conflict management strategies in workplce

Upload: sonysamsung

Post on 06-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Conflict Management Strategies in Workplce

    1/25

    A Theory-Based Measure of Conflict Management Strategies in the WorkplaceAuthor(s): Carsten K. W. De Dreu, Arne Evers, Bianca Beersma, Esther S. Kluwer, Aukje NautaReviewed work(s):Source: Journal of Organizational Behavior, Vol. 22, No. 6 (Sep., 2001), pp. 645-668Published by: John Wiley & SonsStable URL: http://www.jstor.org/stable/3649613 .

    Accessed: 19/02/2012 23:00

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    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 [email protected].

    John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access toJournal of

    Organizational Behavior.

    http://www.jstor.org/action/showPublisher?publisherCode=jwileyhttp://www.jstor.org/stable/3649613?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3649613?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=jwiley
  • 8/3/2019 Conflict Management Strategies in Workplce

    2/25

    Journalof OrganizationalBehaviorJ. Organiz.Behav.22, 645-668 (2001)DOI: 10.1002/job. 07

    A theory-based measure of conflictmanagement strategies in the workplaceCARSTEN K. W. DE DREUI*, ARNE EVERS', BIANCA BEERSMA'ESTHER S. KLUWER2 AND AUKJE NAUTA31Universityof Amsterdam,Amsterdam,The Netherlands2 UtrechtUniversity,Utrecht,TheNetherlands3 Universityof Groningen,Groningen,The Netherlands

    Summary Conflictmanagement nfluencesindividualwellbeing, groupperformanceandorganizationaleffectiveness. This researchexamined hepsychometricqualitiesof two versions of thenewlydeveloped test for conflicthandling.The lean version(Study 1 and2) includedproblemsol-ving, forcing, yielding and avoiding as distinct conflict management strategies, and theexpanded version (Study 3) also included compromising.A negotiation study (Study 1)showed substantialconvergence between self-reports,opponent-reports nd observerratedbehavior orproblemsolving, forcingandyielding,butnot foravoiding.In Study2 andStudy3 the psychometricpropertieswere examinedof the lean and the expandedversion,respec-tively.Confirmatoryactoranalysesrevealedgood to excellentpsychometricqualitiesof bothversionsof thescale. Weconcludethat hescaleis aparsimonious,lexible andvalidinstrumentto assess conflictmanagement trategiesatwork.Copyright? 2001 JohnWiley & Sons,Ltd.

    IntroductionThe effectiveness of individual employees, teams and entire organizations depends on how they man-age interpersonal conflict at work (Tjosvold, 1998). Managers spend an average of 20 per cent of theirtime managing conflict (Thomas, 1992), and evidence suggests conflict and conflict management atwork substantially influences individual, group and organizational effectiveness, as well as wellbeing,as indicated by health complaints and doctor visits (De Dreu et al., 1999; Spector and Jex, 1998).Given the importance of conflict management in organizations, it is vital to have and develop reliableand valid measurement instruments. Such instruments help researchers to obtain valid data throughself and peer-report, and practitioners to diagnose conflict management strategies at work. Unfortu-nately, the measurement instruments that have been described in the literature either suffer fromlow psychometric quality, or the psychometric quality is unknown. The present research was designed

    *Correspondenceo: CarstenK.W.De Dreu,Universityof Amsterdam,Department f Psychology,Roetersstraat 5, 1018 WBAmsterdam,The Netherlands.E-mail:[email protected]

    Copyright? 2001 JohnWiley & Sons, Ltd.Received 25 February2000Accepted 9 April2001

  • 8/3/2019 Conflict Management Strategies in Workplce

    3/25

    646 C. K.W.DEDREUETAL.to assess the psychometricqualitiesof a revisedandupdatedversionof the DutchTest for ConflictHandling(DUTCH)designedby Vande Vliert(1997). The testhas beendevelopedby Dutchscholarsbut is not necessarily applicableonly to people from the Dutch culture.As we will elaborateuponbelow, the theoretical basis for the test generalizes across cultureand the original test correspondsclosely with tests developedin the United States (Vande Vliert, 1997).As anoutline,we firstdiscussthe theoreticalbasisunderlying he instrument, ndwe discussseveralconflictmanagement ests thathave been described n the literatureandprecededthe DUTCH. Sub-sequently,wereport hreestudiesdesigned o assessthepsychometric ualitiesof theDUTCH.InStudies1and2 we examineda 'lean'versionof theDUTCHwhich measuresproblem olving, forcing,yieldingandavoiding.InStudy3 we examinedan 'expanded'versionwhich includescompromising s adistinctstrategy.Weconcludewitha discussion of some practicalandtheoretical mplicationsof the results.

    Conflict management at work: dual concern theoryConflictmanagements whatpeoplewho experienceconflictintendto do as well as whattheyactuallydo (Vande Vliert, 1997).Althoughan infinitenumberof conflictmanagement trategiesmaybe con-ceived of, conflictresearchandtheorytends to convergeon Dual ConcernTheory(PruittandRubin,1986). Dual ConcernTheoryis relatedto earlierworkby Blake andMouton(1964) and to Deutsch'sTheoryof CooperationandCompetition Deutsch,1973). Itargues hatconflictmanagements a func-tion of highorlow concernfor self, combinedwith highorlow concernfor others.As FigureI shows,high concern for self andlow concernfor othersresultsin a preference or orcing, focusedon impos-ing one's will on others.Forcing nvolves threatsandbluffs,persuasivearguments, ndpositionalcom-mitments.Low concernfor self andhighconcernfor othersresultsin a preference oryielding,whichis orientedtowardsacceptingandincorporating therswill. It involvesunilateralconcessions,uncon-ditionalpromises,andofferinghelp.Low concernfor self and othersresultsin a preference oravoid-ing, whichinvolvesreducing he importanceof the issues, andattempts o suppress hinkingabouttheissues. High concern for self andothersproducesa preference orproblemsolving, which is orientedtowardsanagreement hatsatisfiesboth own andothers'aspirationsas muchaspossible.Itinvolvesan

    High Yielding Problem Solving

    Concern (Compromising)For Other

    Low Avoiding ForcingLow HighConcernorSelf

    Figure1. Theoreticalepresentationf the five conflictmanagementtrategiess a function f concernor selfandconcern or other

    Copyright 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    4/25

    CONFLICT ANAGEMENT 647exchange of informationaboutprioritiesand preferences,showing insights, and making trade-offsbetween importantandunimportantssues.Recently, some authorshave suggestedthat intermediateconcernfor self, pairedto intermediateconcernfor others results in a preference or compromising.Some see compromisingas 'half-heartedproblem solving (e.g., Pruittand Rubin, 1986). Others,however, see it as a distinct strategythatinvolves the matchingof others'concessions,makingconditionalpromisesandthreats,andan activesearchfor a middleground(e.g., Van de Vliert, 1997). Rather hanseekingto settle this debateat thetheoretical evel, we choose to test two versionsof the DUTCH.The 'lean' versionincludesproblemsolving, forcing,yielding andavoiding(Studies 1 and2); the 'expanded'version(Study 3) includescompromisingas well. Doing so yields informationaboutthe psychometricpropertiesof two relatedmeasurement cales. In addition,Study 3 will providesome empiricalarguments or or againsttheconceptualizationof compromisingas a distinctconflictmanagementstrategy.

    Conflict management testsDual Concern Theory has received good support in field research in organizations(Blake andMouton, 1964;Thomas, 1992; Van de Vliert, 1997) as well as in experimentalresearch n the socialpsychological laboratory(Carnevaleand Pruitt, 1992; De Dreu et al., 2000). As such, the theoryprovides a solid basis for the development of instruments to assess conflict managementstrategies at work. Several such instruments exist. One of the oldest methods is the ConflictMeasurementSurvey (e.g., Kilmannand Thomas, 1977), but it has been criticized and replacedby other measures because of disappointing psychometric qualities (Landy, 1978; Thomasand Kilmann, 1978; Shockley-Zalabak, 1988). Substantial improvement was reached by thedevelopmentof the Rahim OrganizationalConflict Inventory(ROCI-2).The instrumentcomes inthree forms, one concerned with conflicts with one's supervisor (Form A), one concerned withconflicts with one's peer (Form B), and one concerned with conflicts with one's subordinates(FormC). In each case the instrumentmeasuresthe five conflict managementstrategiesdiscussedabove.

    Rahim andMagner(1995) conductedconfirmatory nalysisof data from one studentandfourpro-fessionalsamples,andshowedconvergentand discriminant alidityof the five sub-scalesof measuringconflictmanagement.Close inspectionof thegoodness-of-fit(GFI)indices in bothstudentandprofes-sional samplessuggestsmoderate itfallingbelow the standard ut-offcriterionof 0.90. Satisfactory itwasreachedonlywhen withineach sub-scale temswereaggregatedntotwo sets (RahimandMagner,1995, p. 126). Although this procedure s statisticallyand methodologicallycorrect, it makes theinstrumentunfit for applied purposes in which one wishes to quickly and unequivocally answerthe questionwhich conflict managementstrategyrespondentsuse within a particular etting. Thus,the ROCI-2 s widely used but it suffers from a numberof disadvantages ncludingless thanoptimalpsychometricproperties.To redressthese shortcomings,Vande Vliertandcolleagues developeda test for conflicthandling(EuwemaandVande Vliert,1990;JanssenandVande Vliert,1996;Vande Vliert, 1997). In interactivesessions with focus groups,new items were designed anditems derivedfrom the ROCI-2and otherinstrumentswere revised andadaptedup to a pointwhere the coefficientalphasandpatternsof inter-correlationswere sufficient to good (Vande Vliert, personalcommunication,June 14, 1999). Com-paredto the ROCI-2,the DUTCHhas fewer items (16 or 20 insteadof 28), which is importantwhenwe want to measureconflictmanagementas partof a largeorganizational urveyand need as shortameasureas possible. Also, the DUTCH leaves the hierarchicalrelationshipbetween conflict partiesunspecified,which increases the investigator's lexibility.Copyright 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    5/25

    648 C.K.W.DEDREUETAL.The problem solving (forcing) scale of the DUTCH is positively (negatively) related to inte-grative agreements in group negotiation (Beersma and De Dreu, 1999). However, we have noinformationabout the convergencebetween self and other reportsof conflict managementstrate-gies, or between self-reportsand actual conflict behavior.Study 1 was designed to fill this voidandexamined the convergentvalidityof the DUTCHby comparingself-ratingsof forcing,problemsolving, yielding and avoiding during a negotiation task with ratings by one's opponent andwith observer ratings of conflict behavior. Also, Study 1 assessed the extent to which self-reportsare susceptible to self-serving bias - the tendency to see one's own conflict managementas more constructive and less destructive than the conflict management of one's opponent(De Dreu et al., 1995).In addition,the psychometricqualitiesof the DUTCH areunknown.To fill this void, we assessedin Study 2 the psychometricqualities of the lean version of the DUTCH. Study 3 was designedto assess the psychometricqualitiesof the expandedversionof the DUTCH (includingcompromis-ing as a distinct strategy).In Studies 2 and 3, we focused on (1) the unidimensionalityof the fouror five sub-scales,(2) the discriminantvalidity,and(3) the invarianceacrossgenderand hierarchicallevel.

    STUDY 1MethodParticipantsandprocedureSeventy eight psychology students(29 males and 49 females) at the Universityof Groningenpartici-patedin this study.They were randomlyassignedto dyads.The mean age of the participantswas 22years,rangingfrom 19 to 26 years.Initialanalysesrevealedno differencesbetweenmale and femaleparticipants.Participantscame to the laboratory n groupsof four andcare was takenthatparticipantsn onesessionwereunacquainted.As in ourpastresearch De Dreuet al., 1998, 1995- Study3), participantswereseated n separate ooms and had 10 minutes o readtheirroleinstructionssee below).Theywereto imaginethatthey were in the describedsituation,and to preparea conversationwith anotherparti-cipant. Hereafter,participantswere pairedand seated in separaterooms where they engaged in anaudiotapednegotiation(see below). After 15 minutes,the experimenter ntered he room,andhandedout a questionnaire.Upon completionof the questionnaire,participantsweredebriefedand dismissed.

    NegotiationtaskThe study employed a role playingmethodologythatwas developedin a pilot studyand was basedon our past research(e.g., De Dreu et al., 1998, 1995 - Study 3). Participantswere providedwithbackground nformation n which a conflict situationwas describedand then freely interactedwitheach otherto discuss the issue. The instructionsstated thatthe participantworkstogetherwith a fel-low studenton a researchproject.The originalscheduleof activities for the next week is that bothstudentsdo statisticalanalyses togetherbutthatthis originalscheduleis no longerapplicableand that

    Copyright 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    6/25

    CONFLICT ANAGEMENT 649they need to reach an agreementabout how to change the schedule of activities. Thus, memberswithin one dyad needed to reach an agreementaboutwho would do what kind of activity the nextweek. Throughinstructions,partieswere given opposed preferencesfor certainactivities and theywere given arguments hey could use to defend their particularpreferences.Participantswere told,finally,thattheyhad 15 minutesto discuss theiropposingpreferencesandto reachanagreementon anew schedule of activities. (In contrast to some of our past research,participantswere not given apay-off schedule indicatingthe value of a particularagreement.Rather,they were told what agree-ments they preferredand what agreementsthey did not prefer.As a result, the study did not yieldobjectiveperformancedata).

    DependentvariablesSelf-report and opponent-report measuresIn thepost-negotiationquestionnaire,ownconflict managementwas measuredwith the DUTCH.Theinstrument omprised16 items, with four(randomlyordered) temsperconflictmanagement trategy(forcing, problem solving, yielding, andavoiding;see AppendixA). Questionscould be answeredon5-pointscales, with 1 = not at all, to 5 = very much).Cronbach'salphaswere0.86 for forcing,0.81for problemsolving, 0.71 for yielding,and0.69 for avoiding.In addition,we askedparticipantso filloutthe DUTCHwithregard o theiropponent'sconflictmanagement.Cronbach'salphaswere0.83 forforcing,0.82 forproblemsolving, 0.73 foryielding,and0.67 foravoiding.The order n whichown andothers'conflictmanagementwas assessed was variedsystematically.Because it hadno effects it is notdiscussedany further.Observer measuresThe negotiationswere audiotapedand subsequentlycoded by two trainedobserversunawareof thegoals of the study.The trainedobserversindependentlyratedthe conflict behaviorsof each of thetwo partieswithina dyad. Using electronicbeeps, the 38 audiotapedconversationswere divided intotime samplesof two minutes(n = 212 time samples,M = 5.4 time samplesperdyad). Everytwo-minuteinterval,the observers rated fourconflictbehaviors(problemsolving, forcing, yielding, andavoiding;1 = demonstratednot at all, to 4 = demonstrated o a great extent).Inter-observer eli-abilitieswere sufficientto good, with Cronbach'salphasrangingfrom 0.61 to 0.81 for forcing.Twoexceptionswere the interrater eliabilitiesfor yielding (a = 0.53) andfor avoiding(a = 0.32). Anexplanationis that both constructshad low variance (SD = 0.49 and SD = 0.16 respectively).When looking at absolutereliability,it appearedthat 141 out of 212 times (67per cent) observersagreedupon yielding and that 200 out of 212 times (94 per cent) observersagreed upon avoiding(Cohen'sKs > 0.67). We conclude that interrater eliabilitiesfor the four scales were sufficienttogood.

    ResultsWithineachdyadwe had observerratings,andself and otherreports or eachparty(further eferred oas PartyA andPartyB) on each of the foursub-scalesfromthe DUTCH.Sinceparties nteracted heirdata were interdependent nd we decided to analysethe data foreach party ndependentlyandto treatdataforPartyB as a replicationof data forPartyA. This is a conservativeapproachbecauseit reduces

    Copyright ) 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    7/25

    650 C. K. W. DE DREUETAL.ourdegreesof freedomby half. It is, however, ustifiedin lightof the fact that in the negotiationPartyA andPartyB were given differentpositionswith differentarguments.

    Convergentand divergentvalidityThe upperpartof Table1a gives themeans,standarddeviations,andinter-correlationsetweenobser-verratingsof PartyA andopponent-ratings f PartyA on the one hand,andself-reportsby PartyA onthe other.The lowerpartof Table la gives the same informationwith regard o PartyB. Before dis-cussing observedcorrelations,we wish to emphasizethatmoderatelynegativeormoderatelypositivecorrelationsmay be expected,theoretically,betweendifferentsub-scalesbecausetheybothload highon one dimension(e.g., concernfor self) and low on anotherdimension(e.g., concernfor self). Forexample, problemsolving may be positively correlatedwith yielding (bothshare a high concernforother),while forcingandyieldingmaybe expectedto correlatenegatively(theydiffer bothonconcernforself andconcernforother).Inotherwords,we expectedstrongpositivecorrelations n thediagonalin Table la, andsometimesmoderatelynegative,or moderatelypositivecorrelationsoff the diagonalin Table la.

    As can be seen in Table la, self-reported orcingwas positivelyandstronglycorrelatedwithobser-verratingsof forcingandwith opponent-ratings f forcing.Moreover,self-reportedorcingwas nega-tively relatedwith the otherconflictmanagementstrategiesratedby observersor the opponent.Thissuggeststhatself-reportedorcinghashighconvergentanddivergentvalidity.A similarbut somewhatweakerpatternemergedforproblemsolving.Again, self-reportedproblemsolvinghadstrongpositivecorrelationswith observerratingsandopponentratingsof problemsolving, andpositivebut weakercorrelationswithyielding.Interestingly,problemsolving andavoidingwerepositively,butnot signif-icantly,correlated.This may be understood n termsof Dual ConcernTheory,whereproblemsolvingis the result of high dualconcern,andavoidingthe resultof low dualconcern.Thus,problemsolvingtends to have good convergentandacceptabledivergentvalidity.Withregard o self-reported ielding,data aresupportiveof convergentvalidity, n thatself-reportedyielding is positivelycorrelatedwith observerratingsandopponentratingsof yieldingbehavior.Oneexceptionto this is thatself-reportedyielding correlatedas positivewithobserverratingsforyieldingas with observerratings orproblemsolving, suggestingthatobservershavedifficultyseparatingmereconcession makingfrom problemsolving behavior.This can be understoodwhen we realize thatinnegotiation,problemsolving also involves logrolling- makingsmallconcessionson importantssuesand largeconcessions on unimportantssues.Table1ashows, finally,thatavoidinghadpoor convergentvalidity.BothforPartyA andforPartyBdata,self-reportedavoidingdid not correlatewithobserverratingsandwithopponentratingsof avoid-ing. Most likely, the lack of convergentanddivergentvalidityis due to the factthatavoidingbehavioris difficultto observefrom audiotapedconversation.We return o this in the ConclusionandGeneralDiscussion section.Besides correlationsbetweenself-reportandopponent-report nd observerreport, t may be usefulto examine correlationsbetween the sub-scales for self-reports.Table lb providesthese. As can beseen, the patternof intercorrelations losely resembles those reported n Table 1a. For both PartyAandB, problem solving andyielding, problemsolving andavoiding,andyielding andavoidingwerenot related. For both PartyA and PartyB, forcing and problemsolving were negativelyand signi-ficantly related, and forcing and yielding were negatively but not significantlyrelated. Somewhattroublesome is the strongnegative correlationbetween forcing and avoiding for the PartyA data,which were notreplicated n the PartyB data,and was not present n Table 1a either.This mayreflectmeasurementproblemsrelatedto the avoidingsub-scale.

    Copyright ) 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    8/25

    ?

    00

    o

    00

    t'

    ?s

    t"

    .,e2tt

    Table la. Means, standarddeviations for self-reportedconflict management,and correlationsbetween self-report,obconflict behaviors;partyA (top panel) andPartyB (bottompanel)M SD Forcing(Ob) Forcing(Op) ProbSol(Ob) ProbSol(Op) Yield(Ob

    Forcing 3.28 1.36 0.72* 0.33t - 0.55* - 0.45* - 0.38tProblemSolving 3.22 1.24 - 0.45* - 0.36t 0.45* 0.47* 0.17Yielding 2.97 0.97 - 0.03 - 0.41* 0.36t 0.20 0.38tAvoiding 1.72 0.96 - 0.40* -0.11 0.33t 0.22 0.25Forcing 2.90 1.41 0.41* 0.31t - 0.38t - 0.22 - 0.31ProblemSolving 3.55 1.43 -0.18 -0.43* 0.42* 0.54* 0.17Yielding 2.97 1.05 - 0.43* - 0.46* 0.46* 0.35t 0.39tAvoiding 1.91 1.07 0.01 -0.23 -0.17 -0.06 -0.25Note:*p< 0.01;t p < 0.05 (two-tailed,with n = 38; (Ob)= observerratings; (Op)= opponentratings.

  • 8/3/2019 Conflict Management Strategies in Workplce

    9/25

    652 C.K. W. DE DREUETAL.Table b. Correlationsmong elf-reportonflictmanagementtrategiesorPartyA (above hediagonal) ndPartyB (below hediagonal)

    Forcing Problemolving Yielding AvoidingForcing - - 0.47* - 0.23 - 0.61*Problemsolving - 0.54* - 0.29 0.04Yielding - 0.32 0.35 - 0.29Avoiding 0.08 -0.16 0.13Note.*p 0.001,withn= 38.

    Table2. Means orconflictmanagements a function f target fjudgmentConflictmanagement

    Forcing Problemolving Yielding AvoidingTarget f JudgmentSelf 3.09*t 3.38* 2.97t 1.81lOther 3.12*t 2.98t 2.91 1.82'Note:*,t,l eanshat onot harehe ameuperscriptiffertp< 0.01accordingopaired-testsdf=37).

    Self-servingbias in self-reportsof conflictmanagement trategiesTo assess self-servingbias in self-reported onflictmanagementwe conductedanAnalysis of Variancewith Targetof Judgment(self or opponent)and Conflict Management(forcing, problem solving,yielding, avoiding)as within-participantariables.Resultsrevealedsignificantmaineffects forTargetof Judgment,F(1, 37) = 66.11, p < 0.001, andfor ConflictManagement,F(3, 35) = 11.50,p

  • 8/3/2019 Conflict Management Strategies in Workplce

    10/25

    CONFLICT ANAGEMENT 653ContextualSidebar

    Organizational Context for Study 2OrganizationThe site concerns one of the most profitablebusiness units of a largemanufacturing rganizationproducingheavymachinery.At thetime of thestudy(1999), themanufacturing rganization's tockvalue is substantiallyundervalued ndthe possibilityfor take-over s a continuing ssue during hetimeof thestudy.Threeyearsbeforethestudywasconducted,alargerestructuringperationhadbeenimplemented,nvolvingatransition rom ndividual-basedo team-basedpracticesalongwithdown-sizing andreductionof thenumberof management taff.EmployeesEmployees n the firmareunionized.Theunion contractpermitteda limitednumberof sickdaysforwhich the employee receivedfull pay - unlimited additionalsick days (upto two years)could betaken at 70 per cent salary, provideda doctor's certificatewas provided.The restructuringpro-grammehad resulted in a numberof employees workingin new positions that were consideredless interestingandproviding ower status. This was particularlyrue for a small numberof main-tenancestaffandoffshoreengineerswho,because of theirfrequentabsencewerenot included n thestudy.MoraleEmployeestraditionallyderiveda strongsense of prideworkingfor the company,whichprovidedthemwitha strong dentityandrelativelygood workingconditions,pay and benefitspackages.Theeconomicchangesof thenineties,makingold industry ess interesting o workfor,undermined hissense of prideandthe restructuring rogrammenduced some unrestand uncertaintiesas well asrole ambiguitiesandsense of increasedworkpressures.

    MethodSampleRespondentswereemployeesof a companyspecialized n thedevelopmentandconstruction f foodpro-cessing systems.On the initiativeof bothmanagementand laborunionsa projectwithregard o workstresswas started.The DUTCHwas included n the instrumento assessthe level of stressandseveralrelatedconcepts.All employees(includingemployeesfrom the generalmanagementandfinancialandsalesdepartments, s well as employeesin productionor technicalpositions)wereasked to participate.Out of 364 employees308 returned hequestionnaireresponserate = 85percent). Sevenquestion-naireswere not included in the analysesbecause of one or more missings. Mean age and length ofappointment f therespondentswere43.91 (SD = 10.41)and 21.32 (SD - 13.01)years,respectively.Eleven per cent of the respondentswere female.

    Copyright 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    11/25

    654 C. K.W.DE DREUETAL.ProcedureThe DUTCH was partof a booklet whichcontained n all about250 items. Theotheritemsdealtwithsources of stress,health, work satisfaction,control, etc. The items of the DUTCH were in randomorder.A researchassistant nvitedemployees to participate n the study,andan accompanyingcoverletter from the firm'sCEO emphasizedthe importanceof participatingas well as its voluntaryandanonymouscharacter.The cover letterfurtherexplainedthe purposeof the study,which was to geta better understandingof the way employees work together, experience their work environment(e.g., technicalequipment, nternalcommunications) o be ableto improvethe situationwhereneces-sary.The researchassistantadministered he questionnairesn small groups(it was sent to the homeaddressof employeeswhowere notable to attend hesegroupsessions).Thequestionnaireswere filledin anonymouslyandcollected by the researchassistant.

    Analysis and ResultsUnidimensionality nd interrelationsof thefour sub-scalesThe internalconsistencyreliabilitycoefficients(coefficienta) were computed or the fourscales.Theminimalacceptable evel of alphawas set to be 0.60. (Forteststhatareprimarilymeant o studygroupsa reliabilitycoefficientof between 0.60 and 0.70 qualifiesas 'sufficient'and a coefficientof 0.70 orhigher as 'good' [Bartram,1997;Evers, et al., 2000]). Since a high alphais no guarantee or unidi-mensionality(CrockerandAlgina, 1986), LISRELanalyseswere conductedto check for unidimen-sionality.By means of this analysiswe checkedfor unwantedhighcorrelationsbetweenpairsof itemsthatwere too similarin content,that is items that cause an 'unfair'increasein alpha.We used the fitindices of X2(divided by df), the adjustedgoodness-of-fitindex (AGFI) and root mean squareofapproximation RMSEA), which are important or scale development(Bentler, 1990; Schumackerand Lomax, 1996). A low x2 (when judged against degrees of freedom) is suggestive of good fit.An AGFI above 0.90 is indicativeof good fit and above 0.95 of excellent fit. The RMSEAresidualsshouldbe about0.05 or smaller for a good fit.Table3 reports he basic psychometricpropertiesof the DUTCHscales. The internalconsistenciescan be qualifiedas sufficient,and,except for two RMSEAfigureswhicharesomewhat oo high,all fitindices show good fit. The AGFIvalues reveal excellent fit. Apparently he items fit well into theirrespective scales and the high fit indices and low RMSEA values show that no unwantedoverlapbetweenitems within a scale is detected. Fromthese datawe concludethateach scale separatelydeli-vers reliableinformationwith respectto unidimensionallymeasuredconstructs.

    Table3. Internalconsistenciesand LISRELfit indices of the DUTCHscalesa x2 df AGFI RMSEA

    Yielding 0.65 1.95 2 0.99 0.00Dominating 0.83 5.57 2 0.96 0.07Collaborating 0.73 1.32 2 0.99 0.00Avoiding 0.64 5.76 2 0.95 0.08

    Copyright0 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    12/25

    CONFLICT ANAGEMENT 655Table 4. Goodness-of-fit ndices for one-, two-, three-,and four-factormodelsModel 2 df AGFI RMSEANull model 1316.05* 120 0.54 0.21One-factormodela 836.78* 104 0.59 0.18Two-factormodelb 552.14* 103 0.72 0.13Three-factormodelc 356.42* 101 0.81 0.10Four-factormodeld 233.08* 98 0.87 0.07Note.AGFI adjustedoodness-of-fitndex;RMSEA= rootmean quarerror f approximation.*p< 0.001.aDifferencene-factor ndnullmodel:X2(16)= 479.27.*bDifferencewoandone-factormodel:X2(1) 284.64.*cDifferencethreeandtwo-factormodel:X2(2) = 195.72.*dDifferenceourand hree-factorodel:X2(3)= 123.34.*

    The next step was to test the proposedfour-factormodel by means of a maximum likelihoodLISRELanalysisof the covariancematrixof the 16 DUTCH items. In this model each of the itemswas allowed to load on its associated factoronly, and the four factors (representing he four scales)were not allowedto correlate.The same fit indices as mentionedabovewere used. FollowingRahimandMagner(1995), we also computedthe GFIfor the independentor null-modeland one- throughfour-factormodels.The two-factormodel grouped(a) problemsolving andyielding;and (b) forcingandavoiding(cf. Deutsch, 1973). In the three-factormodel only the items of the sub-scalesavoidingandyielding were grouped ogether,whereasthe remaining wo factorscomprisedthe same items asin the four-factormodel (cf., Putnamand Wilson, 1982). Table4 summarizesthe measuresof theGFI for the null- and one- throughfour-factormodels. All measuresshow a gradualincrease, andthe four-factormodel is superior o all other models. The X2/df-ratios good; however,the AGFIisa little bit too low and the RMSEA is a little bit too high. Neverthelessthe values are indicativeofa sufficient o good fit.Inall, theresultsshow that thefour-factormodelconstitutesa good representa-tion of the interrelationsamongthe 16 items of the DUTCH.

    DiscriminantvalidityThe thirdstep in the analyses was to assess the discriminantvalidity of the scales of the DUTCHby computingthe correlationsbetween the latent factors.The values on the diagonalwere set to 1(the completely standardizedsolution). Table 5 shows the factor intercorrelations or the totalgroup.The resultssupport he discriminant alidityof the scales, in thatthe intercorrelations etweenscales are generally low (the correlationof 0.33 between yielding and avoiding being the highestcorrelation).

    Table5. Factorntercorrelationatrixn= 301)DUTCH-factors 1 2 3 41. Yielding2. Dominating 0.12(0.07)3. Collaborating 0.12 (0.11) 0.00 (- 0.02)4. Avoiding 0.33 (0.23) 0.13 (0.14) - 0.23 (- 0.08)Note: ndicesn brackets recorrelationsetweenmanifest ariables.

    Copyright 0 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    13/25

    656 C.K.W. DE DREUETAL.STUDY 3Study1 and 2 revealedthatthe leanversion of the DUTCHhasacceptable o good psychometricprop-erties.As mentioned at the outset,however,some scholarsarguethatin addition o the fourstrategiesidentifiedin Dual ConcernTheory,compromisingshould be consideredas a distinct strategyratherthan as a 'lazy form of problemsolving' (cf. Pruitt and Rubin, 1986; Vande Vliert, 1997). If true,we shouldbe able to measurecompromising,and confirmatoryactoranalyses shouldrevealbetterfitfor a five-factorrather hana four-factormodel. Inaddition,multidimensional calingshouldreveala patternof inter-correlationshatfitsa two-dimensional pacewithforcing, problemsolving,yieldingandavoiding occupyingcornerpositionsin a 2 x 2 spaceandcompromisingoccupyinga centreposi-tion (cf., Figure 1). To examine these issues, Study3 examinedthe psychometricpropertiesof a five-scale version of the DUTCH. The instrument s identical to the one used in Study 1 and 2 but alsoincludes four items to measurecompromising.

    MethodSampleData were collected from a heterogeneoussampleof professionals(n= 2403). Mean age and workexperienceof therespondentswere 34.14 (SD = 8.41) and 5.52 (SD= 5.73) years,respectively.Thirtyper cent of the respondentswere female.

    Data gathering procedureThe DUTCHwas placedon the web-site of the Dutchweekly magazineIntermediair publishedbyElsevier/ Reed).The magazineis targeted owardshighereducatedprofessionals n the fieldsof man-agement,consulting, personnelcounselling,financialadministrationndcontrolling,andlaw. Its aimsinclude providing informationabout careerpossibilities and updateson technological and humanresourcedevelopments n variousprofessionalfields. The web-site has been installed to support heprintedpaperversion of the magazine,and containsvariouskinds of information, ncludingseveraltestsdesignedto screenone's owntendencies andpolicies.TheDUTCHwaspartof this seriesof tests.Respondentswere askedto answerseveralbackgroundquestionstapping ntotheirdemographicchar-acteristics and their work situation. Subsequently, they were presented with the 20 items of theDUTCH (four items measuringforcing, problem solving, compromising,yielding, and avoiding;see AppendixA). Questionswerepresented n a randomorder. nbetweenDecember1998 andMarch1999, 2403 respondentscompletedthe questionnaire.We screenedthe answers(including biographi-cal data)to eliminaterespondents hatfilled out the questionnairemore than once. This reduced thesample size to 2182.MeasuresThe backgroundquestions assessed gender and age. In addition,respondentswere asked whetherthey experiencedmost conflicts in their work situationwith (1) theirsuperiors), (2) theirpeers and

    Copyright 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    14/25

    CONFLICTANAGEMENT 57colleagues,(3) theirsubordinate(s), r (4) other.Subsequently, espondentswere asked 'howtheytendto deal with conflict in theirdaily worksituations.'The 20 items of the DUTCHwere presented n arandomorder.Whenrespondentshadansweredall questions,theywereprovidedwith a shortdescrip-tionof the fiveconflictmanagement trategies respondents ould notenterthis site whentheyhad notanswered all questions).This descriptionalso contained some referencesfor furtherreading,in caserespondentswereinterested n learningmoreaboutconflict and conflictmanagementn organizations.

    Analysis and ResultsUnidimensionality nd interrelationsof thefive sub-scalesTable6 reports he basicpsychometricpropertiesof theDUTCHsub-scales.Theinternalconsistenciescan be qualifiedas sufficient,andexcept for one X2/df-ratio, ll fit indices show good fit. The AGFIvalues revealexcellent fit. Apparently he items fit well into theirrespectivescales and no unwantedoverlapbetween items within a scale is detected.From these datawe concludethat each scale sepa-ratelydeliversreliableinformationwith respectto unidimensionallymeasuredconstructs.The next step was to test the proposedfive-factormodel by meansof a maximum ikelihood LIS-REL analysis of the covariancematrixof the 20 DUTCHitems. In this model each of the items wasallowed to load on its associatedfactoronly,andthefive factors(representinghefive scales)werenotallowed to correlate.In addition o the models computed n Study2 we computeda five-factormodelwhich includedcompromisingas a distinctstrategy.Table7 summarizes he measuresof the GFI forTable6. InternalonsistenciesndLISRELit indicesof the DUTCH cales

    a X2 df AGFI RMSEAYielding 0.65 0.44 2 1.00 0.00Compromising 0.66 8.49 2 0.99 0.04Forcing 0.70 1.73 2 1.00 0.00Problemolving 0.68 19.50t 2 0.98 0.06Avoiding 0.73 9.25* 2 0.99 0.04*p< 0.01; tp < 0.001.Table7. Goodness-of-fitndices orone-,two-, three-,our-,and ive-factormodelsModel X2 df AGFI RMSEANull model 9513.84** 190 0.55 0.19One-factormodela 5798.63** 170 0.64 0.16Two-factormodelb 4358.11** 169 0.74 0.12Three-factormodelc 2752.42** 167 0.83 0.10Four-factormodeld 1980.36** 164 0.87 0.08Five-factormodele 1214.77** 160 0.93 0.05Note. AGFI= AdjustedGoodness-of-FitIndex;RMSEA = Root Mean SquareErrorof Approximation.*p < 0.001.aDifferenceone-factor and null model:X2(20) 3815.21.*bDifference wo and one-factormodel:X2(1)= 1440.52.*cDifferencethree andtwo-factormodel:X2(2)= 1605.69.*dDifference four and three-factor model: 2(3)= 772.06.*eDifference ourand five-factormodel:X'(4) = 765.59.*

    Copyright 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    15/25

    658 C. K. W. DE DREU ETAL.the null-andone- through ive-factormodels.All measuresshowa gradualncrease,andthefive-factormodel is superior o all othermodels. The AGFIandRMSEAvaluesindicatea good fit.AlthoughtheX2/df-ratio s too high, our very large sample size can be held responsible for this high value of X2,since X2 s dependenton samplesize. A reanalysiswitha randomsampleof 1000respondents esultedin a decreaseof x2to 556, whereasthe other ndices remained he same. This shows that thefive-factormodel constitutesa good representation f the interrelationsamongthe 20 items of the DUTCH.

    DiscriminantvalidityThe thirdstep in the analyseswas to assess the discriminantvalidityof the scales of the DUTCHbycomputingthe correlationsbetween the latent factors.The values on the diagonalwere set to 1 (thecompletely standardized olution).Tables8 and 9 show the factorintercorrelationsor the total andgendergroups,andforthegroupsthat differ in hierarchicalevel of theconflictopponentrespectively.Over all sub-samples he correlations how a consistentpattern.Althoughthere aresome fluctuations,theserarelyexceed 0.10. The resultssupport he discriminant alidityof the scales,as theintercorrela-tions betweenscales aregenerally ow (the correlationsbetweenyieldingandcompromising,yieldingandavoiding,andcompromisingandproblemsolvingaremoderate,butforeach scale enoughspecificvarianceremains).

    Invarianceacross genderand hierarchicallevelThe fourthstepin the analyseswas to examine the invarianceof the five-factormodel of the DUTCHacrossdifferentgroups.Two analyses were performed:one for gender,and one for the hierarchicallevel of the conflictopponent.For eachanalysisthecovariancematrices orthe 20 DUTCHitemswerecomputed.Weverifiedwhetheragewas related o conflictmanagement,andresultsshowed it was not.Table 8. FactorIntercorrelationmatrix or total and gender groupsDUTCH-factors 1 2 3 4 5Totalgroup (n= 2182)1. Yielding2. Compromising 0.41 (0.20)3. Forcing -0.28 ( - 0.15) 0.02 (0.04)4. Problemsolving 0.19 (0.10) 0.50 (0.33) 0.06 (0.08)5. Avoiding 0.51 (0.39) 0.22 (0.20) - 0.21 (- 0.15) - 0.20 (-0.11)Men (n = 1543)1. Yielding2. Compromising 0.37 (0.23)3. Forcing -0.26 (-0.16) 0.08 (0.08)4. Problemsolving 0.19 (0.11) 0.52 (0.35) 0.12 (0.11) -5. Avoiding 0.51 (0.39) 0.20 (0.19) - 0.21 (- 0.16) - 0.18 (-0.10)Women n= 639)1. Yielding2. Compromising 0.50 (0.30) -3. Forcing - 0.35 ( - 0.20) - 0.12 (- 0.04) -4. Problemsolving 0.20 (0.07) 0.46 (0.30) - 0.06 (0.00) -5. Avoiding 0.51 (0.39) 0.27 (0.25) - 0.21 (- 0.14) - 0.24 (- 0.13)Note: ndices nbracketsrecorrelationsetweenmanifest ariables.

    Copyright 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    16/25

    CONFLICTANAGEMENT 59Table9. Factorntercorrelationatrixorhierarchicalelation roupsDUTCH-factors 1 2 3 4 5Conflictwithsuperior(n = 564)1. Yielding2. Compromising 0.46(0.31)3. Forcing - 0.17(- 0.09) 0.15(0.14)4. Problemolving 0.22(0.12) 0.43(0.29) 0.06(0.08)5. Avoiding 0.50(0.38) 0.24(0.21) - 0.18(-0.13) - 0.17(-0.10)Conflictwithcolleague (n= 736)1. Yielding2. Compromising 0.40(0.23) -3. Forcing - 0.41 ( - 0.22) - 0.07 (0.00)4. Problemolving 0.03(0.01) 0.50(0.34) 0.07(0.08) -5. Avoiding 0.62 (0.43) 0.22 (0.20) - 0.22 ( - 0.16) - 0.28 ( - 0.16)Conflictwithsubordinate n = 310)1. Yielding2. Compromising 0.51(0.33) -3. Forcing - 0.40( - 0.27) - 0.16( - 0.08) -4. Problemolving 0.33(0.18) 0.54(0.36) - 0.09(-0.02) -5. Avoiding 0.47(0.38) 0.28(0.25) -0.19 (-0.14) -0.05 (-0.01)Note: ndicesnbracketsre orrelationsetween anifestariables.For the totalsample,the correlationwithage was r = 0.06, - 0.03, - 0.08, 0.06, and 0.05 for avoid-ing, dominating,compromising,yieldingandproblemsolving, respectively.Further nalysesrevealedno differencesbetween male andfemale respondents.We estimatedsix modelsusingLISRELandcompared hemsequentially. n Model 1 thepatternoffactorloadingsis held invariant. n Model 2 the patternof factorloadings,andthe factorloadingsareheld invariant. nModel3 thepatternof factor oadings,the factor oadings,andthe iteminterceptsareheld invariant. n Model 4 the patternof factorloadings,the factorloadings,the item intercepts,andthe errorsareheld invariant. n Model 5 the patternof factorloadings, the factorloadings,the itemintercepts, he errors,and the variances/covariances re held invariant. nModel 6, finally,thepatternof factor oadings,the factor oadings,theitemintercepts, heerrors, hevariances/covariances,ndthemean are held invariant.Besides 2/ df, and the RMSEA for these analysesthe GFI- and Akaike'sinformationcriterion(AIC) indices are reported.The interpretation f the GFI index is similar toAGFI.AIC is a badness-of-fit ndicator,with small values indicatinggood fits andlargevalues poorones. It is addedbecause thedifference n AIC whenimposingmorerestrictionsgives additionalnfor-mation about heparsimonyof themodel.DecreasingX2by solvingformoreparameterswill onlybe abenefit if X2is decreasedby more than 2.0 for each parameteradded. If this is the case AIC willdecrease as well (Loehlin, 1998). Tables 10 and 11 presentthe results with respect to the analysisof the factorinvariance.Table 10 shows the results for gender,andTable 11 for the hierarchicalevelof the conflict opponent. (Participantswere allowed to indicate whether they primarily hadconflicts with their superior(s), heircolleague(s), their subordinate(s),or 'other.'Those classifyingthemselvesin the lattercategoryareleft out of these analyses.)Note thatthe first three models in thisanalysis are the most critical. The difference in x2 between Model 2 and Model 1 was significantforgender, ndicating hat hehypothesisof equalfactor oadingsformen andwomenmust be rejected.However,the magnitudeof the differences in Table 10 suggests this is not a severe problem.Thetests for the differences of all other consecutive models were negative, implying equal itemintercepts,errors,variances/covariances,nd meansfor menand women.As AICgraduallydecreases(except for the step from Model 1 to Model 2), one can conclude that the more restrictionsare

    Copyright 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2,645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    17/25

    660 C.K. W.DE DREUETAL.Table 10. InvarianceanalysisacrossgenderModel AIC GFI X2 df RMSEA1. Equalfactorpattern 1751.39 0.92 1394.94* 320 0.062. Model 1 + equal factorloadingst 1752.72 0.92 1427.50* 335 0.063. Model 2 + equal item interceptst 1731.14 0.92 1436.96* 350 0.064. Model 3 + equal errors? 1722.97 0.92 1464.18* 370 0.055. Model 4 + equal variances/covariances1 1710.95 0.92 1483.49* 385 0.056. Model 5 + equal means: 1706.79 0.92 1490.16* 390 0.05Note. Male= 1543; female= 639. AIC= Akaike'sinformationcriterion;GFI= goodness-of-fit;RMSEA = root mean squareerror f approximation.*p< 0.001.TModel2 - Model1:X2(15)= 32.56(p< 0.01).tModel 3 - model 2: X2(15)= 9.46 (p

  • 8/3/2019 Conflict Management Strategies in Workplce

    18/25

    CONFLICT ANAGEMENT 6611.5

    ProblemSolvingo

    0.5Yielding oo CompromisingConcernfor Other

    -0.5

    Avoiding

    Forcing-1.5-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

    Concern or SelfFigure2. Empirical epresentationf the fiveconflictmanagementtrategiess a function f two dimensions(concernorself andconcern orother)basis of the Dual ConcernTheory (see Figure 1). To answer this question, we conducteda multi-dimensionalscaling of the data. The purposeof this procedure s to find a configurationof pointswhose Euclideanoutputdistancesreflect as closely as possiblethe rankorderof inputdissimilarities.Assuming the data fit one dimension (e.g., competition-cooperation)yielded an unsatisfactoryfit(Young'sS = 0.19, with R2 = 0.87). Assumingtwo dimensions (i.e., concern for one's own out-comes, and concern for other)resultsrevealedthatYoung'sstress index was 0.00251 (0.00147 forthe matrix),with R2 = 0.99, which is both very good. The solution is displayed in Figure 2. Ascan be seen, compromising ends to be in the middle,while forcing,avoiding,problem solving andyieldingtendtowards he cornersof the 2 x 2 matrix.It should be mentioned,however,thatyieldingis too close to avoiding.Nevertheless,the patternof (Euclidean)distancestends to corroborate hetheoreticalpatterndepictedin Figure 1.

    Conclusionsand GeneralDiscussionTo studyconflictmanagementstrategies n the workplace,reliableandvalid instrumentsarecrucial.Unfortunately,nstruments hatsurveyin the literatureandareused by scientistsandpractitionersoassess (preferences or) conflictmanagement trategiesmay be criticizedon methodologicalor prac-tical grounds.The currentresearchbuilds upon earlierworkby Vande Vliert (1997) andwas con-cerned with the psychometricqualities in the Dutch Test for Conflict Handling (DUTCH). Wesoughtanswers o two questions.First,do self-reportsof conflictmanagement trategiesconvergewith

    Copyright 2001JohnWiley&Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    19/25

    662 C. K. W. DE DREU ETAL.

    other i.e.,opponent)eports nddoself-reportsonvergewithactual onflict ehavior? econd,whatare hepsychometricualities f theDUTCH?Weexaminedwoversions f theDUTCH.n Studies1and2 weuseda 16-itemnstrumenthatmeasures roblemolving, orcing, ielding, ndavoiding.nStudy3 we examined 20-itemnstrumenthatmeasuredompromisingn additiono theother ourconflictmanagementtrategies.

    Theoretical mplicationsStudy1showed onvergenceetween elf-reportsf problemolving, orcing, ndyielding ntheonehand,andopponent-reportsf thesestrategies ntheother.Moreover, tudy1revealedheexpectedcorrelationsetween elf-reportsfforcing ndproblemolvingandobserveratings f conflict eha-viorduringnegotiation. ess supportiveesultswere obtainedor avoiding.Althoughhe psycho-metricqualitiesof thesub-scale oravoidingweregood,self-reportsf avoidingdidnotconvergewithreports ytheconflict pponent,r withobserveratings.A possible xplanations thatavoidingis themoreambiguoustrategy pen o multiple ttributions.orexample, conflictpartywhocon-sistentlydownplayshe mportancef the conflictssuemaydothis nordero avoid he ssue and oreducenteractiono a minimum.Theopponent, owever,mayperceive uchbehavior s a cunningwaytogetone'sway, obuytimeand oimposeone'swill onothers i.e.,forcing).Perhapsvoiding,morethananyof theotherconflictmanagementtrategies,nvolvesbehaviorshatare difficultojudge,making ccuratenderstandingfunderlyingntentionsmoremportant.ecause he ndividualhasbetterknowledge boutone'sownintentionshanopponents ndneutral bserversonvergencebetween elf-reportsndother, eports f avoidings likelyto be low.An issue hathasgeneratedomedebaten theconflictiteratures whetherompromisings a 'lazyformof problem olving'(Pruitt ndRubin,1986)or a trulydistinct onflictmanagementtrategy(Vande Vliert,1997).Upto now, hedebatehas beenconceptualatherhanempirical.However,fcompromisings indeeda distinct trategy ne shouldbe ableto design temsthataredistinct romitemsdesignedo measure roblemolvingandconfirmatoryactoranalysis hould eveala better itforafive-factorhana four-factor odel.Study3 showedbothconditionsanbe satisfied.naddition,if compromisings a trulydistinct trategyhatresults rom ntermediateoncernorself and nter-mediate oncernorothers,multidimensionalcaling hould evealapatternf inter-correlationshatcloselyfitsthe DualConcernTheorydepictednFigure1.Again,Study3 (i.e.,Figure ) showed hiscondition anbe satisfied: voiding,orcing,problem olvingandyieldingoccupiedpositionsn thefourcomersof a two-dimensionalpace,whilecompromisingended owards midpoint osition.Thispattern f results peaks o theconstruct alidity f the instrumentnd he sub-scalemeasuringcompromisingn particular.More mportantly,heseresults ontributeo conflict heoryn that heylendfurtherempirical)upportortheargumenthatcompromisings distinct romproblemolving(cf.,VandeVliertandKabanoff,990;VandeVliert,1997). t s concludedhat utureesearch,heorydevelopment,ndpracticewouldbenefit romcloserattentiono compromisings a distinct trategyformanagingonflictat work.

    Thepsychometricvidence ortheDUTCH rovidesndirectupportorDualConcern heory nd,as such, complementsexperimentaland field research for reviews, see De Dreu et al., 2000; VandeVliert, 1997). AlthoughDual ConcernTheoryhas been aroundfor quite some time, many scholarscontinue to work with related but differentmodels. For instance,the basic distinction betweencom-petitionandcooperationcontinues to inspiremuchtheoreticalandempiricalwork(for a review,seeTjosvold, 1998). The currentresearch ndicatesthis basic distinctionto be too broad and we suggestthat future researchandpracticewould benefitfrom the more fine-grainedanalysis offeredby DualConcernTheory.Copyright 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    20/25

    CONFLICTANAGEMENT 63The results of the threestudiestogethersuggestthatfor problemsolving, forcingandyielding (butnot for avoiding) self-reportsconverge with opponent-reportsand with actual conflict behavior.Resultsof Studies2 and3 showthatthe leanandexpandedversion of the instrumenthasgood psycho-metricqualities.Convergentand discriminantvaliditywere good, and invarianceacross genderandhierarchical elationshipwas high. These studies show that the DUTCHis a psychometricallysoundinstrumenthathas,exceptforavoiding,good predictivevalidity.The DUTCHappears o be a flexibleand parsimonious nstrument hat may be of use to both researchers nterestedin the relationshipsbetweenconflictmanagement trategiesandrelatedaspectsof organizationalife, as well as to practi-tioners nterested n diagnosing(preferences or)conflictmanagement trategies n organizationalet-tings. Althoughdirect,empiricalcomparisonsawait futureresearch, he psychometricqualitiesof theDUTCHappear o comparefavorablyto otherinstruments n the literatureandfuture researchmaybenefitfromusing the DUTCH.

    Self-servingbiasesStudy 1 revealedthatself-ratings or problem solving were significantlyhigherthanratingsby one'sconflictopponent.Thismayreflecta self-servingbias in the assessmentof one's conflictmanagement.Self-servingbias has been observedin a wide varietyof settingsincludingconflictsin organizationalsettings, n close relationshipsand in interpersonal egotiation(fora review,see Johns,1998).Currentresults are consistent with this past research. Our results suggest social desirabilitydid not play animportant olebecause social desirabilitywouldleadone to over-estimateone's tendencytowardspro-blem solvingandto underestimate ne's tendencyto engagein forcing.The latterwas notobserved.Itmay well be thatpeople tend to underestimate he extent to which theiropponentengages in problemsolving. Conflictresearchsuggestspeople tend to overly emphasizeother'snegativetraitsandbeha-viors (De Dreuet al., 1995) and tendto engage in 'hostileattributions' suspectingulteriormotivesunderlyingother'sambiguousbehavior(Baron,1997).An interestingavenue for futureresearch s to examinethe conditionsthatfosteror inhibitthe indi-vidual'stendencyto underestimatehe opponent'stendencyto engage in problemsolving. The studyby De Dreu et al. (1995) suggestedthat as conflict intensifies self-servingbias becomes more pro-nounced,presumablybecauseconflictparties eel morethreatenedanddevelopa morenegativeviewof theiropponent cf., Rubinet al., 1994). A recentstudyby Gelfandet al. (2000 - manuscriptunderreview)showedthatnegotiators n anindividualisticculture,such as the U.S. or the Netherlands,weremorelikely to develop self-servingperceptionsof own andother'sconflictmanagement hannegotia-tors in a collectivistculture,suchas Japan.Theyexplainedthis culturaldifference n termsof the col-lectivist negotiator's strongerinclinationto define oneself in terms of interdependentratherthanindependent ocial structures.Together, hese studiessuggestthatself-serving perceptionsof conflictmanagementdevelopespeciallywhenconflictpartiesdefinethemselvesas differentrather hansimilarfrom their conflict opponent.

    Limitationsand avenuesfor future researchThe current esearchhas strengthsas well as weaknesses.We examinedthe psychometricqualitiesofthe DUTCHin a varietyof samples(psychology students n Study 1; employees in a manufacturingfirm n Study2; and a large sampleof highereducatedprofessionals n Study3). In addition,we usedmultiple ways to gatherdata, including paper-pencil and computerizedmethodology.Consistencyacross studies suggests the specific way in which data are gathereddoes not substantially mpact

    Copyright 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    21/25

    664 C. K. W. DE DREU ETAL.the validityof the (sub)scales,but morecontrolledstudies areneededto settle this issue. Also, resultsin Study3 mayhavebeeninfluencedby self-selection,in thatwe hadno controloverwho did,and whodid notvisit the web site, and who diddecide (not)to participaten the study.Finally,we includedtwoversions of the DUTCH,a lean versionthat includedproblemsolving, forcing, yielding andavoiding,and an expandedversionthat also includedcompromising.Althoughboth versionsappeared o havesound psychometric qualities, the drawbackof this approach s thatwe have no data regarding heconvergence between self-reports and other-reportsfor compromising. This issue needs to beaddressed n futureresearch.To understand he resultsfor avoidingwe distinguishedbetweenintentionsand(observable)beha-viors. Likewise, in discussingthe self-serving perceptionsof conflictmanagementstrategies,we dis-tinguishedbetween intentionsand behavior.In the conflict literature, his distinctionis not alwaysexplicit. Althoughwe askedpeople what they do when in conflict, it may be safe to assume thattheDUTCH is a measureof the individual'sbehavioral ntentions,rather hanhis or her actualbehavior.Sometimes,intentions ranslate nto actualbehavior,but sometimestheydo not.Sometimes, ntentionsdo translate nto behaviorbut it is notperceivedas such. Conflicttheory,and futureresearchon scaledevelopment, may benefitfrom a more thoroughanalysisof what individuals n conflictwant to do,what they believe they shoulddo, andwhat they actuallydo.

    As mentionedat theoutset, heDUTCH s notnecessarily imitedto theDutchculture.The theoreticalbasis forthe nstrument eneralizesacrosscultural oundaries, ndweexpectthepsychometric ropertiesto beconstantacrossculture.Thisis not tosay,however, hat ndividualsromdifferent ultures espond oconflict nsimilarways.Cultural ifferences houldreveal tself,however,ndifferencesn themeansandnotindifferences n factorstructuresndrelatedpsychometric roperties.Future esearchs needed,how-ever,to test thepsychometricpropertiesof the DUTCHin differentcultures.

    Preferences or conflictmanagement:person or situation?Before closing,we wish to address hequestionof whetherconflictmanagements stableovertime,orprimarilyheproductof thesituation. nDual ConcernTheoryconflictmanagements seenastheproductof concernforself andconcern orothers(see Figure1).Concern or self and concernforothersare,inturn,predictedbyone'spersonality nd hesituation DeDreuetal., 2000;Pruitt ndRubin,1986;VandeVliert,1997).For nstance,stable ndividualdifferences n socialvalueorientation orrelatewithconcernfor others:Partieswith a pro-socialorientationhavehigherconcernfor othersthanpartieswith an indi-vidualisticorcompetitivevalueorientation.Likewise,evidencesuggestsstable ndividualdifferences ncollectivistversus ndividualistic aluespredictconcern orothersandthatconflictparties hatscorehighon agreeablenessalso tend to havehigherconcern or others.At thesametime,instructions y constitu-ents,incentivestructures,heexpectationof cooperative uture nteraction, ndpositivemood inductionsall increaseconcernforothers(De Dreuet al., 2000).The fact that conflictmanagements the productof bothpersonalityandthe situationdoes not sug-gest necessarilythat within a particularwork settingconflictmanagementcannotbe accuratelypre-dictedon the basis of instrumentsike theDUTCH.Worksettingstend to remainrelativelystable overtime. Employeesinteractwith the same co-workers, ncentivestructuresdo not changeovernightandemployeesdo the same kind of work for longer periodsof time andthusface the same (interpersonal)problemson arecurringbasis. Inaddition, ndividualswithin thesameunit,teamordepartmentend toinfluence one another(SalancikandPfeffer, 1977), thus creatingtheir own social environmentwith,most likely,rather tablepreferences ordealingwith conflict.The consequence s that anindividual'sactual and preferredconflict managementstrategiesare likely to be relativelystable over time. Wecannotknow,however,whether his is due to stableindividualdifferences, o the fact that thesituation

    Copyright 2001JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    22/25

    CONFLICT ANAGEMENT 665is relatively mpervious o change,or both.Consistentwith Dual ConcernTheory,however,we expectthatthe individual'sconflictmanagement trategiesat work arerelativelystableover time due to bothstableindividualdifferencesandthe relativelystable situations n which people work.

    ConclusionThe current esearchreveals,firstof all, that the DUTCH is a validandreliableinstrumenthat can beused to measureconflictmanagement trategies n theworkplace.The DUTCHmay well complementthe existing conflict management ests especially when flexibility and parsimonyare importantandpsychometricquality is valued. Second, the current researchindicates that conflict theory wouldimprove by incorporating ompromisingas a separateconflict managementstrategythat is distinctfrom problemsolving. Third,this study indicatesthatfutureresearchwould benefit from using theDUTCH to assess an individual'sconflictmanagement trategies.Finally,this researchsuggests thatpractitioners hould no longer rely on a broadcompetition-cooperationdistinction,butconsequentlyandconsistentlyresortto a five-factor axonomyof conflict managementstrategieswhen interveningin conflictsituations,or assessing conflictmanagementstrategies n a particularorganization.

    AuthorbiographiesCarsten K.W.De Dreu is Professorof Organizational sychologyattheUniversityof AmsterdamandDirector of Research of the KurtLewin GraduateSchool for Social Psychology.His research s con-cerned with social influence,groupdecision making,negotiationandorganizationalconflict.Arne Evers is AssociateProfessorof OrganizationalPsychologyat theUniversityof Amsterdam.Hisresearch s concernedwith testdevelopmentandvalidation,occupational tressand vocationalchoice.Bianca Beersma is a PhD student n OrganizationalPsychologyat the Universityof Amsterdam.Herresearchconcerns the motivational,cognitive and structuralaspects of groupnegotiation,and teamperformance.Esther Kluwer is AssistantProfessor n Social andOrganizationalPsychology at UtrechtUniversity.Her research nterestsare social conflict and social justice in close relationshipsandgenderissues.Aukje Nauta is Assistant Professor in the Departmentof Managementand Organizationat theUniversity of Groningen,and Researcherat TNO Work and Employment.Her researchinterestsinclude conflict and negotiationin organizations,organizationalcommitment,and humanresourcepractices.

    ReferencesAndersonC,Gerbing W.1988.Structuralquationmodellingnpractice:review nd ecommendedpproach.PsychologicalBulletin 103: 411-423.BaronRA.1997.Positive ffectsof conflict:nsightsromsocialcognition.nDe DreuCKW,Vande VliertE(eds). Using Conflict n Organizations,Sage: London;177-191.Bartram D. 1997. Review of ability and aptitude tests (Level A) for use in occupational settings. BritishPsychological ocietyBooks:Leicester.

    Copyright 2001 JohnWiley& Sons,Ltd. J. Organiz. ehav. 2, 645-668(2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    23/25

    666 C. K.W.DEDREUETAL.BeersmaB, De DreuCKW.1999. Negotiationprocessesand outcomesin prosociallyandegoistically motivatedgroups.InternationalJournalof ConflictManagement10: 385-402.Bentler PM. 1990. Comparative it indexes in structuralmodels.PsychologicalBulletin 107: 238-246.Blake R, Mouton JS. 1964. TheManagerial Grid. Gulf:Houston,TX.CarnevalePJD, PruittDG. 1992. Negotiationandmediation.AnnualReviewof Psychology43: 531-582.Crocker L, Algina J. 1986. Introductionto Classical and Modern Test Theory.Holt, Rinehart& Winston:Orlando,FL.De DreuCKW,NautaA, Vande VliertE. 1995.Self-serving evaluationof conflictbehaviorand escalationof thedispute.Journal of AppliedSocial Psychology25: 2049-2066.De Dreu CKW, HarinckF, Van Vianen AEM. 1999. Conflict and performance n groups and organizations.InternationalReviewof Industrialand OrganizationalPsychology 14: 376-405.De DreuCKW,GiebelsE, Vande Vliert E. 1998. Social motives andtrust n integrativenegotiation: hedisruptiveeffects of punitivecapability.Journalof Applied Psyhcology 83: 408-422.De Dreu CKW,WeingartLR, Kwon S. 2000. Influenceof social motives on integrativenegotiation:a meta-analyticalreview andtest of two theories. Journalof Personalityand Social Psychology78: 889-905.Deutsch M. 1973. Theresolutionof conflict:constructiveand destructiveprocesses. YaleUniversityPress:NewHaven, CT.EuwemaMC,Van de Vliert E. 1990.Gedragen escalatiebij hierarchische onflicten(Behaviorandescalation nhierarchicalconflicts). ToegepasteSociale Psychologie 4: 28-41.Evers A, Frese M, CooperCL. 2000. Revisions and furtherdevelopmentsof the OccupationalStressIndicator(OCI):LISRELresultsfromfourDUTCH studies.Journalof Occupationaland OrganizationalPsychology73:221-240.JanssenO, Vande VliertE. 1996. Concern orother'sgoals: key to de-escalationof conflict.International ournalof ConflictManagement7: 99-120.Johns G. 1998.A multi-leveltheoryof self-servingbehavior n andby organizations.Research n OrganizationalBehavior21: 1-38.KillmanRH,Thomas KW.1977. Developinga forcedchoice measureof conflicthandlingbehavior.Educationaland PsychologicalMeasurement37: 309-325.Landy F. 1978. Conflictmanagementsurvey.In Buros OK (ed). Eigth mentalMeasurementYearbook,Vol. 2,Gryphon:HighlandPark,N.J; 1173-1174.Loehlin JC. 1998. Latent VariableModels. An Introduction o Factor,Path, and StructuralAnalysis, 3rd ed.LawrenceErlbaum:Mahwah.NJ.PruittDG. 1998. Social conflict. InGilbertD, Fiske ST,LindzeyG. (eds). Handbookof Social Psychology,4thedn, Vol.2, McGraw-Hill:New York;89-150.PruittDG, Rubin J. 1986. Social Conflict:Escalation,Stalemate and Settlement.RandomHouse:New York.PutnamLL, Wilson CE. 1982. Communicative trategies n organizational onflicts:reliabilityandvalidityof ameasurement cale. In CommunicationYearbook,Vol.6, BurgoonM. (ed). Sage:BeverlyHills, CA;629-652.RahimA, MagnerNR. 1995. Confirmatoryactoranalysisof the styles of handling nterpersonal onflict:first-order factormodel and its invarianceacross groups.Journalof AppliedPsychology80: 122-132.Rubin JZ, Pruitt DG, Kim S. 1994. Social Conflict: Escalation, Stalemate, and Settlement.McGraw-Hill:New York.Salancik GR, Pfeffer J. 1977. An examinationof need satisfactionmodels of job satisfaction.AdministrativeScience Quarterly22: 427-456.SchumackerRE, Lomax RG. 1996. A BeginnersGuide to StructuralEquationModelling.LawrenceErlbaum:Mahwah,NJ.Shockley-ZalabakP. 1988. Assessing the Hall Conflict Management Survey. Management CommunicationQuarterly1: 302-320.SpectorPE, Jex SM. 1998. Development of four self-reportmeasuresof job stressorsand strain: nterpersonalconflictatworkscale,organizationalonstraints cale,quantitativeworkload nventory, ndphysicalsymptomsinventory.Journalof OccupationalHealth Psychology3: 356-367.ThomasKW. 1992. Conflict andnegotiationprocessesin organizations. nDunnette MD, Hough LM (eds). Handbook of Industrial and OrganizationalPsychology, 2nd ed, Vol. 3,ConsultingPsychologistsPress:Palo Alto, CA; 651-717.Thomas KW, KillmanRH. 1978. Comparisonof four instrumentsmeasuringconflict behavior.PsychologicalReports42: 1139-1145.

    Copyright? 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    24/25

    CONFLICT ANAGEMENT 667Tjosvold D. 1998. Cooperativeand competitive goal approachto conflict: accomplishmentsand challenges.AppliedPsychology:An InternationalReview 47: 285-342.Vande VliertE. 1997. Complex nterpersonalConflictBehavior PsychologyPress: London.Van de Vliert E, Kabanoff B. 1990. Toward theory-basedmeasures of conflict management.Academy ofManagementJournal33: 199-209.

    Copyright 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav.22, 645-668 (2001)

  • 8/3/2019 Conflict Management Strategies in Workplce

    25/25

    668 C.K. W.DEDREUETAL.Appendix ATheDutch Test or ConflictHandling(DUTCH)WhenI have a conflict at work,I do thefollowing:Yielding1. I give in to the wishes of the otherparty.2. I concurwith the otherparty.3. I try to accommodate he otherparty.4. I adaptto the otherparties' goals and interests.Compromising5. I try to realize a middle-of-the-road olution.6. I emphasizethat we have to finda compromisesolution.7. I insist we both give in a little.8. I strive wheneverpossible towardsa fifty-fiftycompromise.Forcing9. I push my own point of view.10. I search for gains.11. I fightfor a good outcomefor myself.12. I do everythingto win.Problem solving13. I examineissues until I finda solutionthatreally satisfiesme andthe otherparty.14. I standfor my own and other'sgoals andinterests.15. I examine ideas fromboth sides to finda mutually optimalsolution.16. I work out a solutionthat serves my own as well as other'sinterestsas good as possible.Avoiding17. I avoid a confrontationabout ourdifferences.18. I avoid differencesof opinionas much as possible.19. I try to make differences loom less severe.20. I try to avoid a confrontationwith the other.Note. Items could be answeredon a 5-pointscale (1 = not atall, to 5 = verymuch).Items are translated rom DUTCH and werepresented n a randomorder.The lean version(Study 1 and2) does not includethe compromising cale; the expandedversion(Study3) does include thecompromising cale.

    Copyright? 2001 JohnWiley & Sons, Ltd. J. Organiz.Behav. 22, 645-668 (2001)