Experience in integrative negotiations: What needs to be learned

Journal of Experimental Social Psychology 43 (2007) 77–90

www.elsevier.com/locate/jesp

Experience in integrative negotiations: What needs to be learned? �

Simone Moran a,¤, Ilana Ritov b

a Department of Management, School of Management, Ben Gurion University of the Negev, Beer Sheva, Israelb Department of Psychology, Hebrew University, Jerusalem, Israel

Received 14 March 2004; revised 8 September 2005Available online 2 March 2006

Abstract

In this research we focus on the roles of experience and understanding in fostering integrative negotiation performance. We report ontwo experiments in which we distinguish between understanding opponents’ general priorities among issues versus understanding their spe-ciWc gains for particular oVers. Although experience enhanced integrative performance even in the absence of understanding, we found thatunderstanding the speciWc gains had an incremental eVect on performance. We conclude that while generally acknowledging opponents’interests is not suYcient, the additional inferential step of assessing their speciWc gains throughout the negotiation process is advantageous.© 2006 Elsevier Inc. All rights reserved.

Keywords: Negotiations; Experience; Learning; Perspective taking

Negotiation is a continual interpersonal interactive pro-cess of decision making, in which participants perceive andinterpret the information available to them and proceed toact on these perceptions. Improving negotiation skills andincreasing the ability to negotiate eVectively is crucial notonly in managerial, political, and business contexts, but inall avenues of life. A primary topic of negotiation researchconcerns the achievement of integrative agreements asopposed to distributive agreements. Distributive agree-ments represent agreed divisions of a Wxed pie of resources,where one side’s gains are the other side’s losses. Integrativeagreements, in contrast, reconcile the interests of both par-ties, and lead to higher joint beneWt (Pruitt, 1983; Walton &McKersie, 1965). The present research seeks to gain a better

� Simone Moran is a lecturer at the School of Management, a member ofthe Decision Making and Economic Psychology Center, and of the Centerfor Studies of Populations in ConXict, Ben Gurion University of the Ne-gev, Israel. Her research interests include decision-making, negotiations,organizational behavior, and consumer behavior. Ilana Ritov is an associ-ate professor at the Hebrew University, Jerusalem. Her research interestsinclude the role of emotions in decisions, temporal changes in outcomeevaluation, attitudes to public issues, and cognitive biases in negotiations.

* Corresponding author. Fax: +972 8 6477697.E-mail addresses: [email protected] (S. Moran), [email protected]

(I. Ritov).

0022-1031/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.jesp.2006.01.003

understanding of the roles of experience and of cognitiveprocessing in fostering integrative negotiation performance.

Many negotiation situations contain integrative poten-tial, i.e., an increase in the joint gain available to the negoti-ators over and above the joint gain aVorded by a Wxed-sumsolution (Thompson, 2001). In negotiations involving com-plex agendas where several issues are under consideration,integrative agreements can be achieved through diVerentspeciWc strategies (e.g., Bazerman, 2001; Thompson, 2001).In this research we focus primarily on one of these integra-tive strategies, namely “logrolling,” which is an appropriatestrategy for creating value when parties have diVerent pri-orities among issues. It speciWcally refers to a strategywhereby value is created by each party conceding on lowpriority issues in exchange for concessions on issues ofhigher priority.

Although integrative agreements are frequently avail-able, mutually beneWcial, and therefore desirable solutions,the behavioral research on negotiation and conXict resolu-tion reveals that negotiators are often ineYcient in thatthey fail to reach integrative agreements (e.g., Bazerman,Magliozzi, & Neale, 1985; Bazerman & Neale, 1983, 1992;Thompson, 1990b; Thompson, Gentner, & Loewenstein,2000; Thompson & Hastie, 1990). Prior research has shownthat repeated experience with an integrative negotiation

mailto: [email protected]




78 S. Moran, I. Ritov / Journal of Experimental Social Psychology 43 (2007) 77–90

task can improve negotiators’ integrative outcomes (e.g.,Bazerman et al., 1985; Neale & Bazerman, 1991; Thomp-son, 1990a, 1990c; Thompson & DeHarpport, 1994;Thompson & Hastie, 1990). However, important questionsconcerning the nature of the learning process involved insuch repeated experience remain unanswered, particularlyquestions about the precise link between experience, cogni-tive understanding, and performance.

An important and relevant distinction has been made inthe decision-making literature between experience andexpertise (Bazerman, 2001; Dawes & Corrigan, 1974; Ham-mond & Grassia, 1985; Neale & Bazerman, 1991; Neale &Northcraft, 1990). Experience is deWned as repeated feed-back that typically leads to mindless behavioral modiWca-tion and to habitual learning that does not necessarilyrequire understanding. Expertise, on the other hand, isdeWned as developing a “strategic conceptualization,”understanding what constititutes an eVective strategy andwhy, and consciously monitoring the process. While experi-ence is mainly eVective for performing a speciWc task, a cen-tral feature of expertise is the ability to be adaptive tochanges in situational demands and to eVectively transferskills across diVerent tasks.

As Neale and Bazerman (1991) note, negotiators’ highquality outcomes can be a result of one of three processes:(1) randomly selecting an eVective strategy without anyknowledge of what he or she has done correctly, (2) learn-ing a particularly eVective script from experience, but hav-ing little understanding why this particular script works inthis particular situation, or (3) acquiring a strategic con-ceptualization of negotiation situations that summarizeswhen and why particular strategies are eVective (i.e.,expertise). Similarly, Thompson (1990b, 1990c) points outthat in repeated negotiations with a particular task thathas a limited number of optimal solutions, reaching anintegrative agreement does not necessarily require thenegotiator’s improved understanding of the other partyor of the underlying integrative potential. It can also bereached by the negotiator merely stumbling upon an opti-mal agreement, which is a speciWc script that works in theparticular task, without comprehensively understandingthe situation.

Hence, examining learning by solely looking at perfor-mance in an unchanging setting limits the ability to capturethe nature of the underlying learning process involved.Indeed, research on learning integrative negotiation skillshas not been limited to the study of performance in a spe-ciWc task. The degree to which experience enhances thedepth of understanding and expertise has also been exam-ined, by measuring negotiators’ assessments of the otherparties’ interests and utility functions (e.g., Thompson,1990a, 1990c, 1992; Thompson & DeHarpport, 1994;Thompson & Hastie, 1990), as well as by examining thetransferability of learned principles to a diVerent negotia-tion context (e.g., Bereby-Meyer, Moran, & Unger-Aviram,2004; Loewenstein & Thompson, 2000; Loewenstein,Thompson, & Gentner, 1999; Mannix, Northcraft, & Neale,

1991; Thompson, 1990a, 1990c). However, to date, the Wnd-ings of such studies are inconclusive.

Although experience has commonly been found to resultin improved assessments of the opponent’s interests (e.g.,Thompson, 1990a, 1990c; Thompson & DeHarpport, 1994;Thompson & Hastie, 1990), there are some indications thatthe improvement in understanding due to mere experienceis rather limited. For example, in a study by Thompson(1992), even when negotiators bargained under ideal condi-tions—with immediate, clear, and complete process feed-back—their judgments of the other party remained faulty.Moreover, positive transfer of integrative negotiation skillsto diVerently structured tasks was not easily achieved, andtypically required more than simple repeated experience(e.g., Bereby-Meyer et al., 2004; Loewenstein & Thompson,2000; Loewenstein et al., 1999; Mannix et al., 1991; Thomp-son, 1990a, 1990c).

Previous studies are also inconclusive regarding the roleof improved understanding in enhancing subsequent inte-grative behavior. On one hand, many authors have empha-sized the key role of judgment accuracy for achievingimprovement in integrative negotiation performance (e.g.,Thompson, 1990a, 1990c; Thompson & DeHarpport, 1994;Thompson & Hastie, 1990). Gaining information about theother party has been shown to be an important factor ineYcient bargaining (e.g., Bazerman & Carroll, 1987; Pruitt& Lewis, 1975; RaiVa, 1982; Weingart, Thompson, Bazer-man, & Carroll, 1990; Zeng & Sycara, 1998). On the otherhand, some work, including work referring to perspectivetaking, suggests that even if negotiators possess knowledgeabout their opponents’ priorities, they may not have theability or the motivation to actually use this informationduring the negotiation process (e.g., Bazerman, 2001; Baz-erman & Neale, 1983; Carroll, Bazerman, & Maury, 1988;De Dreu, Koole, & Steinel, 2000; Pinkley, GriYth, &Northcraft, 1995; Samuelson & Bazerman, 1985; Shubik,1971).

A number of recent studies provide additional indica-tions that acknowledging the other parties’ priorities maynot be suYcient for accomplishing integrative outcomes,and that while using this information appropriately at thenegotiation table may be a key factor, it is not straightfor-ward (Gentner, Loewenstein, & Thompson, 2003; Nadler,Thompson, & Van Boven, 2003). These studies typicallydemonstrated instances where integrative negotiation per-formance was inferior, even though negotiators had fullinformation and/or were highly accurate about the priori-ties of the other party.

In line with the distinction between possessing and usinginformation, we distinguish in the present study betweentwo types of judgment accuracy: (a) judging the other par-ties’ general proWt schedules, and (b) assessing their speciWcgains for particular oVers that are exchanged during thenegotiation process. As mentioned above, previous studiesrelied mostly on the elicited assessment of the other party’sgeneral proWt schedule—i.e., assessment of the opponent’spayoV schedule for each of the negotiating issues—as their

S. Moran, I. Ritov / Journal of Experimental Social Psychology 43 (2007) 77–90 79

measure for judgment accuracy (e.g., Thompson &DeHarpport, 1994; Thompson & Hastie, 1990). This mea-sure can indicate the degree to which the negotiator under-stands the other party’s priorities, i.e., whether thenegotiator accurately assesses which issues are more impor-tant and which are less important to the other party. How-ever, we argue that negotiators may understand theiropponents’ proWt schedules in general terms, yet still notapply and further process this knowledge in order to esti-mate what their opponents might gain from diVerent pro-posals throughout the negotiation process.

We maintain that the processing of information fromthe opponent’s perspective can be an important contribut-ing factor to successful negotiations. This idea is consistentwith the literature on the role of perspective taking in nego-tiations. Previous research (Bazerman & Neale, 1982, 1983;Neale & Bazerman, 1983) found that negotiators who had agreater tendency to think about the perspective of otherswere more successful in their negotiations. This increasedattention to the perspective of the other party allowed themto better predict their opponents’ goals, expectations, andreservation points, all keys to negotiating successfully. Inthe present study, we therefore examined the role of under-standing, and in particular the role of increased informa-tion processing that entails focusing on and thinking aboutthe other’s perspective. The latter was examined by measur-ing assessments of the other party’s speciWc gains for partic-ular oVers (also used in the study by Moran & Ritov, 2002).We used this measure in addition to the more commonmeasure—judgment of the other party’s general proWtschedule.

Finally, we ask whether negotiators’ integrative perfor-mance can improve as they gain experience, without anytype of understanding. There are several indications in theliterature that even if enhanced understanding and moreaccurate processing of information does improve integra-tive negotiation performance, such improvement may alsooccur without increased understanding. For example,Moran and Ritov (2002) found that proposing highly inte-grative counter-oVers did not necessarily require improvedunderstanding of the other parties’ interests. Even in thestudy by Thompson and DeHarpport (1994), where themain Wnding supported the notion that judgment accuracyplays a key role, there is evidence that it may not be a neces-sary condition. Although in their study integrative perfor-mance improved to the greatest degree for thoseparticipants who received maximum feedback and showedthe largest improvement in judgment accuracy, the perfor-mance of negotiators who received no feedback at all alsoimproved over time, even though there was no parallelimprovement in their judgment accuracy.

Taken together, previous Wndings suggest that althoughrepeated integrative experience may lead to improvedassessment of the other parties’ general priorities, it maynot lead to high levels of cognitive processing and under-standing of the other parties’ perspective. Moreover,although improved understanding can enhance integrative

negotiation performance, it may not be a necessary condi-tion. In the present research employing two experiments, weaim to further explore the links between experience, cogni-tive understanding and processing, and performance inboth unchanging (Experiment 1) and changing (Experi-ment 2) environments.

Experiment 1

In the Wrst experiment we concentrated on the initialstages of the negotiation process, and examined the eVectsof experience in an unchanging setting. Our focus on theinitial stage was motivated by the intriguing evidence pro-vided by Thompson and Hastie (1990) that most of theeVective learning for integrative negotiations occurs duringthe Wrst few minutes of the negotiation. An additional moti-vation for focusing on the initial stage stems from therobust Wndings regarding the crucial role of initial oVers innegotiations. The fact that the value of the initial oVer hasbeen found to have a great impact on the Wnal agreement indistributive negotiations (see Benton, Kelley, & Liebling,1972; ChertkoV & Conley, 1967; Galinsky & Mussweiler,2001; Liebert, Smith, Hill, & KeiVer, 1968; Rubin & Brown,1975; Yukl, 1974), suggests that the antecedent of an inte-grative agreement may already be revealed in the initialoVer. Indeed, this notion was supported in a recent study byMoran and Ritov (2002), where participants with no previ-ous experience with the negotiation task evaluated andresponded to diVerent initial oVers forwarded to them byan opponent. In that study, the speciWc composition of theinitial oVer was found to have an eVect beyond the alreadydocumented eVect of its total value. In particular, logrollingopening oVers (characterized by high inter-issue variance)led to more logrolling counter-oVers and integrative agree-ments than equal value non-logrolling opening oVers (withlow inter-issue variance).

In the present study, we employed the same methodol-ogy but focused on the role of experience. We Wrst lookedat the eVect of experience on understanding, and then con-tinued to explore the eVects of experience and of under-standing on promoting integrative performance.

Based on the above literature review, we hypothesizethat:

H1. Integrative experience will increase the accuracy ofassessing the other parties’ general priorities.

Given the robust Wndings on the positive eVect of experi-ence on integrative negotiation performance, our secondhypothesis is:

H2. Integrative experience will lead to more integrative per-formance in the initial stage of the negotiation

As presented in the introduction, there are some indica-tions that having information about the opponent does notguarantee improved integrative performance. Notwith-standing, if, as proposed by many researchers, understand-ing enhances integrative performance, then:


H3a. Greater accuracy in assessing the other parties’ gen-eral priorities should lead to more integrative performance,over and above the eVect of integrative experience, and

H3b. Greater accuracy in assessing the other parties’ spe-ciWc oVer values should lead to more integrative perfor-mance, over and above the eVect of integrative experience.

Lastly, if as suggested in the introduction improvedunderstanding is not a necessary condition for improvedintegrative performance over time, then we predict:

H4. Integrative experience will enhance integrative perfor-mance, over and above the eVects of accuracy in assessingthe other parties’ general priorities and of accuracy in assess-ing the other parties’ speciWc values for the given oVers.

Method

ParticipantsOne-hundred and thirty undergraduate Management

and Industrial Engineering students participated in theexperiment as part of a class exercise. Eighty-six partici-pated in the experimental condition and an additional 44participated in an external control condition. All of thesestudents were inexperienced with regard to participation inmulti-issue laboratory negotiation tasks.

ProcedureThe research paradigm we employed was adapted from

Bazerman et al. (1985) (modiWed from the methodologyoriginally developed by Kelley, 1966, Pruitt & Lewis, 1975).The experimental condition was run in three separate ses-sions (N’s were 22, 28, and 36). At the beginning of eachsession, participants were given an information packagethat included a description of the task as a free marketnegotiation. They were told that the market consists of rep-resentatives of two diVerent company types, “Type A” and“Type B,” who shall engage in two-party negotiations inwhich three issues were negotiable. Participants were ran-domly and evenly assigned to be representatives of Type Aor Type B companies. All Type A’s and all Type B’sreceived identical information. All participants were told toattempt to gain maximum proWt during the Wxed marketperiod (20 min), by reaching agreements with as many rep-resentatives as possible of the other type of company (TypeB or Type A, respectively). Each participant also receivedan “individual” proWt schedule specifying his/her proWt foreach level of each of the three issues (see Appendix A). Asin Bazerman et al. (1985), the proWt span (maximum poten-tial proWt) diVered between issues, and participants hadincomplete information: they did not receive informationconcerning their opponents’ proWt schedule. As can be seenin Appendix A, issue “R” was a distributive issue, andissues “D” and “F” were logrolling issues. The minimumagreement pie for this negotiation was $8000. Splitting thisminimum pie equally between the two parties would aVordeach of them $4000. Integrative agreements—i.e., agree-

ments resulting in a pie greater than $8000, might beobtained by parties logrolling on issues D and F. An opti-mal integrative agreement could be reached when partieslogrolled and agreed on transaction terms of level 1 forissue D and level 9 for issue F. Such an agreement wouldlead to a combined proWt of $10,400 for both negotiators,with each of them gaining at least $4000—the amount he orshe would gain from an equitable non-integrative agree-ment (i.e., equally splitting the minimum $8000 pie).1

After 10 min of reading instructions and proWt schedules,participants were randomly matched for the Wrst negotia-tion and proceeded to engage in a 20-min simulation of themarket. They Wlled out a negotiation report after eachnegotiation, in which they speciWed the sequential numberof the negotiation (Wrst, second, etc.), whether an agreementwas reached, and the terms agreed upon (if agreement wasachieved). Upon completion of the report, participantswere randomly rematched by experimenters with a partnerwith whom they had not previously negotiated.

When the market terminated, participants proceeded toengage in a test phase. We gave each participant an addi-tional page of instructions, in which we requested him orher to presume that he or she was negotiating the identicaltask with another opponent in the same market. At thisstage, participants were presented with a single initial oVer,supposedly forwarded by this new opponent, and spent upto twenty minutes completing a questionnaire referring tothe initial oVer.

Two initial oVer types were manipulated between sub-jects, one high variance—logrolling, and one low vari-ance—distributive (no logrolling). The overall value of thetwo types of oVers from the participants’ perspective washeld constant. The larger pie (combined value) of the log-rolling oVer was due to the fact that it was worth more tothe other party than the distributive oVer. Appendix Ashows the proWt schedules and the initial oVers that weresubmitted. As can be seen, both the integrative and thedistributive oVers forwarded to “Type A” participants(“2-3-9” and “5-6-6,” respectively) were worth a total valueof $4800 for their recipient. However for the other party(the supposed initiator), the integrative oVer was worthmore than the distributive one: $5300 and $3500, respec-tively.2 The same was true with respect to the two oVertypes forwarded to “Type B” participants (“4-4-5” and “1-7-8”).

In the questionnaire, participants Wrst assessed the initialoVer’s attractiveness for themselves and for the otherparty,3 and then were requested to:

1 Such an optimal outcome could also be reached when parties agreedon level 9 for issues R and F and level 1 for issue D, or on level 1 for issuesD and R and level 9 for issue F.

2 This information was not explicitly given to the participants.3 Results of analyzing these attractiveness rating measures did not pro-

vide additional information to the main Wndings and conclusions of thecurrent study. Hence, for reasons of simplicity and readability, we did notreport them.


(a) Assess the initial oVer’s value for the initiator on eachof the issues: They were requested to assess what theinitiator’s proWts would be on each of the three issues:“D,” “R,” and “F,” for the level he or she had pre-sumably oVered.

(b) Propose a counter-oVer.(c) Estimate the other party’s general proWt schedule, by

assessing the lowest and highest value for him or heron each of the issues. Participants were again pre-sented with their own proWt schedule, together withan identically formatted blank proWt schedule. Theywere asked to assume that the blank schedule wasthat of the other party. Their task was to Wll in thevalues of the minimum and the maximum levels ofeach of the issues, reXecting their assessment of theother party’s proWt schedule.

A potential limitation of the above described method isthat although proWt schedules were given as private infor-mation, we did not include safeguards against participantssharing this information. Consequently, in order to directlytest the role of information availability, we included anexternal control condition with an additional 44 partici-pants from the same subject pool as the participants of theexperimental condition. The procedure for these partici-pants was similar to the procedure used for the experimen-tal condition, except that after they read the instructionsand their own proWt schedules, the participants were giventhe full proWt schedule of the other party. They wereallowed to examine the other party’s proWt schedule for Wveminutes, after which it was collected by the experimenter.Then, without engaging in any actual negotiations, partici-pants proceeded to engage in a test phase identical to thetest phase in the experimental condition.

MeasuresIntegrative experience. Two measures of integrative experi-ence were included:

(1) Experiencing integration: In tasks of the type we usedin the present experiment, i.e., repeated negotiationsin unchanging integrative markets, Wnding the opti-mal integrative solution could be an “aha!” experi-ence, where the proverbial light goes on. Therefore,we chose to measure integrative experience primarilyby classifying participants into those that did (codedas 1) versus those that did not (coded as 0) reach atleast one optimal integrative agreement during themarket session. An optimal agreement was deWned asone for which the pie score, i.e., combined proWts ofboth parties, was $10,400.

(2) Extent of prior integrative experience: Since learningthe optimal agreement might involve a more gradual,accumulative process (rather than an insight type ofexperience), we additionally measured the extent ofeach participant’s integrative experience by comput-ing his or her aggregate pie score for the entire market

(i.e., summing up all his or her agreement pie scores).Hence, participants with higher aggregate pie scoreshad more integrative negotiation experience thanthose with lower aggregate pie scores.

Integrative performance. Our measure of integrative perfor-mance was the counter-oVer response: Integrative perfor-mance in the test phase was measured in terms of thecounter-oVer’s pie score, i.e., the combined proWts of bothparties associated with the counter-oVer they proposed inthe questionnaire. The more integrative (logrolling) thecounter-oVer, the higher the pie score.

Understanding. Two judgmental measures were included tomeasure the diVerent levels of understanding and process-ing the other party’s interests:

(1) The general estimation error: Our measure is arevised version of the common judgmental measureused in several previous studies by Thompson and hercolleagues to measure understanding the other party’sinterests (Thompson, 1991; Thompson & DeHarp-port, 1994). As reported in the Procedure section,when estimating the other party’s proWt schedule, par-ticipants assessed only the lowest and highest valueon each of the issues. We used these values to com-pute the assessed proWt span of the other party foreach issue by subtracting the lowest value from thehighest one. For example, if the participant assessedthe highest possible value (the maximum potentialproWt) for the other party on issue “D” to be $2000and the lowest possible value (the minimum potentialproWt) on that issue to be $0, the assessed proWtpotential span for issue “D” would be $2000 ($2000minus $0). We then computed the diVerence betweenthe other party’s assumed proWt span and the otherparty’s actual proWt span (as determined by theirproWt schedule) for each issue. The sum of thesediVerences across the three issues is the estimationerror, which served as our measure of the accuracy ofthe estimation of the other party’s proWt schedule.It isimportant to note that the critical integrative aspectnegotiators should understand concerning the otherparties’ general proWt schedule is their prioritiesamong the issues. In the present study this wouldmanifest itself by negotiators of Type A estimatingthat for their opponents—i.e., Type B—issue F has asmaller proWt span (and is therefore less important)than issue D. Similarly, negotiators of Type B shouldestimate that for their opponents—i.e., Type A—issueD has a smaller proWt span than issue F (see Appen-dix A). For speciWcally assessing whether participantsunderstood this point, we classiWed them into twogroups: those who did versus those who did not accu-rately assess the other party’s proWt span to besmaller for their least important issue than for theirmost important issue. The correlation between this


measure and the general estimation error wasextremely high (rD .923), suggesting that the latter isalso an appropriate measure of understanding thiscritical integrative aspect. Given the high correlation,using both measures would be redundant, and wechose the general estimation error, which is morecommonly used.

(2) The speciWc estimation error: As explained in theintroduction section, negotiators may acknowledgetheir opponents’ proWt schedules in general terms, butthis does not necessarily mean they can or will applythis knowledge accurately in estimating what theiropponents might gain from diVerent speciWc propos-als. In order to examine the accuracy of applyinginformation about the other party, we included theestimation of the initial oVer’s value for the otherparty as an additional judgmental measure. For eachissue, we computed the absolute diVerence betweenthe actual and the assumed value of the initial oVerfor the other party. The sum of these diVerencesacross the three issues is the speciWc estimation error,which served as our measure of the accuracy ofassessment of the oVer’s value for the other party.

Results

The zero order correlations among the diVerent poten-tial predictors of integrative performance are presented inTable 1. These include the two measures of experience(experiencing integration and the extent of integrative expe-rience), the initial oVer type, and the two estimation errors(the general estimation error and the speciWc oVer estima-tion error).

Not surprisingly, the two measures of understanding—the general estimation error and the speciWc oVer estimationerror—were signiWcantly correlated. Given the theoreticalimportance of distinguishing between these constructs, andsince the correlations between them—although signiW-cant—were not extreme, we continued in our subsequentanalyses to consider both and to assess the unique role ofeach of them.

There was also a signiWcant correlation between the twomeasures of integrative experience—experiencing integra-tion and the extent of integrative experience. As explained inthe Method, the extent of prior integrative experience is ameasure of the negotiator’s aggregate pie score (or joint

proWt score) for the entire previous market, which indicatesthe overall integrative experience that each participantaccumulated throughout the market. Experiencing integra-tion, on the other hand, merely depicts whether the partici-pant did or did not reach at least one optimal integrativeagreement during the previous market session. If discover-ing the optimal integrative solution is typically an insight-ful, “aha!” experience, then this would be a moreappropriate measure of experience. Hence we next exploredwhether or not there was evidence in our market sessiondata to support the “aha!” hypothesis. We primarily testedthis notion by examining whether the probability of reach-ing an optimal agreement signiWcantly increased after pre-viously succeeding in reaching one such agreement.

Initially, we found that with the exception of one partici-pant, all reached at least one agreement during the marketperiod, with an average of 4.8 (SDD2) agreements per par-ticipant. Moreover, on average, 36% (SDD 0.3) of the par-ticipants’ agreements were optimal, with most participants(73%, ND62) reaching at least one optimal agreement dur-ing the market. For each of the participants that reached atleast one optimal agreement, we computed the proportionof optimal agreements that he or she reached after accom-plishing the Wrst optimal agreement (excluding ten partici-pants whose Wrst optimal agreement was also their lastagreement). In line with the “aha!” notion, a majority ofparticipants (56%; 29 of the 52) always reached optimalagreements after reaching their Wrst optimal one, suggestingthat an insight type of experience is highly plausible.

In our subsequent analyses we therefore regard the expe-riencing integration measure, which corresponds to aninsight type of experience as our primary measure of inte-grative experience. However, given the theoretical impor-tance of distinguishing between the two types of learning—insight versus gradual, and since we cannot rule out thepossibility that accumulative gradual learning is alsoapparent, we brieXy report results regarding the extent ofintegrative experience measure as well.

We began by examining whether experiencing integra-tion promoted understanding of the other parties’ interests.Comparing the understanding measures of participantsthat did (coded as 1) versus those that did not (coded as 0)reach an optimal integrative solution during the marketsession revealed that experiencing integration did to somedegree promote understanding. In line with H1, the generalestimation error was smaller for participants who did expe-

Table 1Correlations among potential predictors of integrative performance

Number of observations: 78 (case wise deletion of missing data).¤ p < .05.

Experiencing integration Extent of integrative experience Initial oVer type General est. error SpeciWc oVer est. error

Experiencing integration 1.00Extent of integrative experience 0.531¤ 1.00Initial oVer type 0.102 0.040 1.00General est. error ¡0.175 ¡0.131 0.047 1.00SpeciWc oVer est. error 0.038 ¡0.052 0.068 0.496¤ 1.00


rience integration (MD3177, SDD2392) than for thosewho did not (MD 4145, SDD1722), t (79)D1.74, p < .05,one tail. The speciWc oVer estimation error, however, wasnot signiWcantly diVerent for these two groups (MD2445,SDD 2081 versus MD2452, SDD1352, respectively),t (82)D0.01, pD .99. Hence, experiencing integrationimproved the ability to assess the other parties’ general pri-orities but not to evaluate the other parties’ outcomes forspeciWc oVers.

We subsequently examined the inXuence of the extent ofprevious integrative experience on understanding the otherparties’ interests. Two separate general linear models fortesting the eVect of the aggregate pie score (as the indepen-dent variable) on each of the understanding measuresrevealed no signiWcant eVects. Hence, having a greaterextent of integrative experience did not promote under-standing of the other party.

We next explored the roles of integrative experience andof the two types of understanding in improving integrativeperformance. With regard to this question we particularlywanted to determine whether experience promoted integra-tive responses at the initial negotiation stage, whetherunderstanding enhanced integrative behavior beyond theeVect of previous integrative experience, and whetherimproved understanding was a necessary condition formore integrative behavior.

We addressed these questions by means of a general lin-ear model for predicting counter-oVers (i.e., the counter-oVer pie) from experiencing integration (i.e., whether partic-ipants did or did not experience integration, coded as 1 or 0,respectively), the two measures of understanding (i.e., thegeneral estimation error of the other party’s proWt scheduleand the speciWc oVer estimation error of the other parties’oVer outcomes), and the initial oVer type (coded as 0 fordistributive and 1 for integrative).

Results of this analysis are reported in Table 2. As can beseen in the table, the signiWcant predictors of counter-oVerintegrativeness were experiencing integration, type of initialoVer, and the speciWc oVer estimation error. The generalestimation error (i.e., the assessment of the other parties’general proWt schedule) did not qualify as a signiWcant pre-dictor.

A parallel analysis using the extent of previous integra-tive experience measure (i.e., the aggregate agreement piescore) instead of the experiencing integration measure

Table 2Results of general linear model for predicting counter-oVer pie

Dependent Variable: Counter-oVer pie, R Squared D .226, F (4, 4) D 5.4,p < .001.

Parameter B t P Partial eta squared

Intercept 8745.831 28.62 .00 .917Experiencing integration 741.301 2.86 .00 .100Initial oVer type 559.547 2.51 .01 .079General est. error ¡1.086E¡02 ¡.19 .85 .000SpeciWc oVer est. error ¡.130 ¡1.94 .05 .049

revealed a similar pattern of results (although the speciWcoVer estimation was only marginally signiWcant, p < .1).4

We next turned to asses the role of information as adeterminant of the increase in integrative responses. Thiswas tested by comparing the counter-oVers of the experi-enced participants in the experimental condition with thoseof the fully informed but inexperienced participants in theexternal control condition. We found that the meancounter-oVer pie of the original experienced group was sig-niWcantly higher than the mean oVer pie of the controlgroup (MD9186, SDD1062 versus MD8545, SDD715),t (128)D 3.6, p < .001. Hence, complete information withoutexperience in an integrative market does not seem to fosteran overall increase in the tendency to respond with integra-tive counter-oVers. This result suggests that acquiring moreinformation about the other party’s priorities is not suY-cient and cannot fully account for the higher degree of log-rolling we observed among experienced negotiators.

To summarize, in the Wrst experiment we found thatexperiencing integration improved the ability to assess theother parties’ general priorities (in line with hypothesis H1)but not the ability to evaluate the other parties’ outcomesfor speciWc oVers. Moreover, the extent of previous integra-tive experience did not signiWcantly inXuence understand-ing. We also found that, as predicted in hypothesis H2,integrative negotiation experience, and the extent of thisexperience, led to an overall increase in logrolling responsesalready noticeable at the initial negotiation stage. Thisenhancing eVect was apparent even when controlling foreVects of understanding (in line with H4) and of informa-tion. In addition, while understanding the other party’sgeneral interests did not signiWcantly contribute to the ten-dency to respond in an integrative manner (contrary tohypothesis H3a), being able to correctly assess the speciWcvalues of the diVerent oVers to the other party did contrib-ute (providing support for H3b, although the eVect was notlarge).

Discussion

Results of the Wrst experiment imply that much of theimprovement in integrative negotiation performance in anunchanging environment can be accounted for by experi-

4 An alternative explanation for the null eVect of general understandingcould be that participants who had, even if by accident, come to recognizethe “best solution” during the experience phase assumed that this was theexpected “correct response” in the test phase as well. In that case, theeVects of understanding would necessarily be bounded. To test for thispossibility, we ran an additional general linear model, in which we exam-ined the eVects of the two understanding measures on performance (i.e., onthe counter-oVer pie), excluding all participants that gave the optimal “1-5-9” solution as their counter-oVer. The pattern of results for the remain-ing participants (59 out of 86) did not diVer from that obtained with thefull sample. As before, the speciWc oVer estimation was signiWcant,F1,56 D 5.3, p < .03, while the general proWt schedule error was not. Hencedemand characteristics do not seem to be the explanation for the insigniW-cant eVect of understanding general priorities.


ence, most likely an insightful, “aha!” type of experience,rather than by merely understanding the other party’s pri-orities and values. Moreover, they indicate that thisimprovement in integrative performance is already appar-ent at the initial stage of the negotiation. Indeed, experiencepromoted not only integrative performance but also assess-ment of the other parties’ general priorities. However, thefact that integrative experience enhanced integrative per-formance, even when controlling for the eVects of accuracyin assessing the other party’s interests, implies that under-standing the other party is not a necessary condition for theimprovement in integrative behavior.

Results further suggest that while being able to correctlyestimate the other parties’ general priorities is not suYcientfor improving integrative behavior, increased accuracy inassessing the speciWc values of the diVerent oVers for theother party (i.e., accurate processing of the information) isbeneWcial. Hence, although the relationship between experi-ence and integrative performance in an unchanging settingdoes not seem to be mediated by understanding the oppo-nents’ general priorities or their speciWc outcomes, the lattermay have a signiWcant incremental eVect on integrativenegotiations.

It is noteworthy, however, that the way in which wesolicited understanding and performance measures in thisstudy constrains its conclusions. Participants were Wrstasked to provide initial oVer estimates, then to makecounter-oVers, and Wnally to assess general proWt schedules.Given this order, a possible explanation for the Wnding thatinitial oVer estimates predicted counter-oVers while generalproWt schedule estimations did not, could be that theformer were provided before making counter-oVers, whilethe latter were not. In other words, participants could haveadjusted their counter-oVers to match the initial oVer esti-mates they already made. However, it seems equally likelythat participants could also have adjusted their generalproWt schedule estimates to their preceding counter-oVerresponses. In that case one would expect the correlationbetween initial oVer and its preceding estimate (i.e., speciWcoVer estimation), and the correlation between initial oVerand its succeeding estimate (i.e., general proWt schedule esti-mation) to be similarly high. Notwithstanding, given thecorrelational nature of this study, which limits the causalconclusions that can be inferred, we continued to examinethe notion that understanding the opponents’ speciWc out-comes has a greater inXuence on integrative performancethan understanding their general priorities. This was donein a second study, in which we employed a controlledexperiment.

Experiment 2

In the second experiment we further assessed the impor-tance of distinguishing between two levels of understandingthe opponent’s interests: (a) estimating the opponent’s gen-eral interest structure, and (b) estimating the opponent’soutcomes for speciWc oVers. Results of the Wrst experiment

suggest that acknowledging the other party’s proWt sched-ule in general terms is not suYcient for achieving anincrease in integrative responses. However, applying thisknowledge accurately in order to estimate what the oppo-nent might gain from diVerent proposals, seems to beimportant and to signiWcantly contribute to the tendency torespond in an integrative manner. The second experimentsought to further examine the impact of considering theopponent’s outcomes for speciWc oVers, using a controlledexperimental manipulation.

In addition, in Experiment 1 we found that accuratelyassessing the other parties’ speciWc oVer outcomes canimprove performance within an unchanging environment.In Experiment 2, we tested whether such increased cogni-tive processing could also enhance performance in a trans-fer task. Appropriate transfer of acquired skills to a novelcontext is commonly assumed to indicate better under-standing and successful learning of general principles, i.e.,expertise. Indeed, there is a growing tendency in the litera-ture to explore learning and evaluate training by examiningtransferability of skills to new contexts (e.g., Day, Arthur, &Gettman, 2001; Ford, Smith, Weissbein, Gully, & Salas,1998; Singley & Anderson, 1989). This tendency is alsonoticeable in the literature on negotiations (e.g., Bereby-Meyer et al., 2004; Loewenstein et al., 1999; Thompson,1990c). These studies have typically shown that althoughgaining expertise and transferring learned negotiation skillsis possible, it is not easily achieved. Examples of eVectivemechanisms and training schemes that have been proposedare analogical encoding and motivating learning by provid-ing learning goals and values. The present study aimed toinvestigate the eVectiveness of another training mecha-nism—focusing the learner on considering the other party’sspeciWc outcomes throughout the negotiation process.

In an initial training negotiation task we manipulatedthe negotiators’ focus on the other parties’ interests—focusing them either on their opponents’ general intereststructure or on their opponents’ speciWc oVer outcomes. Wethen examined their performance and understanding of anew task, which was diVerent and more complex than thetraining task. Consistent with Wndings on role reversalwhich suggest that having negotiators verbalize eachother’s viewpoint is beneWcial and increases the likelihoodof resolution (e.g., Ball, Bazerman, & Carroll, 1991), weexpect training for perspective taking to be eVective.

Based on the Wndings in our Wrst experiment, we hypoth-esize that:

H5. Negotiators trained to focus on assessing their oppo-nents’ outcomes for speciWc oVers will achieve more inte-grative outcomes in a transfer task than negotiators trainedto focus on understanding their opponents’ general proWtschedules.

H6. Negotiators trained to focus on speciWc outcomes willassess the other party’s speciWc outcomes for diVerent oVersmore accurately than negotiators trained to focus on gen-eral proWt schedules.


Method

ParticipantsThirty-six students participated in the experiment. We

recruited the students via posted advertisements oVering apayment of $12 for participation in an experiment on nego-tiations. Upon signing up, participants were randomlyassigned to one of two experimental conditions.

ProcedureThe experiment had two between-subject conditions that

were conducted separately: A general interest training con-dition and a speciWc oVer training condition. Both condi-tions included two stages. The Wrst was a training stage, inwhich participants repeatedly negotiated the same integra-tive task. The second stage was a test stage, in which partic-ipants negotiated a new integrative task.

In the Wrst stage of both conditions, we randomlyassigned participants to the role of buyer or seller in a nego-tiation market. We then gave each of them an informationpackage which included instructions, a description of thetask and of the market procedure, individual proWt sched-ules, and negotiation report forms. They were told that dur-ing the market period, representatives of buying and sellingcompanies would engage in three successive two-partynegotiations, with diVerent opponents each time. The nego-tiation task was similar to the one used in our Wrst experi-ment. It included three issues (delivery terms, discount level,and terms of payment), and the values of the “individual”proWt schedules were identical to those in Experiment 1 (seeAppendix B). As can be seen in the Table, delivery and pay-ment terms were logrolling issues, and discount level was adistributive issue. Hence, an optimal integrative agreementcould be reached when parties logrolled and agreed ontransaction terms of level 1 for delivery terms (60 days), andlevel 9 for payment terms (9 payments).

For each negotiation, participants were given a detailedreport form to Wll out during the negotiation process. In theseforms participants were required to report each oVer thatwas exchanged during the negotiation, specifying the partywho made the oVer (buyer/seller) and the levels oVered oneach issue. The negotiator’s focus on the other party’s inter-ests was manipulated by varying the report requirements inthe two experimental conditions, in the following way: In thegeneral interests training condition, after every oVer, partici-pants were requested to provide their estimation of the otherparty’s general proWt schedule, by assessing the lowest andhighest possible proWt for him or her on each of the issues. Inthe speciWc oVer training condition, for every oVer, partici-pants were requested to report—in addition to their ownproWts for each of the three issues—their assessment of theoVer’s value for their opponent, for each of the three issues.In both conditions, when the negotiation ended participantswere requested to report whether an agreement was accom-plished and if so to specify its terms.

Prior to the market, participants were given 10 min forreading instructions and proWt schedules. Before each of the

three negotiations, the experimenter randomly matchedeach seller (buyer) with a diVerent buyer (seller) and thesedyads were then given 10 min for negotiating. When thethird negotiation period terminated, the experimenter col-lected all materials from the Wrst stage and gave partici-pants a new package with materials for the second stage.

In the second stage participants were assigned roles in anew negotiation task. In this task, representatives of a realestate development company and of a city council negoti-ated four issues concerning a community development pro-ject (adapted from Thompson, Peterson, & Brodt, 1996).Participants were given individual proWt schedules thatindicated their utility points on each of the issues (seeAppendix C). As can be seen in the Appendix, among thefour issues to be negotiated, payment terms and number ofsub-contractors were logrolling issues, retail space was acompatible issue (no conXict of interest was apparent sinceboth parties preferred larger retail space), and open spacewas a distributive issue. This task diVered from the Wrst oneon several dimensions, including context, number of issues,number of alternatives for each issue, and potential integra-tive strategies (compatible issues in addition to logrolling).

An optimal integrative agreement could be reached byagreeing on seven payments (level 5), four sub-contractors(level 1), and 6000 m2 of retail space (level 5). In such anagreement negotiators could collectively gain a total of15,300 points.

In this second task only one negotiation took place.Before negotiating, the experimenter matched dyads thathad not previously negotiated with each other during stage1, and they were allowed 15 min to negotiate. At the end ofthe negotiation (when time was up or agreement wasreached), participants Wlled out a “transaction form” inwhich they reported whether an agreement was achieved,and if so, detailed the four terms they agreed upon.

At the end of this negotiation, as in Experiment 1, eachparticipant was given an additional page of instructions, inwhich he or she was requested to presume that he or shewas now negotiating the identical task with another oppo-nent. Participants were then presented with a single initialoVer, supposedly forwarded by this new opponent. As inthe Wrst Experiment, two oVer types were manipulatedbetween subjects, one integrative and one distributive. Theoverall value of the two types of oVers, from the partici-pants’ perspective, was held constant. However, for theother party, the integrative oVer was worth more pointsthan the distributive oVer. As can be seen in Appendix C,which shows the initial oVers that were submitted, the dis-tributive and the integrative oVers forwarded to the realestate development company representatives (“3-3-3-3”and “4-3-1-2,” respectively) were both worth a total valueof 5250 points for their recipient. However, for the oppo-nent (the city council representative who supposedly for-warded the oVer), the integrative oVer was worth 7050points, while the distributive one was worth only 5250points. The same was true with respect to the two oVertypes submitted to recipients representing the city council.


Upon receiving the initial oVer, participants were askedto: (a) assess the initial oVer’s value for the initiator on eachof the four issues, and (b) estimate the other party’s generalproWt schedule, by assessing the lowest and highest valuefor him or her on each of the four issues.

Results

We Wrst assessed our training manipulation by examin-ing whether over the three training negotiations, partici-pants’ improved in their assigned tasks in each condition—i.e., in estimating the other parties’ speciWc oVer values inthe speciWc oVer training condition and general interests inthe general interests training conditions, respectively. Asnegotiations varied in the number of oVers exchanged, weused the decrease in the estimation errors that were madefollowing the Wrst oVer exchange in each of the three con-secutive negotiation rounds as a criterion for learning. Inboth conditions the mean estimation error decreased fromthe Wrst to the third negotiation. t Tests conWrmed that themean errors in the third negotiations were signiWcantlysmaller than in the Wrst negotiations for the speciWc trainingcondition (MD 2219, SDD 1848 versus MD 3244,SDD1282), t (15)D1.8, p < .05 (one tail), and also for thegeneral interests training condition (MD3126, SDD2390versus MD4800, SDD1045), t (14)D2.85, p < .005 (onetail). These results imply that our manipulation was eVec-tive in training the participants as intended.

Although in this research we are primarily interested inparticipants’ performance in the transfer task, we also ana-lyzed the data on performance in the training phase. Giventhe low incidence of optimal agreements that were achievedthroughout the training session: 4 out of 24 agreements(17%) in the general interests training condition and 2 out of20 (10%) in the speciWc oVer training condition, we continuedby focusing on the mean pie size rather than on optimalagreements. We analyzed the mean pie size as a function ofnegotiation number and found that although the mean pie inboth conditions increased over the three training negotia-tions, this increase was not statistically signiWcant. This insig-niWcant increase in performance is not surprising, given thatthis was a training phase in which participants were requiredto focus on learning rather than on performance. Indeed,prior research on achievement motivation goals found themain beneWts of learning goals to be apparent in transfertasks that follow training, rather than in the speciWc trainingtasks themselves (e.g., Bereby-Meyer et al., 2004).

Turning to our focal interest in the current study, wenext examined the participants’ performance in the transfertask. Our main hypothesis (H5) was that negotiatorstrained to focus on assessing their opponents’ outcomes forspeciWc oVers will achieve more integrative outcomes in atransfer task than negotiators trained to focus on under-standing their opponents’ general proWt schedules. In sup-port of this hypothesis, we found the mean combinedagreement points to be signiWcantly higher in the speciWcoVer training condition (MD 14,325, SDD959) than in the

general interests training condition (MD13,440, SDD 718),t (16)D 2.2, p < .04.5

We next turn to analyze the responses to the question-naire, beginning with the general proWt schedule estimationand continuing with the speciWc oVer estimation. Althoughparticipants in the general estimation condition weretrained to estimate the general priorities of the other party,the mean general estimation error in that group was greater(MD4308, SDD1666) than in the speciWc training condi-tion (MD3181, SDD 2049), and this diVerence approachedsigniWcance, t (32)D 1.8, pD .09. The diVerence between themean speciWc oVer estimation errors in the two trainingconditions, although also not signiWcant, was in the pre-dicted direction; lower in the speciWc training conditionthan in the general training one (MD2416, SDD 1432 andMD2988, SDD1510, respectively), t (34)D1.16, pD .3.

Although the estimation errors of the two groups werenot signiWcantly diVerent, only participants in the speciWctraining condition correctly assessed the integrative oVer tobe signiWcantly more valuable to the other party than the dis-tributive one. While in this condition, value assessments ofthe distributive and integrative initial oVers were signiWcantlydiVerent from each other (MD5261, SDD870 andMD6400, SDD937), t (14)D2.51, p < .03, in the generaltraining condition participants’ estimations of the two typesof oVers did not signiWcantly diVer from each other(MD5459, SDD715 for the distributive oVer and MD5989,SDD1177 for the integrative oVer), t (18)D1.24, pD .2.Hence, providing some support for hypothesis H6, negotia-tors who focused on speciWc outcomes more than those whofocused on general proWt schedules understood the crucialaspect of integrative oVers—i.e., that integrative oVers areworth more to the other party than are distributive oVers.

Discussion

Results of the second experiment strengthen and extendWndings of our Wrst experiment. First, they provide additionalevidence for the importance of distinguishing between judgingthe other party’s general priorities and assessing his or herspeciWc oVer outcomes. Second, they support the notion thatconsidering the opponent’s interests by assessing his or herspeciWc oVer outcomes during the negotiation process may bemore eVective for integrative negotiation than merely assess-ing the other party’s general interest structure, without neces-sarily taking it into consideration when exchanging oVers.These Wndings expand those of our Wrst experiment, by show-ing that such cognitive processing has a positive eVect onnegotiation outcomes and not only on initial negotiationresponses. Finally, they also suggest that it is beneWcial totrain negotiators to consider their opponents’ interests (i.e., totake their opponents’ perspective) throughout the negotiationprocess, rather than simply recognize them. This is possible

5 All dyads reached an agreement. Among the 10 dyads in the general es-timation condition, one reached the optimal solution. Among the eight inthe speciWc oVer estimation condition, two reached the optimal solution.


and eVective not only for improving performance in the train-ing task, but for other integrative negotiations as well.

General discussion and conclusions

Although much has been written about integrative negoti-ation, knowledge is still insuYcient regarding the psychologi-cal and cognitive determinants of integrative solutions. Thedistinction we propose between negotiators’ understanding ofthe opponents’ general proWt schedules versus their gains forparticular oVers relates to this question. From a methodologi-cal point of view, the diVerential eVectiveness of these twojudgment measures suggests that future research should simi-larly adopt a range of understanding measures reXectingdiVerent levels and forms of processing. In addition to thesetheory-testing and methodological contributions, our Wndingsarticulate aspects of negotiator training that can increase thelikelihood of achieving integrative outcomes.

The Wndings of the present research go beyond thealready-established eVect of experience on Wnal negotiationagreements (Bazerman et al., 1985; Neale & Bazerman,1991; Thompson, 1990a, 1990c; Thompson & DeHarpport,1994; Thompson & Hastie, 1990), by showing that theincreased tendency of the negotiators to logroll as theygained more experience was apparent, not only in the Wnalagreements that they reached, but also during the initialstages of their transactions.

While many earlier studies emphasized the key role ofunderstanding the interests of the opponent in achievingmore integrative performance over time (Thompson, 1990a,1990c; Thompson & DeHarpport, 1994; Thompson & Has-tie, 1990), the present Wndings suggest that at least in anunchanging environment, understanding is not a necessarycondition. This conclusion is supported in several ways.First, the eVects of repeated experience on understandingwere quite limited. Although experience did to some degreeimprove the estimations of the other party’s general intereststructure, it did not improve assessments of the otherparty’s speciWc oVer outcomes. Second, the improvement inunderstanding the other party’s general interest structuredid not lead to more integrative counter-oVers. Lastly,although improved understanding of the opponent’s spe-ciWc oVer outcomes did contribute to logrolling responses,integrative experience signiWcantly enhanced integrativecounter-oVer responses, even when controlling for the eVectof such improved understanding. Hence, even though accu-rately understanding the other party’s speciWc outcomescan advance logrolling responses, such understanding doesnot seem to be a necessary condition for advancing integra-tive responses of negotiators with experience in a particulartask. This Wnding is in line with the notion put forward byNeale and Bazerman (1991), as well as by Thompson(1990b, 1990c), that experience does not necessarily lead toexpertise. Indeed, in repeated negotiations with a particulartask, negotiators might learn a particularly eVective strat-egy, without comprehensively understanding when andwhy this strategy is eVective (i.e., expertise).

The Wnding that improved assessment of the other party’sinterests in general did not enhance integrative performance,while accurate assessment of the other party’s speciWc oVervalues did, suggests that considering and accurately process-ing available information about the opponent is a signiWcantdeterminant of successful integrative negotiations. Thisappears to be consistent with the motivated information pro-cessing model of negotiation developed by De Dreu and Car-nevale (2003). Their model underlines the importance ofinducing motivation for engaging in deep and systematicprocessing of information that is or becomes available duringthe negotiation, for achieving more integrative agreementsand revisions of Wxed-pie perceptions.

A distinction between availability of information andprocessing of information was also suggested by Pinkleyet al. (1995). They found that while full information raisedthe probability of negotiating integrative agreements, it wasnot suYcient to ensure achieving them. Although informa-tion processing was not directly manipulated or measuredin their study, they argue that inferior performance of fullyinformed negotiators can be explained by, and should beattributed to information processing errors.

In the present study, we directly measured and manipu-lated information processing, thereby providing more directevidence for the importance of accurate information process-ing. In addition, our study addresses a unique form of infor-mation processing. We propose that processing informationabout the other parties’ general priorities by assessing theirspeciWc values for the diVerent oVers throughout the negotia-tion process—i.e., taking into consideration the opponents’perspective—is beneWcial and improves integrative perfor-mance and outcomes. This is consistent with previous Wnd-ings that negotiators’ attention to their opponents’ goals andexpectations, and their ability to adopt the opponents’ per-spective, is a key to successful negotiation.

Finally, regarding implications for negotiation training,our Wndings imply that simply allowing negotiators toexperience integrative negotiations can enhance their ten-dency to engage in more integrative behavior in repeatednegotiations within an unchanging environment. They alsoindicate that for achieving more, it is not suYcient to teachnegotiators to consider the other parties’ general intereststructure. Merely acknowledging the other parties’ inter-ests, without additional cognitive processing activity of thelearners, is not enough. It is important that trainees alsolearn to infer and estimate the other parties’ speciWc out-comes. Training negotiators to do so, i.e., to consider theiropponents’ interests by considering their speciWc outcomesthroughout the negotiation process, was shown to be possi-ble and eVective for improving performance, not only in thetraining task but in other integrative negotiations as well.These results complement several other studies in the nego-tiation learning literature, which show that although gain-ing expertise and being able to transfer learning innegotiations may not be easily achieved, it is possible (e.g.,Bereby-Meyer et al., 2004; Loewenstein et al., 1999; Mannixet al., 1991; Thompson, 1990c). The present research pro-


vides encouraging evidence concerning an additional train-ing method to those previously proposed (such asanalogical reasoning, learning goals, etc.) for achieving suchexpertise.

We end by noting some limitations to our research. First,our sample consisted of students without previous negotia-tion experience. Future research should further explore therole of experience and of understanding by examining expe-rienced negotiators as well. A second limitation is that thenegotiation tasks we used were artiWcial, and negotiators’utility functions were predetermined by the experimenters,rather than self-generated. Although this is a commonlyused method in the negotiation literature, it constrains gen-eralizability. Notwithstanding these limitations, we believethis research has potentially important implications inhighlighting the importance, both theoretical and methodo-

logical, of diVerentiating between levels and forms of cogni-tive processing and understanding. This researchspeciWcally contributes to the literature on social cognitionand perspective taking in interpersonal contexts such asnegotiations. It shows that mere acknowledgement of oth-ers’ general interests may not be suYcient, and that theadditional inferential step of assessing their speciWc out-comes is important and can be advantageous. Finally, froma practical point of view, our Wndings convey aspects ofnegotiator training that may increase the likelihood ofachieving integrative outcomes.

Acknowledgment

This research was supported by the Israel Science Foun-dation (Grant No. 774/03).

Appendix A. Experiment 1: ProWt schedules and initial oVer types (adapted from Bazerman et al., 1985)

“Party a” ProWt Schedule: ($)

Level Issue D Issue R Issue F

1 0 0 02 200 300 5003 400 600 10004 600 900 15005 800 1200 20006 1000 1500 25007 1200 1800 30008 1400 2100 35009 1600 2400 4000

Initial oVer Types forwarded to “Type A” participants ProWts for recipient (party a) ProWts for other party (party b)

Distributive: 5-6-6 ($800 + $1500 + $2500 D $4800) ($2000 + $900 + $600 D $3500)Logrolling: 2-3-9 ($200 + $600 + $4000 D $4800) ($3500 + 1800 + $0 D $5300)

“Party b” ProWt Schedule: ($)

Level Issue D Issue R Issue F

1 4000 2400 16002 3500 2100 14003 3000 1800 12004 2500 1500 10005 2000 1200 8006 1500 900 6007 1000 600 4008 500 300 2009 0 0 0

Appendix B. Experiment 2—stage 1: ProWt schedules (adapted from Bazerman et al., 1985)

Initial oVer Types forwarded to “Type B” participants ProWts for recipient (party b) ProWts for other party (party a)

Distributive: 4-4-5 ($2500 + $1500 + $800 D $4800) ($600 + $900 + $2000D $3500)Logrolling: 1-7-8 ($4000 + $600 + $200 D $4800) ($0 + $1800 + $3500 D $5300)

Buyers’ ProWt Schedule

Delivery Terms (No. of days for delivery) Discount Level Payment Terms

Level Alternative ProWt Level Alternative (%) ProWt Level Alternative ProWt

1 60 days $0 1 0 $0 1 Cash $02 55 days $200 2 1 $300 2 2 payments $500


Sellers’ ProWt Schedule

Appendix B (continued)



3 50 days $400 3 2 $600 3 3 payments $10004 45 days $600 4 3 $900 4 4 payments $15005 40 days $800 5 4 $1200 5 5 payments $20006 35 days $1000 6 5 $1500 6 6 payments $25007 30 days $1200 7 6 $1800 7 7 payments $30008 25 days $1400 8 7 $2100 8 8 payments $35009 20 days $1600 9 8 $2400 9 9 payments $4000

Appendix C. Experiment 2—stage 2: ProWt schedules (adapted from Thompson et al., 1996)



1 60 days $4000 1 0 $2400 1 Cash $16002 55 days $3500 2 1 $2100 2 2 payments $14003 50 days $3000 3 2 $1800 3 3 payments $12004 45 days $2500 4 3 $1500 4 4 payments $10005 40 days $2000 5 4 $1200 5 5 payments $8006 35 days $1500 6 5 $900 6 6 payments $6007 30 days $1000 7 6 $600 7 7 payments $4008 25 days $500 8 7 $300 8 8 payments $2009 20 days $0 9 8 $0 9 9 payments $0

Real Estate Development Company’s ProWt Schedule2
Payment terms (no. of payments) Retail space (m ) Number of subcontractors Open space
Level Option Points Level Option Points Level Option Points Level Option (%) Points

1 3 500 1 0 100 1 4 200 1 30 502 4 1500 2 1500 700 2 3 600 2 25 2503 5 2500 3 3000 1300 3 2 1000 3 20 4504 6 3500 4 4500 1900 4 1 1400 4 15 6505 7 4500 5 6000 2500 5 0 1800 5 10 850

Initial oVer Types forwarded to “real estate development company” Points for recipient (real estate) Points for other party (City Council)

Distributive: 3-3-3-3 2500 + 1300 + 1000 + 450 D 5250 1000 + 1300 + 2500 + 450D 5250Integrative: 4-3-1-2 3500 + 1300 + 200 + 250 D 5250 600 + 1300 + 4500 + 650 D 7050

City Council’s ProWt Schedule2
Payment terms (no. of payments) Retail space (m ) Number of subcontractors Open space
Level Option Points Level Option Points Level Option Points Level Option (%) Points

1 3 1800 1 0 100 1 4 4500 1 30 8502 4 1400 2 1500 700 2 3 3500 2 25 6503 5 1000 3 3000 1300 3 2 2500 3 20 4504 6 600 4 4500 1900 4 1 1500 4 15 2505 7 200 5 6000 2500 5 0 500 5 10 50

Initial oVer Types forwarded to “City Council” Points for recipient (City Council) Points for other party (real estate)

Distributive: 3-3-3-3 1000 + 1300 + 2500 + 450 D 5250 2500 + 1300 + 1000 + 450 D 5250Integrative: 5-3-2-4 200 + 1300 + 3500 + 250 D 5250 4500 + 1300 + 600 + 650 D 7050

References

Ball, S. B., Bazerman, M. H., & Carroll, J. S. . (1991). An evaluation oflearning in the bilateral winner’s curse. Organizational Behavior andHuman Decision Processes, 48, 1–22.

Bazerman, M. H. (2001). Judgment in managerial decision making (5th ed.).New York: John Wiley & Sons.

Bazerman, M. H., & Carroll, J. S. . (1987). Negotiator cognition. Researchin Organizational Behavior, 9, 247–288.

Bazerman, M. H., Magliozzi, T., & Neale, M. A. . (1985). Integrative bar-gaining in a competitive market. Organizational Behavior and HumanDecision Processes, 35, 294–313.

Bazerman, M. H., & Neale, M. A. . (1982). Improving negotiation eVective-ness under Wnal oVer arbitration: The role of selection and training.Journal of Applied Psychology, 67, 543–548.


Bazerman, M. H., & Neale, M. A. . (1983). Heuristics in negotiation: Limi-tations to dispute resolution eVectiveness. In M. H. Bazerman & R. J.Lewicki (Eds.), Negotiating in organizations. Beverly Hills, CA: Sage.

Bazerman, M. H., & Neale, M. A. (1992). Negotiating rationally. NewYork: The Free Press.

Benton, A. A., Kelley, H. H., & Liebling, B. (1972). EVects of extremity ofoVers and concession rate on the outcomes of bargaining. Journal ofPersonality and Social Psychology, 24, 73–83.

Bereby-Meyer, Y., Moran, S., & Unger-Aviram, E. (2004). When perfor-mance goals deter performance: Transfer of skills in integrative negoti-ations. Organizational Behavior and Human Decision Processes, 93,142–154.

Carroll, J. S., Bazerman, M. H., & Maury, R. . (1988). Negotiator cognitions:A descriptive approach to negotiators’ understanding of their opponents.Organizational Behavior and Human Decision Processes, 41, 352–370.

ChertkoV, J. M., & Conley, M. . (1967). Opening oVer and frequency ofconcessions as bargaining strategies. Journal of Personality and SocialPsychology, 7, 181–185.

Dawes, R. M., & Corrigan, B. (1974). Linear models in decision making.Psychological Bulletin, 81, 95–106.

Day, E. A., Arthur, W. J., & Gettman, D. (2001). Knowledge structures andthe acquisition of a complex skill. Journal of Applied Psychology, 86,1022–1033.

De Dreu, C. K. W., & Carnevale, P. J. D. (2003). Motivational bases forinformation processing and strategy in conXict and negotiation. In M.P. Zanna (Ed.), Advances in experimental social psychology (Vol. 35, pp.235–291). New York: Academic Press.

De Dreu, C. K. W., Koole, S. L., & Steinel, W. (2000). UnWxing the Wxedpie: A motivated information processing approach to integrative nego-tiation. Journal of Personality and Social Psychology, 79, 975–987.

Ford, J. K., Smith, E. M., Weissbein, D. A., Gully, S. M., & Salas, E. .(1998). Relationships of goal orientation, meta-cognitive activity, andpractice strategies with learning outcomes and transfer. Journal ofApplied Psychology, 83, 218–233.

Galinsky, A. D., & Mussweiler, T. (2001). First oVers as anchors: The roleof perspective-taking and negotiator focus. Journal of Personality andSocial Psychology, 81, 657–669.

Gentner, D., Loewenstein, J., & Thompson, L. (2003). Learning and trans-fer: A general role for analogical encoding. Journal of Educational Psy-chology, 95, 393–408.

Hammond, K., & Grassia, J. (1985). The cognitive side of conXict: From the-ory to resolution of policy disputes. In S. Oskamp (Ed.), Applied socialpsychology annual (Vol. 6, pp. 233–254). Beverly Hills, CA: Sage.

Kelley, H. H. (1966). A classroom study of the dilemmas in interper-sonal negotiations. In K. Archibald (Ed.), Strategic interaction andconXict. Berkeley, CA: Institute of International Studies, Universityof California.

Liebert, R. M., Smith, W. P., Hill, J. H., & KeiVer, M. (1968). The eVects ofinformation and magnitude of initial oVer on interpersonal negotia-tion. Journal of Experimental and Social Psychology, 4, 431–441.

Loewenstein, J., & Thompson, L. (2000). The challenge of learning. Negoti-ation Journal, 16, 399–408.

Loewenstein, J., Thompson, L., & Gentner, D. (1999). Analogical encodingfacilitates knowledge transfer in negotiation. Psychonomic Bulletin andReview, 6, 586–597.

Mannix, E. A., Northcraft, G. B., & Neale, M. A. (1991). Integrative anddistributive negotiation training and experience: A competitive advan-tage? Working Paper, Northwestern University.

Moran, S., & Ritov, I. (2002). Initial perceptions in negotiations: Evalua-tion and response to “logrolling” initial oVers. Journal of BehavioralDecision Making, 15, 101–124.

Nadler, J., Thompson, L., & Van Boven, L. (2003). Learning negotiationskills: Four models of knowledge creation and transfer. ManagementScience, 49, 529–540.

Neale, M. A., & Bazerman, M. H. (1983). The role of perspective takingability in negotiating under diVerent forms of arbitration. Industrialand Labor Relations Review, 36, 378–388.

Neale, M. A., & Bazerman, M. H. (1991). Cognition and rationality in nego-tiation. New York: The Free Press.

Neale, M. A., & Northcraft, G. B. (1990). Experience, expertise and deci-sion bias in negotiation: The role of strategic conceptualization. In B.Sheppard, M. Bazerman, & R. Lewicki (Eds.), Research on negotiationsin organizations (Vol. 2). Greenwich, CT: JAI Press.

Pinkley, R. L., GriYth, T. L., & Northcraft, G. B. (1995). “Fixed-pie” a lamode: Information availability, information processing, and the nego-tiation of sub-optimal agreements. Organizational Behavior and HumanDecision Processes, 62, 101–112.

Pruitt, D. G. (1983). Achieving integrative agreements. In M. H. Bazer-man & R. J. Lewicki (Eds.), Negotiating in organizations. BeverlyHills, CA: Sage.

Pruitt, D. G., & Lewis, S. A. (1975). Development of integrative solutionsin bilateral negotiation. Journal of Personality and Social Psychology,31, 621–633.

RaiVa, H. (1982). The art and science of negotiation. Cambridge, MA:Bellknap.

Rubin, J. Z., & Brown, B. R. (1975). The social psychology of bargainingand negotiation. New York: Academic Press.

Samuelson, W. F., & Bazerman, M. H. (1985). Negotiating under the win-ner’s curse. In V. Smith (Ed.), Research in experimental economics (Vol.3). Greenwich, CT: JAI Press.

Shubik, M. (1971). The dollar auction game: A paradox in noncooperativebehavior and escalation. Journal of ConXict Resolution, 15, 109–111.

Singley, M. K., & Anderson, J. R. (1989). The transfer of cognitive skill.Cambridge, MA: Harvard University Press.

Thompson, L. (1990a). An examination of naive and experienced negotia-tors. Journal of Personality and Social Psychology, 59, 82–90.

Thompson, L. (1990b). Negotiation behavior and outcomes: Empiricalevidence and theoretical issues. Psychological Bulletin, 108, 515–532.

Thompson, L. (1990c). The inXuence of experience on negotiation perfor-mance. Journal of Experimental Social Psychology, 26, 258–544.

Thompson, L. (1991). Information exchange in negotiation. Journal ofExperimental Social Psychology, 27, 161–179.

Thompson, L. (2001). The mind and heart of the negotiator (2nd ed.). UpperSaddle River, NJ: Prentice-Hall.

Thompson, L. (1992). A method for examining learning in negotiation.Group Decision and Negotiation, 1, 71–84.

Thompson, L., & DeHarpport, T. (1994). Social judgment, feedback, andinterpersonal learning in negotiation. Organizational Behavior andHuman Decision Processes, 58, 327–345.

Thompson, L., Gentner, D., & Loewenstein, J. (2000). Avoiding missedopportunities in managerial life: Analogical training more powerfulthan individual case training. Organizational Behavior and HumanDecision Processes, 82, 60–75.

Thompson, L., & Hastie, R. (1990). Social perception in negotiation. Orga-nizational Behavior and Human Decision Processes, 47, 98–123.

Thompson, L., Peterson, E., & Brodt, S. E. (1996). Team negotiation: Anexamination of integrative and distributive bargaining. Journal of Per-sonality and Social Psychology, 70, 66–78.

Walton, R. E., & McKersie, R. B. (1965). A behavioral theory of labor nego-tiations. New York: McGraw-Hill.

Weingart, L. R., Thompson, L., Bazerman, M. H., & Carroll, J. S. (1990).Tactical behavior and negotiation outcomes. The International Journalof ConXict Management, 1, 7–31.

Yukl, G. (1974). EVects of the opponent’s initial oVer, concession magni-tude, and concession frequency on bargaining behavior. Journal of Per-sonality and Social Psychology, 30, 323–335.

Zeng, D., & Sycara, K. (1998). Bayesian learning in negotiation. Human-Computer Studies, 48, 125–141.

Experience in integrative negotiations: What needs to be learned

Documents

Transcript of Experience in integrative negotiations: What needs to be learned