Vous êtes sur la page 1sur 16

Team Reward Attitude: Construct Development and Initial Validation

Author(s): Jason D. Shaw, Michelle K. Duffy and Eric M. Stark

Source: Journal of Organizational Behavior, Vol. 22, No. 8 (Dec., 2001), pp. 903-917
Published by: Wiley
Stable URL: http://www.jstor.org/stable/3649578
Accessed: 09-04-2015 04:23 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Wiley is collaborating with JSTOR to digitize, preserve and extend access to Journal of Organizational Behavior.


This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions

Journal of OrganizationalBehavior
J. Organiz.Behav. 22, 903-917 (2001)
DOI: 10.1002/job.121

Team reward attitude: construct

development and initial validation

'University of Kentucky,Lexington,U.S.A.
2JamesMadison University,U.S.A.


of a teamrewardattitudeconstructandintialvalidation
a measurein a longitudinal
factoranastudyof teammembers(initialn = 566).Confirmatory
lysis resultsprovidesupportfora revised9-itemscalein twodifferenttimeperiodsseparated
by fourmonths.Themeasurewaspositivelyrelatedto otherteam-related
forgroupworkandperceivedefficacyof teams)andlocusof control,andnegativelyrelatedto
a proxyforability.Implications
of theresearchandfutureresearchdirectionsareaddressed.
Copyright0 2001 JohnWiley& Sons,Ltd.

The use of team-basedrewardsis possibly the fastestgrowingrewardpractice,with up to 70 per cent of
US organizationsnow using some type of team-basedrewards(DeMatteoet al., 1998; Ledfordet al.,
1995). Conceptually,team-basedrewardsare often arguedto have positive effects on cooperationand
collective motivation(e.g., Shamir,1990), behaviorscritical to the success or 'smooth functioning'of
the group(DeMatteoet al., 1998: 144). But line of sight issues (MilkovichandWigdor,1991), the limits
of team-based rewards in fostering individual motivation (DeMatteo et al., 1998), and potential
increasedcompetitionbetweengroups(Mohrmanet al., 1992) areprevalentcontrarianarguments.Thus,
the argumentthatinterdependentrewardsareuniversallyeffective (or not so) has conceptuallimitations,
althoughsuch treatmentscontinueto appear(e.g., Campionet al., 1993). Interestingly,otherresearchers
have long noted the contingencies involved in developing effective rewardsystems (e.g., Miller and
Hamblin,1963), with particularattentionpaid to the consistencyof rewardand task structures.Designs
that incorporateeither interdependentrewardsand tasks, or independentrewardsand tasks, tend to be
more effective than mismatcheddesigns (e.g., see Cotton and Cook, 1982 for a review).
With exceptions (e.g., Cable and Judge, 1994; DeMatteo and Eby, 1997-paper presentedat the
Annual Meetings of the Academy of Management,Boston, U.S.A.; Yamagishi, 1988), little research
attentionhas been paid to individualdifferences in rewardpreferencesor receptivity to team-based
rewards. The few studies that have examined reward attitudes tend to focus on satisfaction with
team-based rewards as an outcome, using personal characteristics,individual difference variables,
* Correspondenceto: J. D. Shaw,
Universityof Kentucky,GattonCollege of Business and Economics, School of Management
445L, Lexington, KY 40506-0034, U.S.A.

Copyright? 2001 JohnWiley& Sons,Ltd.

Received 12 February 1999

Revised 19 May 2000

Accepted13 September

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



and personality as predictorsof such satisfaction (see DeMatteo et al., 1998 for a review). These
studies tend to overlook an importantlink in the process of forming team-basedrewardsatisfaction,
namely, individual attitudes about receiving team- or individually-basedrewards.Individualdifferences such as personalcharacteristics,other attitudes,priorpositive or negative experiences in teams,
andpersonalityvariableslikely contributeto the formationof rewardattitudeswhich, in turn,influence
satisfactionor dissatisfactionwith this type of reward.We begin to addressthese issues by: (1) describing the team rewardattitude(TRA) construct;(2) distinguishingit from related,but conceptuallydistinct, constructs;(3) outlininga potentialset of correlatesof TRA; (4) describingthe developmentof a
TRA measure;and (5) conducting tests for discriminantand convergentvalidity of the construct.

Team Reward Attitude (TRA)

TRA is an individual'sgeneralevaluationof receiving rewardsbased on the performanceof the team;
high TRA reflects a positive evaluationof receiving team-basedrewardsand low TRA reflects a positive evaluationof receiving individually-basedrewardsin team situations.TRA is anchoredat one end
by a positive evaluation of rewardsdistributedbased on an equality principle and on the other by
rewardsdistributedbased on equity (Chen and Church,1993). High TRA individualspositively evaluate equally distributedrewards in teams; low TRA individuals positively evaluate equity-based
rewardsin teams. Equity-or individually-basedrewarddistributionsystems place an emphasison individual differencesin performancewithinteams, andconsequentlyrewardteam membersdifferentially.
Low TRA individualsare disposed towardthis equity principle.By contrast,equality- or team-based
reward distributionsystems tend to punctuatethe common elements and similarities among group
members;members are rewardedequally as a result. High TRA individualsbelieve that teammates
are 'all in this together' and prefer rewardsbased solely on team performance.Between these two
poles are individualswhose attitudesare 'moreevenly matched'such that the attitudeis a compromise
between differentways of distributingrewards(Leventhalet al., 1980: 176). Chen and Church(1993),
noting a scarce literaturebase, arguedthatrewardattitudesshould 'be given the same weight and attention as more rational and observable variables when exploring the effective application of various
forms of distributivesystems and principles' (p. 43).
We elaborateon TRA in accordancewith the tripartiteview of attitudes,i.e., individual attitudes
have affective, cognitive, and behavioralcomponents.The affective componentof attitudesconcerns
how the attitudinalobject makes an individualfeel, i.e., a liking or disliking dimension (Hogan and
Blake, 1996). The cognitive componentof attitudesinvolves an individual'sbelief about the object,
while the behavioralcomponent involves behavioralintentionswith respect to the attitudinalobject.
Therefore, individuals high on the TRA construct should: (1) like or prefer receiving team-based
rewardsin team contexts; (2) believe that such rewards are more effective than individually-based
rewardsin team contexts; and (3) intend to behave in ways consistent with their evaluationof attitudinal object (Bem, 1968).
A criticalfirststep in constructdevelopmentis to conceptuallydistinguishit from othersimilarconcepts. Thus, it is necessary to understandhow TRA conceptuallyrelates to similar constructsand to
fully delineate what TRA is not. We based our selection of potential correlatesof TRA, in part, on
Leventhalet al.'s (1980) theoryof allocationpreferences.These authorsnote that,beyond culturaldifferences, the primarydeterminantsof allocation preferences are prior experiences and other related
attitudes,i.e., rewardattitudesare enmeshed in an arrayof experiences and relatedattitudes.The theory of allocationpreferencesalso allows for the inclusion of stable individualdifferencesin ability and
Copyright@ 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions


personalityas correlatesof rewardattitudes;high ability team membersandthose concernedwith individual productivityshould favour equity-basedrewardsystems. Moreover, since team-basedreward
structuresinvolve a certain loss of control at the individuallevel, personalitydimensions that relate
to stable individualdifferencesin controlshouldalso be associatedwith TRA. These criteriawere used
to guide the generationof variablesthat could logically be expected to relate to TRA. The initial set
includes ability and experience factors, other team-relatedattitudes, and two personality variables
(locus of control and proactivepersonality).

Correlatesof TRA
Personal and experience factors
Severalauthorshave proposedthatindividualability levels may be the most importantindividualcharacteristicin formingreactionsto team-basedrewardsystems (e.g., DeMatteoet al., 1998). The reasoning here is two-fold. First,high abilitymembersmay reactnegativelyto equality-basedrewardsystems
since theirperformanceis likely to be higherthan the averagegroupmember,yet undersuch systems,
the rewardsare equal (DeMatteoet al., 1997). Rewardsystems in teams thatensurethathigh-performing membersget a greaterproportionof the rewardsare likely to be viewed more positively by high
performers.High ability individualsmay also be more concerned than those of average ability with
free-ridereffects in equality-basedsystems (Albanese and Van Fleet, 1985). The second aspect of the
role of ability in forming TRA is one of individual recognition (DeMatteo et al., 1998), i.e., higher
ability individualsmay preferenvirons where the cognizance associated with their individualcontributions is not masked by equality-basedrewards (see Bretz and Judge, 1994). While not directly
relatedto rewardattitudes,two recent studies provideevidence in supportof this position. In a laboratory study,Yamagishi(1988) found that high-performinggroupmembersoften chose to leave a group
and work individuallywhen rewardswere distributedequally to all groupmembers.Parket al. (1994)
found that turnoverwas greateramong high performingindividualsunder a group incentive system,
and greateramong low performerswhen incentives were individuallybased.
The quality of one's prior team experiences is also a factor that, interestingly,has not received a
great deal of attentionas a predictorof reactionsto team-basedrewardsystems, but the theoryof allocation preferencessuggests thatpriorexperiences are centralin mouldingallocationattitudes.Experience shapes perceptions of the utility of an attitudinalobject such that those seen as furthering
importantgoals are preferred(Brief, 1998). Individualswith priorpositive team experiencesare likely
to have developedmorepositive attitudestowardteam-basedrewards.Thus, the following two hypotheses are proposed:
Hypothesis 1: Ability will be negatively related to TRA.
Hypothesis2: Quality of prior team experiences will be positively related to TRA.
Other team-related attitudes
Two dimensions of the broader individualism/collectivism construct specific to work in teams
(preferencefor group work and perceived efficacy of teams) may also relate to TRA. Preferencefor
group work refers to the degree to which individuals have strong preferences for group ratherthan
independentwork. More specifically,preferencefor groupwork concernsthe relativeimportancethat
an individualholds for work design or structure,ratherthan the structureof rewardsystems, per se.
While we expect that preferencefor group work will be positively related to TRA, the variablesare
conceptuallydistinct. Most important,the object of TRA is the design of rewardstructures,while the
CopyrightC 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



object of preferencefor groupwork is the design of task structures,i.e., interdependentor independent

tasks. Individuals high in preference for group work like to interact with others in task situations
(Wagner, 1995), while individuals low in preference for group work prefer the opposite and are
attractedto situationswhere they can control their own work pace (Wagemen, 1995).
When comparingthe conceptualdifferencesbetween TRA andpreferencefor groupwork,it is clear
that an individualmay preferthe atmosphere,social interaction,and comraderyof teams, but still may
prefer and believe in the receipt of individually-basedrewardsin these contexts, and vice versa. A
professional athlete may enjoy playing and interactingwith teammates (high preference for group
work), but may preferto be paid on the basis of individualstatisticsand accomplishmentsratherthan
team performance(low TRA). Similarly,an individualin an organizationmay preferworkingin teams
ratherthanworkingin isolation because of social interaction,but may have a negative attitudetowards
receiving rewardsbased only on team performance.For these individuals,team-basedrewardsmay be
less motivatingthan rewardsfor individualperformance(Hayes, 1976). Moreover,working in teams
but being rewardedfor individualcontribution,or a hybridapproach(Wageman,1995), may be viewed
more positively among individualswho prefer individually-basedrewards.
Perceivedefficacyof teams concernsan individual'sbeliefs regardingthe ability of teams to accomplish work more effectively than individuals working alone, not an individual's attitudeconcerning
rewards.An individual'sbelief in the use of teams as a general concept does not necessarily inform
us as to how thatpersonmay or may not react to a given rewardsystem, althoughthe two concepts are
likely to be positively related. At a more general level, Bandura(1989) argues that individualswith
high efficacy beliefs are less likely to fear or be unhappywith factorsrelatedto a target(teams, in this
case) since they believe they controlthem. It seems plausiblethatthe morepositive one's belief toward
work accomplished in teams, the more positive one's attitudetowardthe rewardsassociated with it.
Hypothesis 3: Preferencefor group work will be positively related to TRA.
Hypothesis4: Perceived efficacy of teams will be positively related to TRA.
Personality characteristics
A numberof dispositions may relate to TRA, but as a point of departure,we focus on two specific
facets (locus of control and proactive personality).We chose specific dimensions in lieu of broader
ones (e.g., Big Five dimensions) since they more specifically concern control and environmental
mastery-central issues in team rewardsituations.Locus of control is an enduringtendency to attribute the causality of events either to internalor external factors (Rotter, 1966). Initial evidence that
those with an internallocus of control would preferindividually-basedrewardsystems comes from a
studyby Wittinget al. (1981). These authorsfound that when subjectsattributedperformanceto luck,
they tended to prefer equality-basedsystems, but preferredequity-basedsystems when performance
was attributedto theirown effort.Equity-basedrewardshighlightindividualcontributionsto the whole
and tend to emphasizea connectionbetween effort and the proportionalityof rewards.Internallocus of
control individuals should have a more positive attitudetoward rewardsystems that allow a clearer
effort to outcome link; externalsperceive a generalizeddisconnectionbetween effort and performance
and should thus be more positively inclined towardteam-basedrewards.Proactivepersonality refers
to an individual's stable tendency to affect environmentalchange (Bateman and Crant, 1993). High
proactivepersonalityis rootedin individualdifferencesin the need to manipulateandcontrolthe environment (e.g., Langer, 1983). Those low in proactivepersonalitytend to be reactive, shaped by their
environment,and adaptive.Operatingunderteam-basedrewardsystems involves a certainloss of control, i.e., the link between a proactiveindividual'smanipulationattemptsand outcomes is less certain
Copyright? 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions


than with individually-basedrewards.Hence, we expect that proactive personalityand TRA will be

negatively related.
Hypothesis5: Locusof control(higherscoresindicatingexternallocus)will be positivelyrelatedto TRA.
Hypothesis 6: Proactivepersonalitywill be negatively related to TRA.
The TRA construct was developed and some initial correlates outlined. The proposed correlates
include ability, quality of prior team experiences, preference for group work, perceived efficacy of
teams, locus of control, and proactivepersonality.The developmentof a TRA measure,a description
of the samples employed, and tests of the predictionsare describedbelow.

Pilot test sample
The pilot sample consisted of 101 junior and senior undergraduatebusiness studentsat a universityin
the southern US. The factor structureof an initial set of items was examined using this sample.
Participantscompleted questionnairesduringclass time and responses were anonymous.The sample
comprised 75 males and 26 females with an average age of 24.21 years.
Primary sample
The primarysample comes from a largerstudy of team-membereffectiveness (e.g., Duffy et al., 2000;
Shaw et al., 2000). The participantswere 566 upper-divisionundergraduatestudentsenrolled in business administrationcourses at a large universityin the southernUnited States. Permissionwas granted
by 11 instructors(a total of 17 classes) employing a group-basedclassroom style for the researchteam
to solicit participationfrom students.To more fully simulate actualwork teams, a class was eligible if
the instructorrequiredgroupsto complete several projects/assignmentsthroughoutthe term, assigned
groupsremainedintactthroughoutthe term, and substantialgroup-memberinteractionswere assured
by the class design. Participantswere guaranteedconfidentialityand were assured that participation
was voluntary.Three phases of self-reportdata were collected. Participantscompleted an initial questionnaire during the first week of class (Time 1; backgroundand demographicinformation,course
expectations, personality and attitudevariables, including TRA items). Time 2 transpiredthe week
following mid-term exams (group processes including task interdependenceitems). Time 3 data
(includingTRA items) were collectedjust priorto finalexaminations(8 weeks afterTime 2; 16 weeks
afterTime 1). The initial sample (n = 566) was 39 per cent female with an averageage of 22 years.The
modal class standingwas junior level. Groupsrangedin size from threeto seven memberswith a mean
size of 4.77 members (SD 1.07).

variables (Time 1)

Ability was measuredwith a commonly used proxy, grade point average (GPA) (e.g., Wagner,1995).
Participantsreportedtheir cumulativeuniversity GPA on the Time 1 questionnaire.Quality of prior
Copyright? 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



team experiences was measuredon the Time 1 questionnairewith the question: 'Not including sports
teams, have your experiences as a member of work teams been negative or positive?' The item has
seven response options ranging from (1) very negative to (7) very positive. Following Wagner's
(1995) approach,we used seven items taken from several sources (Barberet al., 1996-paper presented at the Annual Meeting of the Academy of Management,Cincinnati, OH, U.S.A.; Erez and
Earley, 1987; Wagnerand Moch, 1986; Wagner,1995) to measurepreferencefor group work.A sample item is: 'When I have a choice, I try to work in a group insteadof by myself.' Perceivedefficacyof
teams was measuredwith a 3-item scale from Barberet al. (1996-paper presented at the Annual
Meeting of the Academy of Management,Cincinnati, OH, U.S.A.). A sample item is: 'Teams can
more thoroughlyevaluate options than any one individualcan.' Locus of control was measuredusing
six items from James (1973). Higher scores on the measure indicate a more external orientation.A
sample item is: 'I have found that what will happenwill happen,regardlessof my actions.' Proactive
personality was assessed with six items from Bateman and Crant(1993). A sample item is: 'If I see
somethingI don't like, I fix it.' Because of space and time constraints,we includedonly a subsetof the
items from these measures;preferencewas given to items with higheraverage factor loadings in prior

Measures-control variables (Time 1, Time2, and course instructors)

To accountfor minordifferencesin class design, class structure,and assignments,we includeda set of
instructorcontrols.Instructorswho had more than one class included in the study had identical structures across their classes. Therefore, we created 10 dummy variables (to capture the 11 different
instructors)and enteredthese into the analysis. Severalothercontrols were includedto controlfor differences in individualwork and rewardpreferences,team context, and to synthesize the results with
previouslypublishedstudies from this data set. Age, gender (women = 0, men = 1), and class standing
(sophomore= 1, junior- 2, senior= 3) were included since they may relate to group work preferences, rewardpreferences, and effort levels (e.g., Mason, 1995). These variables were included on
the Time 1 questionnaire.Teamsize, task interdependence,and reward interdependencewere also
includedto controlfor the effects of the team context. Teamsize was collected from course instructors.
Taskinterdependence(e.g., I can't accomplishmy tasks withoutinformationfrom my team members)
was assessed with a 4-item scale from Campionet al. (1993) at Time 2. Rewardinterdependencewas
operationalizedas the per cent of an individual'scourse gradethatwas based on team performanceand
was collected from course instructors.

Initial scale developmentand item selection
The development of the TRA measure proceeded using attitude scale constructiontechniques for
Likert-typescales. After a literaturereview to determineif similarscales existed, 13 initial TRA items
were developed.Each item in the scale was an evaluativestatement,agreementwith which would indicate either a positive or negative attitude regardingreceiving rewardsbased on team performance.
Nine items were worded in a positive direction (i.e., a high score would suggest a positive attitude
Copyright? 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



concerningteam-basedreward)and four were wordedin a negative directionfor reverse scoring. The

items also reflected each aspect of the tripartitestructure;seven items were cognitive (beliefs about
team-based rewards), three were affective (like or dislike), and three were behavioral (behavioral
intentions in team-basedrewardsituations). Scores were recordedin a 7-point Likert-type(strongly
disagree-strongly agree) format.
The initial items were administeredto the pilot sample of 101 participantsand later analysed for
evidence of unidimensionalityand internalconsistency reliability.The initial 13 items displayed high
reliability(a = 0.83), but severalof the inter-itemcorrelationswere substantiallylower thanexpected.
The analysisproceededwith an exploratoryprincipalcomponentsanalysis with varimaxrotation.Two
factors were extractedfrom the procedure(nine and four items in each of the factors, respectively).
The wording of the four items which loaded on the second factor were examined and, based on these
analyses, two of these items were omittedfrom the primarystudy and majoradjustmentswere made to
the otheritems. Minoradjustmentswere made to severalof the remainingnine items in preparationfor
the second data collection.

Primary sample: confirmatory analyses

Time 1 confirmatory factor analyses
The 11 revised TRA items were administeredduringthe Time 1 phase of the longitudinalproject.A
confirmatoryfactor analysis (CFA)procedure(AMOS Version3.6) was used to examine the structure
of the scale. In additionto item factorloadings, severalindicatorsof model fit were assessed including
chi-square(X2).The goal of the X2 test is to fail to reject the null hypothesis, althoughthis is a virtual
impossibility with large sample sizes and many degrees of freedom (Bollen, 1989). To adjustfor parsimony, the ratio of X2 to degrees of freedom (X2/df)was also examined with ratios of less than 3: 1
considered indicatorsof good model fit. Goodness-of-fit(GFI), adjustedgoodness-of-fit (AGFI;GFI
adjustedfor parsimony),and normedfit (NFI) indexes were also assessed. Valuesin excess of 0.90 are
consideredto be acceptablefor GFI,AGFI,and NFI (Medskeret al., 1994). Root mean squareerrorof
approximation(RMSEA) was also assessed with values less than 0.08 indicatingreasonablemodel fit
(Browne and Cudeck, 1992).
The first CFA analysis including all 11 revised TRA items (shown under Analysis 1 in Table 1),
displayed only marginal levels of fit (X2 276.36 (44 df, p < 0.000); X2/df= 6.28, GFI= 0.908;
AGFI= 0.861; NFI = 0.853; RMSEA= 0.145). The loadings for two items (TRA10 and TRAl 1) fell
below the 0.40 lower-boundthresholdcommonly used for standardizedfactor loadings or parameter
estimates (e.g., Ford et al., 1986), but factor loadings for the other nine items fell within acceptable
levels (0.50-0.75). TRA10 and TRAl 1 were droppedand an additionalCFA analysis was conducted.
These results are shown in the column labelled Analysis 2 in Table 1. The model fit of the revised
9-item TRA scale was good (X2= 42.49 (27 df, p < 0.011); X2/df= 1.57, GFI = 0.984; AGFI= 0.970;
NFI = 0.973; RMSEA= 0.060) and factorloadingsfor these items were in the acceptablerange(0.510.76).
AdditionalCFA analyses were conductedin orderto attemptto empirically distinguishTRA from
preferencefor group work and perceived efficacy of teams constructs.In these tests, we compareda
three-factormodel (all items loading on the expected factors) with a one-factormodel (see Table 2).
Better model fit was found when the three-factormodel was estimated (AX2 = 1163.98, p < 0.000).
The three-factorsolution was also superiorto alternativetwo-factor solutions, i.e., when TRA items
loaded on a single factor with preferencefor group work items (AX2= 1032.68, p < 0.000) and perceived efficacy of teams items (Ax2 589.92, p <0.000). These resultsprovide some evidence of the
discriminantvalidity of TRA.
Copyright? 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



Table 1. Confirmatoryfactor analyses results for TRA items*

Analysis 1 Analysis 2 Analysis 3
(Time 1) (Time 1) (Time 3)


It makes sense to give rewardsto team membersbased only

on the overall performanceof the team
TRA2 A team member's rewardsshould be based only on the team's
TRA3 Teams performbetter when all team membersget the same rewards
TRA4 When working on a team, I prefer the rewardsto be based solely on
team performance
TRA5 It's not fair to give every team member the same rewards
regardlessof how each person performs(r)
TRA6 I like to be rewardedbased solely on my performance,not the
team's performance(r)
TRA7 Team memberswork hard when they are rewardedequally
TRA8 Membersof my team should share equally in the team's
successes and failures
TRA9 I exert more effort when rewardsare based solely on the
team's performance
TRAlOt I get upset when poor performersare given the same rewardsas
good team performers(r)
TRAl lt When working on a team, my rewardsshould be based solely
on my contributionto the team (r)

























Model fit indices



(p< 0.000)(p< 0.011) (p< 0.000)





factorloadingsshown.Analyses1 and2 (n= 566).Analysis3 (n= 460).

Time 3 confirmatory factor analysis

The nine TRA items were also subjectedto a CFAprocedureat Time 3. These resultsare shown in the
column labelled Analysis 3 in Table 1. Factors loadings were somewhat lower across the nine items
than those in Analysis 2, but were all above the recommended0.40 threshold. Although model fit
indices were not as good as those at Time 1, the fit was still acceptable on most indicators
(X2= 62.70 (27 df, p < 0.000); X2/df= 2.32; GFI= 0.972; AGFI= 0.937; NFI= 0.948;
RMSEA=0.072). The test-retest reliability of TRA was 0.42 (p < 0.01). The final 9-item scale
includes six cognitive items (TRA1, TRA2, TRA3, TRA5, TRA7, TRA8), two affective items
(TRA4, TRA6), and one behavioralitem (TRA9).

Hypothesis tests
Descriptive statistics and correlationsfor all study variables are shown in Table 3. The tests of the
hypotheses are shown in Table 4. The hypotheses were tested using predictorvariables from Time
1 and Time 3 TRA. A hierarchicalregression approachwas used; instructordummy variables and
Copyright( 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions

Table2. Confirmatoryfactoranalysisresultsfor TRA, preferencefor groupwork, andperceivedefficacy of teams

Factor 1

TRA9 I exert more effort when rewardsare based solely on the

team's performance
TRA2 A team member'srewardsshould be based only on the
team's performance
TRA7 Team members work hardwhen they are rewardedequally
TRA3 Teams performbetter when all team members get the same rewards
TRA6 I like to be rewardedbased solely on my performance,not the
team's performance(r)
TRA4 When working on a team, I preferthe rewardsto be based solely
on team performance
TRA5 It's not fair to give every team member the same rewards
regardlessof how each person performs(r)
TRA8 Membersof my team should share equally in the team's
successes and failures
TRAl It makes sense to give rewardsto team membersbased only on
the overall performanceof the team
I preferto work on team ratherthan individual tasks
When I have a choice, I try to work in a group instead of by myself
I personally enjoy working with others

Workingin a groupis betterthanworkingalone

Given the choice, I would ratherdo a job where I can work alone
ratherthan do a job
where I have to work with others in a group (r)
I like to interactwith others when working on projects
I preferto do my own work and let others do theirs (r)
I believe teamworkcan producebetter results than individualefforts
Teams can more thoroughlyevaluate options than any
one individualcan
Workingin teams stimulatesinnovation

Factor 2

Factor 3



*Model fit (3-factor solution): X2=397.63 (149 df; p<0.000); X2/df=2.67, GFI=0.931; AGFI=0.913; NFI=0.920;
RMSEA= 0.09.
Model fit (1-factor solution): X2= 1561.61 (152 df; p <0.000); X2/df= 10.27, GFI= 0.677; AGFI= 0.596; NFI= 0.665;

RMSEA= 0.26.

Model fit (2-factor solution, TRA items loading with preference for group work items): X2 = 1430.31 (151 df; p <0.000);
x2/df= 9.47, GFI= 0.694;AGFI= 0.614;NFI= 0.693;RMSEA= 0.18.
Model fit (2-factor solution, TRA items loading with perceived efficacy of teams items): X2= 987.55 (151 df; p <0.000);

x2/df= 6.54,GFI= 0.801;AGFI= 0.760;NFI= 0.788;RMSEA=0.12.

tn= 566.

othercontrolswere enteredon the firststep, ability andpersonalityon step 2, and qualityof priorteam
experiences and team-relatedattitudeson the final step. As Table 4 shows, ability and personality
explained 9 per cent of the variancein Time 3 TRA, while experience and attitudesexplained 6 per
cent. Hypothesis I was supported as ability was negatively related to TRA (3 = - 0.23, p < 0.01).
Quality of prior team experiences was not related to TRA (/3= 0.02, n.s.) and thus Hypothesis 2
was not supported.The two team-relatedattitudes were significant predictorsof TRA [preference
for group work (/3= 0.21, p < 0.01); perceived efficacy of teams (/ = 0.09, p < 0.05)]. Hypotheses
3 and 4 were thus supported.Hypothesis5 was supportedas external locus of control was associated
with a higherTRA (/ = 0.10, p < 0.05), but proactivepersonality(Hypothesis6) was not significantly
related (/3= - 0.03, n.s.).
Copyright(0 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions




Table 3. Descriptive statistics and zero-ordercorrelationsamong study variablesl




21.90 4.21 n/a

1. Age
2. Gender
0.39 0.49
3. Class standing
3.19 0.83
4. Team size
4.77 1.07 0.05
4.37 1.05 0.03
5. Task interdependence
6. Rewardinterdependence 38.21 22.71
2.89 0.52
7. Ability (GPA)
8. Quality of priorteam
5.22 1.46 -0.14t
9. Locus of control
3.47 0.88 - 0.10t
10. Proactivepersonality
5.15 0.62
11. Preferencefor
4.66 1.12 - 0.14t
group work
12. Perceived efficacy
5.43 0.92
of teams
4.14 0.96 -0.02
13. TRA (Time 1)
14. TRA (Time 3)
4.62 0.93 -0.05

- 0.13t n/a
0.15t n/a
0.03 (0.70)
0.25t 0.41t 0.15t n/a
0.01 -0.12t
0.23t - 0.13t -0.05
-0.07 -0.14t -0.02 -0.02


-0.07 -0.01 -0.07

- 0.09* -0.04 -0.01


- 0.10t


0.01 - 0.18t -0.05

0.08* -0.15t -0.21t
0.01 - 0.25t
0.60t -0.06




0.04 - 0.17t

0.44t -0.10*

-0.16t -0.04
- 0.22t 0.07



0.08* -0.29t
0.16t - 0.39t



0.08* -0
0.14t -0

in parenthesis
in thediagonalwhereappropriate.
*p< 0.05;tp < 0.01. One-tailed

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions

Table4. Regressionresults
Time3 TRA

Class standing
Team size

Locusof control




- 0.04

- 0.04


- 0.23t
- 0.06
- 0.08

- 0.15t
- 0.07
- 0.08

- 0.13t
- 0.07
- 0.05


- 0.31t

- 0.23t

- 0.04

- 0.03


Qualityof priorteamexperiences
Preferencefor groupwork
Perceivedefficacyof teams






A R2 Block




*p < 0.05; tp < 0.01; n = 351.

IThe set of instructordummyvariablesis denotedbyz t. Gender(male scoredhigher);class standing(1 = sophomore,2 = Junior,

3 = Senior);
locusof control(external
qualityof priorteamexperiences


The results of this study provide initial supportfor a 9-item TRA measure including evidence of its
unidimensionalityand high internal consistency reliability in two time periods separated by four
months. When examined in isolation, the nine TRA items loaded on a single factor, with acceptable
factorloadings and generallyacceptableindicationsof model fit. More important,in CFA analyses,the
TRA items could be empiricallydistinguishedfrom conceptually distinct, but related, preferencefor
groupwork and perceivedefficacy of teams items. TRA was stronglyand negativelyrelatedto a proxy
for individualability (GPA)as predicted,i.e., higherability individualspreferreceiving rewardsbased
on their own performance.Also, as predicted,other team-relatedattitudeswere significantlyrelated
and a significantrelationshipwas also found with locus of controlsuch thatexternalshad morepositive
evaluationsof team-basedrewards.Two predictionswere not supported.Proactivepersonalitywas not
a significantpredictorof TRA at Time 3, and while qualityof priorteam experiencesrelatedto TRA in
bivariatecorrelations,the relationshipbetween quality of team experiences and Time 3 TRA was not
significantwhen other team-relatedattitudeswere taken into account.
Several interestingareas for future research are possible in light of this study. Researcherscould
examine the dynamics of individualattitudesconcerning team-basedrewardsin team contexts with
varying types of tasks and interdependence.In this study, the type of task and the natureof the interdependencewere generally consistent across teams and classes, althoughthere was some variationin
the degree of interdependence.While we would argue that individualshold general evaluationsconcerning how rewardsshould be distributed,differentialreactionsto team-basedrewardsdependingon
type of task being completed and the type of task interdependence(e.g., reciprocal,pooled, or sequential) are certainlypossible. Severalresearchpathsare possible in this regard.More specific measuresof
Copyright(0 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



TRA could be developed that tap situationally-specificrewardattitudes,i.e., the type of rewardindividuals preferin differentsituationalcontexts. Our measuretaps a general evaluationin a broadcontext (work teams), but individualsmay make finer-graineddistinctions.More specific measuresmay
providemore predictivepower. Second, as Leventhalet al. (1980) discuss, rewardallocation attitudes
may vary to the extent that the goals of the team vary.Again, while individualsmay have general attitudes aboutreceiving team-basedrewards,these attitudesmay be embeddedin a system of situational
cues tied to collective goals. Third,rewardallocation attitudesmay be influencedby other individual
differencesnot includedin our study,includingone's perspectiveon self-interestand otherpersonality
Futureresearchcould also explore how TRA operatesin the predictionof individualand team performancein differentcontexts. This researchwill likely focus on the interactiveeffects between TRA
and situationalor contextualvariables.A growing body of researchhighlightsthat the main effects of
individualdifferencesin groupcontexts are not particularlystrongand individualdifferencesby situational factorinteractionsprovidemoreexplanatorypower (e.g., Wagner,1995). Developing interactive
predictionsbetween TRA and situationalvariablesmay lead to more precise theoriesof team functioning. It follows that the most powerful effects of TRA will be evidenced in interactiveeffects with
rewardcontingency variables.As such, investigationsof TRA may also be of practicaluse. Designing
and implementingcompensationand rewardsystems which caterto attitudesof currentorganizational
membersmay improveemployee motivationandperformanceandreduceundesirableworkbehaviors.
Researchon TRA could also be conductedat the group-or team-level. For example, much research
investigatesthe implicationsof diversityprofiles, both in terms of social categoryand value diversity,
on team effectiveness outcomes. It is possible that divergence of team-memberattitudesconcerning
rewards,task preference, and other specific attitudinalvariables may play a role in predictingteam
effectiveness. In this sense, TRA may have implications for the functioning of work teams and
possibly affect the behaviorof other team members.
Some of the hypotheses presentedhere may be culturebound. Kirkmanand Shaprio(1997) argue
that cultural values predict reactions to team-relatedorganizationalinterventions(see also Earley,
1989). Norms about allocation schemes are built into the familial, educational,and political systems
of society and thereforeaffect individual'sattitudesabout allocations (Leventhalet al., 1980). Thus,
while ability may negativelyrelateto TRA in the US culture,the relationshipmay be reversedin more
collectivistic cultures.In general, it seems reasonablethat differences in TRA will appearacross cultures, and that these differences should be investigated.Finally, futureresearchcould also explore the
role of gender in rewardallocation situations,as it was a consistentcorrelateof TRA in our analyses.'
The weaknesses of the study should also be highlighted.This study was an initial step in the validation of TRA and, thus, the measure should be examined across several samples and in differentcontexts. The findings(e.g., modest factor loadings on some items) may suggest thatminormodifications
or perhapsadditionalitems are needed. TRA was also developed using the tripartiteview, although
some researchershave argued that the three aspects will not necessarily apply to all attitudes (e.g.,
Brief, 1998). Futureinvestigationsshould carefullyexamine if this structureis appropriatefor reward
attitudes.While we used studentsamples, the participantsin the primarystudy were within a year of
entering the job market or graduateprogrammesand reportedhaving significant work experience.
These factors should ameliorate concerns about the development of team-relatedattitudes to some
extent. We also used GPA, a common proxy for ability, althoughGPA could be considereda measure
of motivation,or an amalgamof motivationand ability.Futureresearchcould improveon this studyby
including more direct measuresof ability such as cognitive ability or complexity. Moreover,practical
concernsdictatedthatonly a limited set of correlates,personalitycharacteristicsin particular,could be
'We thankan anonymousreviewer for many of these ideas.

Copyright( 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



examined. Futureresearchthat includes more general personalitycharacteristics(e.g., dimensions of

the five FactorModel), other team-relatedand general attitudes,and experience variablesare needed.
Additional correlates, such as Rubin and Brown's (1975) motivation orientationconstruct or social
motives identified by Deutsch (1985), could also be included in these studies. The TRA construct
described here was specific to team-basedrewards,but anotherimportantstep is to develop specific
constructsthat may better tap other types of rewards(monetaryand non-monetary)and recognition.
The proposed TRA measurewas generic in nature,i.e., we did not attemptto discriminatebetween
various types of rewardsin attemptingto measure the overall attitude.Additional samples, contexts,
situationand situationsmay allow researchersto tailor the existing scale to tap related, but more
such as
behavioralvariablesand performance,is needed to more thoroughlyestablishthe constructvalidity of
To summarize,this study reportedthe development and initial validation of a measure of TRA.
While a large literatureinvestigating aspects of team structureand task characteristicsexists, we
attemptedto fill a gap by exploring team-memberattitudesabout differentrewarddistributions.The
measuredemonstratedgenerally acceptablepsychometricpropertiesand could be empiricallydistinguished from other relatedconstructs.Some supportwas found for expected relationshipswith other
constructs.The resultshelp balance the literatureby elaboratingon a potentiallyimportantindividual
attitudinaldifference in team contexts.

The authorswish to thankJon Johnson,Matt Bowler, Scott Droege, Jon Anderson,ConsultingEditor
Linn Van Dyne, and four reviewers for helpful comments on earlier versions of this paper.

Author biographies
Jason D. Shaw is an Assistant Professor in the Gatton College of Business and Economics at the
University of Kentucky.He received his Ph.D. from the Universityof Arkansas.His currentresearch
interests include individual and organizationalconsequences of rewardsystems, work force stability
and turnover,and person-environmentcongruence issues.
Michelle K. Duffy is an Assistant Professorin the GattonCollege of Business and Economics at the
Universityof Kentucky.She received her Ph.D. from the Universityof Arkansas.Her currentresearch
interestsinclude employee stress and well-being, social undermining,and moral disengagement.
Eric M. Stark is an AssistantProfessorin the Programof Managemntat JamesMadison University.
He received his Ph.D. from the University of Arkansas.His currentresearchinterests include team
decision making and performanceand individual-differencecongruenceissues.

AlbaneseR, VanFleetDD. 1985.Rationalbehaviorin groups:thefree-riding
Academyof Management
Review 10: 244-255.

Copyright0 2001 JohnWiley& Sons,Ltd.

J. Organiz.Behav.22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions



BanduraA. 1989. Human agency in social cognitive theory.AmericanPsychologist 44: 1175-1184.

Bateman TS, CrantJM. 1993. The proactive component of organizationalbehavior:a measure and correlates.
Journal of OrganizationalBehavior 14: 103-118.
Bem DJ. 1968. Attitudes as self-descriptions: another look at the attitude-behaviorlink. In Psychological
Foundationsof Attitudes,GreenwaldAG, Brock TC, OstromTM (eds). Academic Press: New York;197-215.
Bollen KA. 1989. StructuralEquations with Latent Variables.John Wiley and Sons: New York.
Bretz RD, Judge TA. 1994. The role of humanresource systems in job applicantdecision processes. Journal of
Management20: 531-551.
Brief AP. 1998. AttitudesIn and AroundOrganizations.Sage: ThousandOaks, CA.
Browne MW, Cudeck R. 1992. Alternativeways of assessing model fit. Sociological Methodsand Research 21:
Cable DM, Judge TA. 1994. Pay preferences and job search decisions: a person-environmentfit perspective.
Personnel Psychology 47: 317-348.
Campion MA, Medsker GJ, Higgs CA. 1993. Relations between work team characteristicsand effectiveness:
implications for the design of effective work groups.Personnel Psychology 46: 823-850.
Chen Y, ChurchAH. 1993. Rewardallocation preferencesin groups and organizations.InternationalJournal of
ConflictManagement4: 25-59.
Cotton MS, Cook JL. 1982. Meta analyses and the effects of variousrewardsystems: some differentconclusions
from Johnsonet al. Psychological Bulletin 92: 176-183.
DeMatteo JS, Rush MC, SundstromE, Eby LT. 1997. Factorsrelatedto the successful implementationof teambased rewards.ACA Journal Winter: 16-28.
DeMatteo JS, Eby LT, SundstromE. 1998. Team-basedrewards:currentempirical evidence and directions for
futureresearch.Research in OrganizationalBehavior 20: 141-183.
Deutsch M. 1985. DistributiveJustice: A Social Psychological Perspective.Yale University Press: New Haven,
Duffy MK, Shaw JD, StarkEM. 2000. Performanceand satisfactionin conflicted, interdependentgroups: when
(and how) does self-esteem make a difference?Academyof ManagementJournal 43: 772-784.
Earley PC. 1989. Social loafing and collectivism: a comparisonof the United States and the People's Republic of
China.AdministrativeScience Quarterly34: 565-581.
Erez M, Earley PC. 1987. Comparativeanalysis of goal setting strategies across cultures. Journal of Applied
Psychology 72: 658-665.
Ford JK, MacCallumRC, Tait M. 1986. The applicationof exploratoryfactor analysis in applied psychology: a
critical review and analysis. Personnel Psychology 39: 291-314.
GladsteinDL. 1984. Groupsin context:a model of task groupeffectiveness.AdministrativeScience Quarterly29:
Hayes LA. 1976. The use of group contingencies for behavioral control: a review. Psychological Bulletin 83:
Hogan R, Blake RJ. 1996. Vocationalinterests:matching self-concept with the work environment.In Individual
Differences and Behavior in Organizations,MurphyKR (ed.). Jossey-Bass: San Francisco;89-144.
James WH. 1973. Internalversus external control of reinforcementas a basic variable in learning theory. In
Measures of Social Psychological Attitudes(revised edition), RobinsonJP, ShaverPR (eds). Survey Research
Center,Institutefor Social Research:Ann Arbor,MI.
KirkmanBL, ShapiroDL. 1997. The impactof culturalvalues on employee resistanceto teams:towarda model of
globalized self-managingwork team effectiveness. Academyof ManagementReview 22: 730-757.
LangerE. 1983. The Psychology of Control.Sage: Beverly Hills, CA.
LedfordGE, LawlerEE III, MohrmanSA. 1995. Rewardinnovationsin Fortune1000 companies. Compensation
and BenefitsReview 27: 76-80.
Leventhal GS, KarusaJ Jr, Fry WR. 1980. Beyond fairness: a theory of allocation preferences.In Justice and
Social Interaction,Mikula G (ed.). Springer-Verlag:New York; 167-217.
Mason ES. 1995. Genderdifferences in job satisfaction.Journal of Social Psychology 135: 143-149.
Medsker GJ, Williams LJ, Holahan PJ. 1994. A review of currentpractices for evaluating causal models in
organizationalbehavior and humanresourcesmanagementresearch.Journal of Management20: 439-464.
Miller LK, HamblinRL. 1963. Interdependence,differentialrewarding,and productivity.AmericanSociological
Review 27: 768-778.
Milkovich GT, Wigdor AK. 1991. Pay for Performance: Evaluating PerformanceAppraisal and Merit Pay.
National Academy Press: Washington,DC.

Copyright ? 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions

Mohrman AM, Mohrman SA, Lawler EE. 1992. The performance management of teams. In Performance
Measurement,Evaluation, and Incentives, Bruns Jr WJ (ed.). HarvardBusiness School Press: Boston, MA;
ParkHY, Ofori-DankwaJ, Bishop DR. 1994. Organizationaland environmentaldeterminantsof functional and
dysfunctionalturnover:practicaland researchimplications.HumanRelations 47: 353-366.
Rotter JB. 1966. Generalizedexpectancies for internalversus externalcontrol of reinforcement.Psychological
Monographs80: (1, whole no. 609).
Rubin JZ, Brown BR. 1975. The Social Psychology of Bargainingand Negotiation. Academic Press: New York.
Shamir B. 1990. Calculations, values and identities: the sources of collectivistic work motivation. Human
Relations 43: 313-332.
Shaw JD, Duffy MK, Stark EM. 2000. Interdependenceand preference for group work: main and congruence
effects on the satisfactionand performanceof group members.Journal of Management26: 259-279.
WagemanR. 1995. Interdependenceand group effectiveness. AdministrativeScience Quarterly40: 145-180.
Wagner JA III. 1995. Studies of individualism-collectivism: effects cooperation in groups. Academy of
ManagementJournal 38: 152-172.
Wagner JA III, Moch MK. 1986. Individualism-collectivism:concept and measure. Group and Organization
Studies 11: 280-303.
Witting MA, MarksG, Jones GA. 1981. Luck versus effort attributions:effect on rewardallocations to self and
other.Personalityand Social Psychology Bulletin 7: 71-78.
YamagishiT. 1988. Exit from the groupas an individualisticsolution to the free-riderproblemin the United States
and Japan.Journal of ExperimentalSocial Psychology 24: 530-542.

Copyright ? 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 903-917 (2001)

This content downloaded from on Thu, 09 Apr 2015 04:23:32 UTC
All use subject to JSTOR Terms and Conditions