and group levels of analysis.

The relation of self-eYcacy to collective eYcacy is consistent with Bandura’s(1997) contention that group members draw partly on estimates of their personaleYcacy in judging their teams’ eYcacy; prior research has also shown that the twoforms of eYcacy belief are interrelated (e.g., Fernandez-Ballesteros et al., 2002;Parker, 1994). The relation of collective eYcacy to team cohesion was also consis-tent with previous Wndings (e.g., Paskevich et al., 1999) and theoretical assumptions(e.g., Zaccaro et al., 1995). That collective eYcacy was strongly related to teamcohesion, while self-eYcacy produced a small correlation with cohesion, supportsthe validity of the social cognitive conception of collective eYcacy as a group-focused construct (Bandura, 1997). In fact, most of the variance in self-eYcacy wasattributable to individuals’ private appraisals rather than to their sharedexperiences as group members.


Consistent with theory (Bandura, 1997) and prior Wndings (Gully et al., 2002), ourresults also indicated that collective eYcacy was predictive of team performance, asjudged by students as well as instructors. The strong relation of collective eYcacy toteam-rated eVectiveness was similar to Gully et al.’s (2002) meta-analytic eVect sizewhere performance was indexed by team ratings. Of course, this relationship waslikely inXated by mono-source bias because team members provided both sets of rat-ings. While the team’s view of its own performance does oVer a useful vantage pointfrom which to assess team functioning, greater weight should probably be placed oninstructor ratings in assessing the magnitude of the collective eYcacy-team perfor-mance relationship. As “insiders,” students have a privileged, in-depth perspective ontheir team’s process. Yet by virtue of their role and subject matter knowledge,instructors have the Wnal say in grading team performance, including determining thequality of team products.

Collective eYcacy-performance relations have been found to be moderated bytask interdependence and level of analysis in past research (Gully et al., 2002). We,unfortunately, could not examine eVect sizes as a function of task interdependencebecause all teams were assigned the same task and degree of inter-member coordina-tion within teams was not assessed. We did Wnd a tendency for collective eYcacy toaccount for more variance in team-rated performance at the group than individuallevel of analysis (49 vs. 36% of variance, respectively), but these correlations werequite large at both levels of analysis. (Cohesion and self-eYcacy also tended toexplain more variation in collective eYcacy at group vs. individual levels of analysis.)The Wnding that collective eYcacy predicted team performance better than did self-eYcacy across performance indicators and levels of analysis supports assumptionsabout the predictive utility of collective eYcacy relative to performances that requirecollective eVort (Bandura, 1997).

Interpretation and generalization of our Wndings should be tempered by severalconsiderations. In particular, while initial data regarding the reliability and validityof our novel measure of collective eYcacy were promising, the distribution of collec-tive eYcacy scores displayed negative skew (i.e., most participants reported relativelyhigh collective eYcacy beliefs) and positive kurtosis in both samples. It may thereforebe useful to revise this measure by adding more challenging team tasks. In addition,though Wndings were consistent with our conceptualization of the antecedents andconsequences of collective eYcacy, cause and eVect relationships cannot be inferredgiven the largely cross-sectional nature of our design. Future research employing lon-gitudinal and experimental designs is needed to better establish the relationshipsbetween variables that are assumed to inform, and result from, collective eYcacy.Finally, multilevel statistical methods, encompassing individual and group levels ofanalysis, oVer valuable tools in the further study of collective eYcacy processes(Moritz & Watson, 1998).

In sum, our Wndings suggest that collective eYcacy in engineering student projectteams (a) can be measured reliably with a brief instrument, (b) is related to teamcohesion and self-eYcacy, and (c) may oVer diVerential predictive validity relative toself-eYcacy in explaining project team performance. Student teams are becoming anincreasingly popular learning medium in engineering, business, and other academic


Welds, mirroring the prominence of teams in the workplace. Thus, it is important toexamine theoretical mechanisms, like collective eYcacy, that may both shed light onteam functioning under natural conditions and suggest developmental or remedialsteps for promoting eVective teamwork.

Acknowledgments

We wish to thank Drs. Neil Goldschmidt, Gary Pertmer, Brad Brenner, HeatherLyons, Daniel Singley, and Dana Treistman for their assistance with this research.This material is based upon work supported by the National Science Foundationunder Grant No. DUE-0089079.

References

ABET (2000). Engineering criteria 2000. Baltimore, MD: Accrediting Board for Engineering and Technology.Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood CliVs,

NJ: Prentice-Hall.Bandura, A. (1997). Self-eYcacy: The exercise of control. New York: W.H. Freeman.Brannick, M. T., Roach, R. M., & Salas, E. (1993). Understanding team performance: A multimethod

study. Human Performance, 6, 287–308.Chen, G., & Bliese, P. D. (2002). The role of diVerent levels of leadership in predicting self- and collective

eYcacy: Evidence for discontinuity. Journal of Applied Psychology, 87, 549–556.Fernandez-Ballesteros, R., Diez-Nicolas, J., Caprara, G. V., Barbaranelli, C., & Bandura, A. (2002). Deter-

minants and structural relation of personal eYcacy to collective eYcacy. Applied Psychology: An Inter-national Review, 51, 107–125.

Gibson, C. B. (2001). Me and us: DiVerential relationships among goal-setting training, eYcacy and eVec-tiveness at the individual and team level. Journal of Organizational Behavior, 22, 789–808.

Goddard, R. D. (2001). Collective eYcacy: A neglected construct in the study of schools and studentachievement. Journal of Educational Psychology, 93, 467–476.

Gully, S. M., Incalcaterra, K. A., Joshi, A., & Beaubien, J. M. (2002). A meta-analysis of team-eYcacy,potency, and performance: Interdependence and level of analysis as moderators of observed relation-ships. Journal of Applied Psychology, 87, 819–832.

Jex, S. M., & Bliese, P. D. (1999). EYcacy beliefs as a moderator of the impact of work-related stressors: Amultilevel study. Journal of Applied Psychology, 84, 349–361.

Lee, C., Tinsley, C. H., & Bobko, P. (2002). An investigation of the antecedents and consequences of group-level conWdence. Journal of Applied Social Psychology, 32, 1628–1652.

Lent, R. W., Brown, S. D., Brenner, B., Chopra, S. B., Davis, T., Talleyrand, R., et al. (2001). The role ofcontextual supports and barriers in the choice of math/science educational options: A test of socialcognitive hypotheses. Journal of Counseling Psychology, 48, 474–483.

Lent, R. W., Brown, S. D., Hackett, G. (2002). Social cognitive career theory. In D. Brown and Associates,Career choice and development. (4th ed., pp. 255–311). San Francisco: Jossey-Bass.

Lent, R. W., Brown, S. D., Schmidt, J., Brenner, B., Lyons, H., & Treistman, D. (2003). Relation of contex-tual supports and barriers to choice behavior in engineering majors: Test of alternative social cognitivemodels. Journal of Counseling Psychology, 50, 458–465.

Moos, R. H. (1986). Group environment scale. Palo Alto, CA: Consulting Psychologists Press.Moritz, S. E., & Watson, C. B. (1998). Levels of analysis issues in group psychology: Using eYcacy as an

example of a multilevel model. Group Dynamics: Theory, Research, and Practice, 2, 285–298.Parker, L. E. (1994). Working together: Perceived self- and collective-eYcacy at the workplace. Journal of

Applied Social Psychology, 24, 43–59.


Paskevich, D. M., Brawley, L. R., Dorsch, K. D., & Widmeyer, W. N. (1999). Relationship between collec-tive eYcacy and team cohesion: Conceptual and measurement issues. Group Dynamics, 3, 210–222.

Peterson, E., Mitchell, T. R., Thompson, L., & Burr, R. (2000). Collective eYcacy and aspects of sharedmental models as predictors of performance over time in work groups. Group Processes and IntergroupRelations, 3, 296–316.

Riggs, M. L., & Knight, P. A. (1994). The impact of perceived group success-failure on motivational beliefsand attitudes: A causal model. Journal of Applied Psychology, 79, 755–766.

Society of manufacturing engineers (1997). Manufacturing education plan: Phase I report. Industry identi-Wes competency gaps among newly hired engineering graduates.

Stajkovic, A. D., & Lee, D. (2001). A meta-analysis of the relationship between collective eYcacy and groupperformance. Paper presented at the meeting of the Academy of Management, Washington, DC.

Stajkovic, A. D., & Luthans, F. (1998). Self-eYcacy and work-related performance: A meta-analysis.Psychological Bulletin, 124, 240–261.

Zaccaro, S. J., Blair, V., Peterson, C., & Zazanis, M. (1995). Collective eYcacy. In J. E. Maddux (Ed.),Self-eYcacy, adaptation, and adjustment: Theory, research, and application (pp. 305–328). New York:Plenum.

collective efficacy beliefs in student work teams: relation to self-efficacy, cohesion, and...

Documents