1 Introduction
“I have had a philosophy for some time in regard to SALT, and it goes like this: the Russians will not accept a SALT treaty that is not in their best interests, and it seems to me that if it is in their best interests, it cannot be in our best interest.”
U.S. Congressman Floyd Spence.Footnote 1Normative models of bargaining and negotiation suggest that if there is potential for mutual benefit, conflicting parties should be able to achieve it (Raiffa, Reference Raiffa1982). Descriptive accounts and empirical investigations of negotiation behavior (e.g., Bazerman & Neale, Reference Bazerman, Neale, Bazerman and Lewicki1983; Thompson, Reference Thompson1990; Thompson & Hrebec, Reference Thompson and Hrebec1996; Walton & McKersie, Reference Walton and McKersie1965), however, suggest that a number of psychological barriers to conflict resolution are likely to make efficient deal making difficult (Bazerman & Neale, 1990; Ross and Stillinger, Reference Ross and Stillinger1991; Thompson & Hastie, Reference Thompson and Hastie1990). For example, research on cognitive biases associated with egocentric perceptions suggests that negotiators and evaluators of negotiated agreements are likely to exhibit a “fixed-pie bias” (Bazerman, Reference Bazerman1986; Bazerman & Neale, Reference Bazerman, Neale, Bazerman and Lewicki1983; Schelling, Reference Schelling1960). The fixed-pie bias refers to the belief that any gain for one party will be associated with an equivalent loss to the other party. This belief is a “bias” when it persists even in contexts where there is a possibility of compatible interests or mutual benefit. A large body of research finds that negotiators are susceptible to the fixed-pie bias prior to, during, and even after negotiations (de Dreu, Koole, & Steinel, Reference de Dreu, Koole and Steinel2000; Pinkley, Griffith, & Northcraft, Reference Pinkley, Griffith and Northcraft1995; Thompson & Hastie, Reference Thompson and Hastie1990; Thompson & Hrebec, Reference Thompson and Hrebec1996). In the current paper we investigate and extend research on fixed pie bias in the context of protracted intergroup conflict. We show that the fixed pie bias can be overcome when disputants understand that successful implementation of negotiated agreements requires an avoidance of outcomes that are favorable to one side.
1.1 Fixed-pie perceptions of negotiated agreements
Fixed-pie perceptions can be detrimental to dispute resolution when parties undervalue or reject proposals simply because they are perceived to be supported by their adversaries (Ross, Reference Ross, Arrow, Mnookin, Ross, Tversky and Wilson1995; Ross & Stillinger, Reference Ross and Stillinger1991). Believing that “what is good for them is bad for us” is not necessarily irrational; when there is uncertainty regarding the value of the proposal, it is reasonable to use information regarding the degree to which others support the proposal when assessing its attractiveness (Maoz, Ward, Katz, & Ross, Reference Maoz, Ward, Katz and Ross2002; Ross, Reference Ross, Arrow, Mnookin, Ross, Tversky and Wilson1995). If the other party is an opponent, competitor, or an enemy, their support for the proposal may indicate that it is of significant benefit only to them. However, as fixed-pie bias increases, it may lead parties to irrationally undervalue proposals simply because they are valued by adversaries. Indeed, research suggests that negotiators devalue others’ offers even when there is little or no ambiguity in the proposal (Maoz et al., Reference Maoz, Ward, Katz and Ross2002) and when the other party is not seen as an opponent (Lepper, Ross, Tsai, & Ward, Reference Lepper, Ross, Tsai and Ward1994). Furthermore, negotiators tend to judge identical proposals more negatively after they have been offered by the other party than before (Ross, Reference Ross, Arrow, Mnookin, Ross, Tversky and Wilson1995), and more negatively if they are offered by the other party rather than by themselves (Oskamp, Reference Oskamp1965).
In the current investigation, we study a potential trigger of fixed-pie bias that is common in the context of many real world negotiations: the degree to which a negotiated agreement is perceived to be favored/opposed by each side. Typically, especially in the domain of ethno-political conflict, formal and informal polling of interested parties begins immediately after a negotiated agreement has been reached. Polling results are made available to the masses (through formal dissemination, leaks, or rumors) and become part of the set of information that individuals might then consider when assessing their own support for the agreement. For example, Nationalists and Unionists in Northern Ireland might have considered not only the text of the Good Friday Agreement when deciding whether to support the peace accord, but also any information they had about the support for the agreement in their own — and in the other — community.
Information regarding who favors the deal will influence support for the agreement in two ways. First, if evaluators believe that any gain for the other party is a loss for themselves, they are less likely to support a deal that has significant support among members of the other party. Second, if most members of one’s own community support the agreement, then the agreement will be evaluated more positively. This is consistent with rational models of decisions making, as well as psychological research on “social proof”, which suggests that, when individuals are uncertain, they look to the behavior and attitudes of similar others for guidance (Cialdini, Reference Cialdini1993).
1.2 Implementation concerns and the preference for balance
While fixed-pie perceptions can affect the degree to which an agreement is perceived as good for one’s own party, agreements are likely to be evaluated not simply on how beneficial they are for one’s own group, but also on how likely they are to be implemented successfully. Because ratifying and enacting an agreement is often costly (both economically and politically), those voting for their group to invest in the endeavor are likely to consider the possibility that implementation will fail. While many factors will affect the likelihood of successful implementation, perhaps the most critical (and most salient) factor is the degree to which each party supports the agreement. If the agreement is entirely one-sided, the party that believes it is getting a bad deal may not be committed to its implementation. The party that is getting the better deal might also withhold support if they believe the other side will not be committed to implementation. In other words, evaluators might realize that a “great deal” is not worth much if it has no chance of being implemented. Thus, the value placed on implementation might offset the effects of fixed-pie perceptions on willingness to support negotiated agreements: disputants may be more willing to support an agreement that is favored equally by both sides — i.e., a balanced deal — rather than one that is perceived as clearly one-sided
Ideally, voters who seek balanced deals will want a deal that is seen favorably by majorities on both sides of the dispute. In some cases, however, deals that are balanced may be those that are equally disappointing to the two sides. Notably, in many negotiations — particularly in difficult negotiations over protracted and seemingly intractable conflicts — both parties will likely perceive significant losses regardless of the outcome that is negotiated. This is because any settlement is seen as a loss relative to the reference point of entrenched aspirations (Ross, Reference Ross, Arrow, Mnookin, Ross, Tversky and Wilson1995), and the parties may therefore frame the outcome in terms of how much was conceded or relinquished (Bazerman & Neale, Reference Bazerman, Neale, Bazerman and Lewicki1983; Kahneman & Tversky, Reference Kahneman and Tversky1979, Reference Kahneman and Tversky1984). The critical factor when considering “balance”, however, is whether each side supports the deal equally, and not the extent to which the deal is seen positively.
Consistent with this, recent research suggests that people may be prepared to accept negative outcomes in inter-group disputes if both sides are seen to be losing equally. For example, Israelis and Palestinians show an enhanced willingness to compromise on issues associated with their “sacred values” if they believe that the other side is having to do the same (Ginges, Atran, Medin & Shikaki, Reference Ginges, Atran, Medin and Shikaki2007). Hayden (Reference Hayden2002) demonstrates that ethnic cooperation tends to flourish under conditions of power balance, where neither ethnic group believes that it can dominate the other. This suggests that in some cases, the appropriate measure of balance may well be “are both parties equally unsatisfied?” rather than “are both parties equally satisfied?” Thus, implementation concerns may lead to increased support for an agreement not only when majorities of both groups favor the agreement, but also when majorities of both groups disfavor it.
1.3 The current study
The current study investigates and extends research on fixed-pie perceptions in a context with substantial ecological validity. The site of the investigation — Israeli attitudes towards the Geneva Accords, an agreement negotiated by unofficial delegations of Israel and the Palestinian Authority — also affords us the opportunity to speak to the dynamics of a current, high-stakes, and real-world conflict. In addition to assessing whether support for a proposal is influenced by information regarding who else supports it, the paper extends fixed-pie research by introducing an important factor that might counteract the effects of fixed-pie bias: implementation concerns.
More specifically, we argue that support for negotiated agreements (e.g., the Geneva Accords) will be influenced by two separate mechanisms. First, people tend to support proposals that they perceive would be good for their own side. As a result, information about the level of support for the Geneva Accords among other Israelis and Palestinians will influence whether Israelis support the accords: Israelis will be more supportive of proposals when they are favored by other Israelis and opposed by Palestinians. Second, parties to a dispute should be more likely to support proposals that they perceive would be likely to be successfully implemented. Thus, they are likely to prefer outcomes that are perceived to be equally attractive to both sides of the dispute rather than one-sided — i.e., balanced agreements. Thus, Israelis should be more supportive of proposals that are equally favored (or disfavored) by Israelis and Palestinians. The current investigation evaluates both sets of propositions.
In the current study, we measured the willingness of Israelis to support the Geneva Accords after they had been told the alleged degree to which Palestinians and other Israelis supported the Accords. Specifically, participants were asked to imagine the results of recent polling that revealed the amount of support for the Geneva Accords among each ethnicity. According to the polls, participants were told, a majority either supported the Accords, opposed the Accords, or there was no information regarding support. This information was provided with regards to Israeli as well as Palestinian polling. Given three possible polling results for each ethnicity, there were 9 (3 × 3) experimental conditions in the study using a between-subjects design.
We tested three sets of hypotheses based on the above discussion. First, the following hypotheses tested the prediction that, if most members of one’s own community support the agreement, then the agreement will be evaluated more positively:
Hypothesis 1a: When a majority of other Israelis support the Geneva Accords, participants will be more likely (a) to view the Geneva Accords as “good for Israel”, and (b) to vote in support of accepting the Geneva Accords.
Hypothesis 1b: The positive effect of Israeli support on willingness to vote for the Geneva Accords will be mediated by perceptions that the accords are “good for Israel”.
Based on the logic of fixed-pie perceptions, we predicted the following:
Hypothesis 2a: When a majority of Palestinians support the Geneva Accords, Israeli participants will be less likely (a) to view the Geneva Accords as “good for Israel”, and (b) to vote in support of accepting the Geneva Accords.
Hypothesis 2b: The negative effect of Palestinian support on willingness to vote for the Geneva Accords will be mediated by perceptions that the accords are “good for Israel”.
To test our reasoning regarding the importance of balanced outcomes and implementation concerns, we created a variable labeled “balance”, which was coded as “1” if the support among Israelis and Palestinians was similar (majorities in both ethnic groups supported the agreement, majorities in both opposed the agreement, or there was no information about support in either group), and coded “0” otherwise. According to the logic of implementation concerns, we predicted the following:
Hypothesis 3a: When support for the Geneva Accords among Israelis and Palestinians is balanced, participants will be more likely to (a) view the Geneva Accords as implementable, and (b) vote in support of the Geneva Accords.
Hypothesis 3b: The positive effect of balance on willingness to vote in support of the Geneva Accords will be mediated by perceptions that the accords are “implementable”.
2 Method
Participants in this experiment were contacted by telephone and asked to answer questions regarding their views on the Geneva Accords. TNS/Teleseker, a survey organization located in Israel, conducted the interviews using “CATI” (computer assisted telephone interviewing) software. Interviews were conducted during the Spring of 2006.
2.1 Participants
450 Israelis participated in this study. 50% of the participants were female. 94.2% of the participants reported that they were Jewish (including Secular, Traditional, Religious, and Ultra-orthodox Jews); the remaining participants reported their religion as Muslim Arab (1.3%), “other” (0.9%), or refused to answer the question (3.6%). The median age of participants was in the range of 40–44 years. All were 18 years or older.
2.2 Procedure
Prior to answering questions regarding the Geneva Accords, participants were asked to imagine that results from recent polling of Israelis indicated that the majority supports the accords, the majority opposes accords, or no information is available. Likewise, results of recent polling of Palestinians indicated that the majority supports, the majority opposes, or there is no information. Each participant was randomly provided one piece of information regarding degree of support for each ethnicity, creating a 3 (Israeli support level) Х 3 (Palestinian support level) between-subjects design.
Participants were then asked to indicate the degree to which they agreed with the following statements: (1) a peace agreement based on the accords would be good for Israel, and (2) a peace agreement based on the accords would be successfully implemented. Participants responded using a 7-point Likert-type scale (1 = “completely disagree”, 7 = “completely agree”). A third question asked whether they would vote to support a peace agreement based on the Geneva Accords (“Yes” or “No”). Participants were also asked a series of demographic questions (as reported above), and questions designed to check whether the experimental manipulations had been effective.
Finally, we sought to investigate whether our predictions regarding balance would be limited to participants who are generally inclined towards favoring peace negotiations. It is clear that those who favor a two state solution (i.e., political liberals, or “doves”) would prefer an implementable deal. It is less obvious that those who oppose such a solution (i.e., political conservatives, or “hawks”) would be positively influenced by implementation concerns. To investigate this, we conducted a supplemental analysis by separating doves from hawks to test whether political leanings of the participants moderated the hypothesized results. To identify participants on the dove-hawk dimension, the survey included the following question (answered using a 7-point scale): “To what extent do you agree with the statement that: Judea and Samaria are integral parts of the Land of Israel and should never be given up.” We regarded respondents as more politically conservative (i.e., hawkish) to the degree that they strongly agreed with this statement.
3 Results
3.1 Manipulation checks
Participants were asked, on a 7-point scale (1= “completely disagree”, to 7 = “completely agree”), the degree to which they agreed with the statements: “Most Israelis (Palestinians) support the Geneva Accords.” Participants agreed more strongly with the statement regarding Israeli support for the accords when they had been informed (in the pre-scenario survey) that a majority of Israelis support the accords (mean response = 3.2) than when they had been told that a majority oppose the accords (mean response = 2.7), [t(284) = 2.631, p < .01].Footnote 2 Similarly, participants agreed more strongly with the statement regarding Palestinian support for the accords when they had been informed (in the pre-scenario survey) that a majority of Palestinians support the accords (mean response = 3.5) than when they had been told that a majority oppose the accords (mean response = 3.0), [t(267) = 1.914, p < .06]. Notably, the Israeli manipulation did not affect perceptions of Palestinian support [t(266) = 0.67, ns] and the Palestinian manipulation did not affect perceptions of Israeli support [t(285) = 0.22, ns].
3.2 Hypothesis testing
Consistent with Hypothesis 1a, participants were more likely to perceive the accords as “good for Israel” [F(2, 445) = 3.35, p < .05], and also more likely to “vote for the accords” [F(2, 419) = 4.84, p < .01], when a majority of Israelis supported the agreement, as shown in Tables 1 and 2. Specifically, 57% said they would vote for the accords when a majority of Israelis supported the accords, compared to 51% when there was no information regarding Israeli support, and 39% when a majority opposed the accords. The mediation model suggested by Hypothesis 1b also received support: the effect of support among Israelis on willingness to “vote for the accords” was partially mediated by perceptions that the accords were “good for Israel”. That is, when both “good for Israel” and support among Israelis were included as predictor variables, “good for Israel” still significantly predicted willingness to “vote for the accords” [F(1, 418) = 572.89, p < .001], but support among Israelis was only marginally significant [F(2, 418) = 2.56, p < .08]. Follow-up analysis using the Sobel Test (Sobel, 1982; Preacher & Leonardelli, Reference Preacher and Leonardelli2001) confirmed that “good for Israel” mediated the effect of Israeli support on “vote for the accords” (z = 3.57, p < .001).
Table 1: Percentage of participants voting in favor of the accords in each condition. The cells in the main diagonal represent “balance” conditions. On average, 55.37% of participants vote in favor of the accords in the 3 balance conditions; 46.02% vote in favor when support is not balanced.
Table 2: Mean scores for “Good for Israel” as a function of experimental condition (standard deviations in parentheses).
Hypotheses 2a & 2b, which expressed the logic of fixed-pie perceptions, were not supported. Information regarding the support (or opposition) for the accords among Palestinians had no effect on perceptions that the accords were “good for Israel” [F(2, 424) = 0.46, p = .633], no effect on the willingness to “vote for the accords” [F(2, 445) = .08, p < .95; F(2, 419) = .32, p = .77], nor any effect on the extent to which participants believed the accords to be implementable [F(2, 424) = 1.25, p = .287.] However, as the results of our manipulation check (above), and our analyses regarding implementation concerns (below) reveal, Israeli participants did not ignore Palestinian attitudes.
As predicted by Hypothesis 3a, balance in the degree of support or opposition among Israelis and Palestinians significantly increased the degree to which the Geneva Accords were perceived as “implementable” [F(1, 448) = 5.20, p < .05]. Table 3 shows the mean “implementable” scores by experimental condition. Balance also marginally increased the willingness of participants to vote for the accords [F(1, 422) = 3.34, p < .07]. On average, 55% of participants were willing to vote for the accords when support was balanced (both support, both oppose, or no information about either); only 46% were willing to vote for the accords when support was not balanced. As predicted by Hypothesis 3b, the effect of balance on voting for the accords was mediated by perceptions that the accords could be successfully implemented: When both balance and implementability were included as predictor variables, balance was no longer a significant predictor of voting (p < .45), but perceived implementability continued to predict voting [F(1, 421) = 118.47, p < .001].
Table 3: Mean scores for perceived “implementability” of the Geneva Accords as a function of experimental condition (standard deviations in parentheses).
A final analysis simultaneously regressed willingness to “vote for the accords” on all of the predictor variables (Israeli support, Palestinian support, good for Israel, balance, and implementability). Analysis revealed significant effects for both mediators: “good for Israel” [F(1, 391) = 346.19, p < .001] and implementability [F(1, 391) = 11.75, p < .001]. No other effects were significant. This analysis provides further support for the mediation models proposed in Hypotheses 1b and 3b.
Table 1 displays the percentage of participants in each of the nine conditions who voted in favor of the accords. It is worth noting that, consistent with the above analyses, a majority of participants (i.e., > 50%) voted in favor of the accords only when a majority of other Israelis supported the accords or when there was a balance in the degree to which Israelis and Palestinians supported the accords (see diagonal). If neither of these conditions was met, a majority of participants voted no. Most notably, perhaps, a majority of participants (52.1%) voted in favor of the accords even when there was balanced opposition to the accords among Israelis and Palestinians.
3.3 Supplemental analysis: Hawks vs. doves
Two types of analyses were conducted using the dove-hawk variable. First, all of the above mediation analyses (testing Hypotheses 1, 2, and 3) were re-run with the inclusion of the dove-hawk variable as a covariate. All of the results remained unchanged. In particular, the dove-hawk variable did not interact with any of the independent variables involved in the testing of Hypotheses 1, 2, or 3. A second analysis regressed willingness to “vote for the accords” on “good for Israel”, “implementability”, and “dove-hawk”. All three variables had a main effect on willingness to vote for the accords, but again there were no interaction effects. This suggests that our measure of “hawkishness” is meaningful because it predicts support for the Geneva Accords [F(1, 412) = 14.69, p < .001], but political leanings did not moderate the hypothesized effects in this study. In other words, the effect of balance on willingness to vote for the accords had a similar effect regardless of the extent to which participants favored the type of two-state solution offered by the Geneva Accords.
4 Discussion
The results confirm that evaluators use information regarding in-group and out-group attitudes towards a proposed agreement when assessing their own willingness to support the agreement. Israeli participants in this study valued the accords more when they were told that other Israelis tended to support the accords. Contrary to the logic of fixed-pie perceptions, support among Palestinians did not lead to negative evaluations of the agreement by Israelis. Indeed, in contrast to prior research on fixed pie perceptions, there was no main effect of Palestinian support or opposition for the accords.
Nonetheless, information regarding Palestinian attitudes was not ignored. The analyses regarding balance provide interesting insights into how this information was incorporated into Israeli evaluations. Israeli votes for the accords actually increased when Palestinians supported the accords, if other Israelis were known to support the accords as well. If other Israelis opposed the accords, Palestinian support for the accords made them less attractive, but Palestinians opposition to the accords (which created balance between the groups) made them much more attractive. In other words, the effect of Palestinian support on Israeli attitudes was contingent upon the degree of Israeli support.
5 Conclusions
These results suggest a degree of complex processing among the participants in this study that would not have been predicted by the extant literature on fixed-pie perceptions. Prior research suggests that partisans will devalue an agreement if there is reason to believe that the other side values it (Ross, Reference Ross, Arrow, Mnookin, Ross, Tversky and Wilson1995). However, this research ignores implementation concerns, which are likely to mitigate the effects of fixed-pie perceptions. To the extent that implementability relies on balancing the level of satisfaction each party feels, negotiators (or, more generally, evaluators of negotiated agreements) are unlikely to use absolute levels of the other party’s satisfaction as particularly informative. In other words, it is not whether the other side is happy (or sad), but is the other side as happy (or sad) as us?
This has a number of practical implications for those seeking to gather support for negotiated agreements (e.g., divorce lawyers, peace negotiators, mediators, etc.). For example, if there is little ambiguity regarding the benefits to one’s own group, or if it is easy to clarify the benefit of the agreement to one’s group, making salient the other party’s satisfaction with the deal might increase the agreement’s perceived implementability and make it more attractive. This advice runs contrary to what is suggested by prior research on fixed-pie perceptions. The results also suggest that, if the parties are negotiating in the domain of losses, or if the constituency is likely to be highly concession-averse, it may be better to embrace and publicize the dissatisfaction felt by both constituencies rather than attempting to spin the agreement as a good thing.
Extensions of the current research might consider when implementation concerns are most likely to displace or overshadow fixed-pie perceptions. For example, implementation concerns are likely to be more important determinants of support for negotiated agreements when the need for an implementation stage becomes salient to evaluators, when evaluators are sophisticated (i.e., educated and aware of the complexities inherent in bargaining) and when the conflict is protracted and entails a history of implementation failures (as is certainly the case in the Israeli-Palestinian conflict). In addition, we note that we did not give participants in this study a detailed account of the content of the Geneva Accords. It remains to be seen whether the effects we have documented would be moderated by the degree of knowledge that evaluators have regarding the agreement. On the one hand, to the extent that a clear understanding of the agreement reduces ambiguity, the effect of polling results should decrease. On the other hand, prior research on reactive devaluation (e.g., Maoz et al, Reference Maoz, Ward, Katz and Ross2002) reveals that the effects of who proposed the peace deal can persist even when participants are given detailed descriptions of the proposal. Finally, if there is a small “zone of possible agreement”, even partisan observers may become less concerned with “winning” and more concerned with reaching any agreement that has a possibility of being implemented successfully.
The current study formulates, in an important context, a framework with which to study the dynamics of fixed-pie perceptions in complex negotiations. In addition, the results indicate that implementation concerns can be critical in understanding partisan attitudes and support for negotiated agreements. The results speak to the possibility that negotiator (or constituent) cognitions may not be as irrational as suggested by extant literature on the topic (cf., Bazerman, Reference Bazerman2005), and also to the possibility that the psychological and strategic shortcomings that do arise in a variety of conflict resolution processes and domains might be overcome.
Open access
