Introduction
In violent conflicts, popular willingness to compromise is a critical yet elusive step for advancing peace. A large literature has examined the factors that shape public opinion for or against peaceful compromises in such contexts. A key argument suggests that improving future expectations about the likelihood of an agreement, via positive top-down or bottom-up signals, can shift public opinion in favour of diplomacy and concessions (for example, Halperin et al. Reference Halperin, Russell, Trzesniewski, Gross and Dweck2011; Hasler et al. Reference Hasler, Leshem, Hasson, Landau, Krayem, Blatansky, Baratz, Friedman, Psaltis, Cakal, Cohen-Chen and Halperin2023; Leshem and Halperin Reference Leshem and Halperin2020; Pruitt Reference Pruitt1997; Shamir and Shikaki Reference Shamir and Shikaki2002; Zartman Reference Zartman, Druckman and Stern2000). Policy preferences, in this view, can change when new information updates prior beliefs about probable developments.
This proposition, however, is complicated by a growing body of work, extending beyond the research of conflicts, which shows that prior ideological beliefs often bias people’s future expectations rather than vice versa (Jerit and Barabas Reference Jerit and Barabas2012; Taber and Lodge Reference Taber and Lodge2006). This counterargument suggests that new information is often screened and interpreted selectively to fit existing views, particularly on salient issues involving strong group identities (Cohen-Chen, Halperin, Porat et al. Reference Cohen-Chen, Halperin, Porat and Bar-Tal2014; Gaines et al. Reference Gaines, Kuklinski, Quirk, Peyton and Verkuilen2007; Nyhan and Reifler Reference Nyhan and Reifler2010; Zaller Reference Zaller1992). Hence, future expectations could reflect prior ideological preferences rather than shape them.
This fundamental disagreement carries important implications for our understanding of public opinion, violent conflicts, and peacemaking. Theoretically, the two arguments differ on the true causal direction between future expectations and preferences for compromise. Policy-wise, they suggest opposing courses of action to promote peace. The former view encourages the cultivation of hopeful signals about the conflict’s potential resolution by leaders and regular citizens. The latter, by contrast, is pessimistic about their value and implies slower and more laborious efforts to form peaceful preferences through early education, collective discourse, and alternative group identities (Coleman et al. Reference Coleman, Vallacher, Nowak and Bui-Wrzosinska2007).
Nevertheless, the competing predictions and potential endogeneity have not been tested in conflictual contexts. Addressing this gap, the current study leverages uniquely rich survey data from Israel to empirically disentangle this relationship in a real-world protracted conflict. Our findings corroborate the first argument. Using vector autoregression (VAR) analysis of monthly time series spanning two decades, we find that aggregate changes in Jewish Israelis’ expectations about future resolution precede shifts in support for compromise but not vice versa. Then, exploiting two exogenous shocks to peace expectations, we validate this influence at the individual level and its uniform impact across partisan lines.
Hence, in violent conflicts, changes in prospective expectations can and do lead to corresponding shifts in public willingness to compromise, with little contrary evidence. This finding is significant for domestic and international peacemaking initiatives, suggesting that hawkish views partly embed practical pessimism. Accordingly, efforts to improve future expectations about an agreement could help overcome one of the most persistent barriers to conflict resolution. This finding further shows that the relationship between preferences and expectations acts uniquely in violent conflicts compared to other policy domains.
The study proceeds with a brief outline of the two opposing arguments and their empirical implications. We then discuss the Israeli case study and present our data and findings in two steps. The first involves VAR analysis of monthly time series of aggregate Jewish-Israeli public opinion from 2001 to 2020. The second examines individual-level preferences before and after two historical shocks to peace expectations. We conclude with several takeaways.
Two Opposing Logics
Previous research on conflicts has identified people’s future expectations as a key factor for resolution. Several studies argue that improved expectations about the practical prospects of peace can increase policy preferences for compromise (for example, Halperin et al. Reference Halperin, Russell, Trzesniewski, Gross and Dweck2011; Hasler et al. Reference Hasler, Leshem, Hasson, Landau, Krayem, Blatansky, Baratz, Friedman, Psaltis, Cakal, Cohen-Chen and Halperin2023; Leshem and Halperin Reference Leshem and Halperin2020; Pruitt Reference Pruitt1997; Shamir and Shikaki Reference Shamir and Shikaki2002; Zartman Reference Zartman, Druckman and Stern2000). Belligerent attitudes are often linked to a sense of personal and national threat, stress, and collective humiliation (Canetti et al. Reference Canetti2019, Reference Canetti, Elad-Strenger, Lavi, Guy and Bar-Tal2017; Zipris et al. Reference Zipris, Pliskin, Canetti and Halperin2019). Greater optimism about the conflict’s future, however, moderates such reactions and reduces threat perceptions, out-group dehumanization, and desire for retaliation (Bar-Tal Reference Bar-Tal2007; Leshem and Halperin Reference Leshem and Halperin2023; Moeschberger et al. Reference Moeschberger, Dixon, Niens and Cairns2005; Rosler, Cohen-Chen, and Halperin Reference Rosler, Cohen-Chen and Halperin2017; Shikaki Reference Shikaki2006). Some conclude, therefore, that increasing optimism is a necessary precondition for peacemaking (Pruitt Reference Pruitt1997).
Building on this framework, multiple studies examine the external factors shaping prospective expectations about peace as a first step for resolution. The findings show that future expectations react to the conflict’s violence and costs (Gayer et al. Reference Gayer, Landman, Halperin and Bar-Tal2009; Gelpi, Feaver and Reifler Reference Gelpi, Feaver and Reifler2006; Godefroidt Reference Godefroidt2023; Zartman Reference Zartman, Druckman and Stern2000) signals about the adversary’s openness to compromise (Berenji Reference Berenji2020; Cohen-Chen, Halperin Crisp et al. Reference Cohen-Chen, Halperin, Crisp and Gross2014; Halperin et al. Reference Halperin, Russell, Trzesniewski, Gross and Dweck2011; Leshem and Halperin Reference Leshem and Halperin2020; Rosler et al. Reference Rosler2023; Yakter and Harsgor Reference Yakter and Harsgor2023), media coverage (Sheafer and Dvir-Gvirsman Reference Sheafer and Dvir-Gvirsman2010), and the international community’s positions (Shelef and Zeira Reference Shelef and Zeira2017).
The argument’s underlying premise, however, stands at the heart of a broader disciplinary disagreement about the causal relationship between prospective expectations and policy preferences. On the one hand, some arguments, as the one noted above, imply rational belief updating and Bayesian learning processes (Coppock Reference Coppock2022; Gerber and Green Reference Gerber and Green1999). In this view, people adjust future expectations following new external information. They then update their prior policy preferences based on the modified evaluations. Economic voting has been a paradigmatic example: voters often observe real-world economic markers, modify their economic expectations accordingly, and subsequently update related policy and electoral preferences (Lewis-Beck Reference Lewis-Beck1988; Lockerbie Reference Lockerbie1991; MacKuen, Erikson and Stimson Reference MacKuen, Erikson and Stimson1992; Nadeau, Lewis-Beck and Bélanger Reference Nadeau, Lewis-Beck and Bélanger2012; Rueda and Stegmueller Reference Rueda and Stegmueller2019).
On the other hand, some works, mainly outside the conflict literature, challenge this premise and suggest that prior preferences shape expectations instead of vice versa. This counterargument argues that people are cognitively motivated to justify and rationalize their long-standing worldviews in the face of new information, particularly on salient issues with strong predispositions (Jerit and Barabas Reference Jerit and Barabas2012; Taber and Lodge Reference Taber and Lodge2006). Such reaffirmation occurs through biased information screening and retrieval, whereby people selectively seek, interpret, and accept messages corroborating prior attitudes and reject opposing evidence (Ditto and Lopez Reference Ditto and Lopez1992; Gaines et al. Reference Gaines, Kuklinski, Quirk, Peyton and Verkuilen2007; Kunda Reference Kunda1990; Nyhan and Reifler Reference Nyhan and Reifler2010; Zaller Reference Zaller1992). Hence, existing preferences often form biased prospective expectations rather than the contrary (Bartels Reference Bartels2002; Duch, Palmer and Anderson Reference Duch, Palmer and Anderson2000; Wlezien, Franklin and Twiggs Reference Wlezien, Franklin and Twiggs1997).
Although typically debated regarding economic and partisan attitudes, this opposing framework is particularly relevant for violent conflicts. Conflicts are politically salient, involve strong collective identities, and tend to solidify rigid and long-standing attitudes validating the in-group and denigrating the adversary (Bar-Tal Reference Bar-Tal2007; Homola, Pereira and Tavits Reference Homola, Pereira and Tavits2020; Lupu and Peisakhin Reference Lupu and Peisakhin2017; Rozenas, Schutte and Zhukov Reference Rozenas, Schutte and Zhukov2017). Studies on long-standing conflicts show that such narratives increase selective information-seeking, dismissal of hopeful messages, and out-group intolerance (Bar-Tal, Reference Bar-Tal, Raviv, Raviv and Dgani-Hirsh2009; Cohen-Chen, Halperin, Porat et al. Reference Cohen-Chen, Halperin, Porat and Bar-Tal2014; Manekin and Mitts Reference Manekin and Mitts2022; Nyhan and Zeitzoff Reference Nyhan and Zeitzoff2018; Peffley, Hutchison and Shamir Reference Peffley, Hutchison and Shamir2015). Consequently, ongoing conflicts often anchor public opinion in pessimistic equilibria regardless of positive signals (Coleman et al. Reference Coleman, Vallacher, Nowak and Bui-Wrzosinska2007).
The disagreement is not just theoretical but predicts different empirical patterns. The first view expects that positive (negative) changes in prospective expectations about resolution will subsequently increase (decrease) support for compromise. The second argument rejects this proposition and implies three possible alternatives. Firstly, there could be a null relationship, in which changes (positive or negative) in future expectations do not move preferences for compromise at all. Secondly, there could be an opposite relationship, wherein positive (negative) changes in prospective expectations react to preceding increases (decreases) in support for compromise. Finally, there could be a heterogeneous relationship, where shifts in expectations only affect the preferences of subgroups ideologically predisposed to favour them. Table 1 summarizes the different predictions.
Expected Empirical Patterns by Theoretical Argument
Nevertheless, these competing hypotheses have not been tested in violent conflicts outside the laboratory. In what follows, we leverage uniquely rich survey data from Israel to empirically disentangle this relationship in real-world settings using both aggregate and individual-level analyses. We first discuss our case selection.
The Israeli Context
Our empirical analysis focuses on Jewish public opinion in Israel, a paradigmatic case study in the research of public opinion and conflict processes (Godefroidt Reference Godefroidt2023; Phillips and Greene Reference Phillips and Greene2022). This context affords several key advantages for our purposes. First, the Israeli-Palestinian conflict is protracted and violent, but also exhibits dynamic trends of escalation and diplomacy. Thus, it displays temporal changes in expectations and preferences about peace while holding the conflict’s nature constant. Second, Israel provides rich public opinion data spanning lengthy periods and critical historical junctures, allowing us to measure popular trends in close and regular intervals. Third, Jewish Israelis are deeply divided on the conflict’s desired solution (Shamir and Arian Reference Shamir and Arian1999; Yakter and Tessler Reference Yakter and Tessler2023), a useful setting to examine potential heterogeneity by ideological predispositions.
The choice of Israel also has limitations, particularly when generalizing to milder, shorter, or more symmetrical conflicts. Specifically, this salient and protracted context may overstate ideological rigidity over change compared to more moderate cases. The conflict’s asymmetry, wherein Israelis experience lower regular violence than Palestinians, may further weaken their urgency for resolution. Hence, null or static findings should be generalized carefully. Nevertheless, this potential rigidity lends greater credibility to patterns of change, as we do find below, informing and possibly understating similar trends in milder or more dynamic cases.
Aggregate-Level Relationship
Data and Method
First, we examine the direction of influence between expectations and preferences at the aggregate level. Longitudinal analyses of aggregate attitudes are particularly useful for revealing consistent patterns of temporal ordering between public views over long periods. This helps identify structural, population-wide dynamics that may be obscured by individual-level fluctuations, measurement error, or short-term noise. Moreover, macro-level shifts in popular opinion are often substantial enough to influence public debates and policy making. Accordingly, their analysis is well-established in research on the temporal dynamics of public opinion vis-à-vis other attitudes and elite behaviour (for recent examples, see Barberá et al. Reference Barberá, Casas, Nagler, Egan, Bonneau, Jost and Tucker2019; Devine and Murphy Reference Devine and Murphy2020; Enns Reference Enns2014; Key and Donovan Reference Key and Donovan2017; Merkley and Stecula Reference Merkley and Stecula2021; Soroka, Stecula and Wlezien Reference Soroka, Stecula and Wlezien2015; Widmann Reference Widmann2022; Wlezien Reference Wlezien2024). Nevertheless, while such data can detect cumulative processes across large segments of the population, as we often expect in active conflicts, they do not establish individual-level correlations and carry the risk of ecological fallacy. We address this limitation in the second step of our study.
Our analysis uses time-series data from the Peace Index project, which polled representative samples of Jewish Israelis every month for two decades. Although each survey includes different respondents, the identical representative sampling ensures a reliable panel of aggregate Jewish-Israeli public opinion. We examine two recurring questions asked monthly from July 2001 to March 2020. Prospective expectations are measured as follows: ‘Do you believe or not believe that negotiations between Israel and the Palestinian Authority will lead in the coming years to peace between Israel and the Palestinians?’Footnote 1 Support for compromise is gauged thusly: ‘What is your position on conducting peace negotiations between Israel and the Palestinian Authority?’ Both questions use a four-point answer scale from strong disbelief/opposition to strong belief/support. We aggregate individual-level responses into time series by summarizing each question’s monthly net belief/support (the total share of positive answers minus the total share of negative answers per month). Figure 1 plots the two series. Supplementary Appendix (SA) A1 elaborates further on the data. Unit-root tests, summarized in SA A2, verify that both series are stationary.
Net Expectations and Ideological Preferences, 7/2001–5/2020.
To analyze whether changes in expectations predict subsequent shifts in preferences and/or vice versa, we estimate a VAR model (Freeman, Williams and Lin Reference Freeman, Williams and Lin1989; Sims Reference Sims1980). VAR models disentangle endogenous relationships using a system of equations wherein each variable is regressed on its lagged values and the lagged values of the other variables. The results are interpreted using a post-estimation Granger Causality test, which determines whether prior levels of one variable significantly predict the current levels of the other, even when controlling for the latter’s past trajectory (Freeman Reference Freeman1983). Our model includes four monthly lags based on several selection statistics discussed in SA A3.
To isolate the interrelations between expectations and preferences, we control for several exogenous factors that may affect both simultaneously and confound their direct relationship. First, we control for monthly violence levels using the current and one-month lag of Israeli casualties by Palestinians, Palestinian casualties by Israelis, and rockets shot from Gaza. Second, we control for elite-level signals about the conflict’s future using the current and one-month lag of negotiation summit occurrences and hawkish changes in the Palestinian and Israeli leadership (the latter in first difference). SA A3 details these covariates and the full model specification.
Findings
Table 2 presents the VAR’s Granger Causality estimates.Footnote 2 Corroborating the first argument, we find that prospective expectations Granger-cause preferences for negotiations. Conversely, we see no statistically significant indications of a null or opposite relationship. To interpret this influence, Figure 2 plots the orthogonalized impulse response function (OIRF) of a hypothetical one-standard-deviation shock to expectations.Footnote 3 Ceteris paribus, it estimates an immediate 4.36-point shift (0.36 of one standard deviation) in net support for negotiations, followed by additional, gradually diminishing effects over 16 months.
Granger Causality Tests Based on VAR Estimates
Orthogonalized Impulse Response Function (OIRF) for Preferences After a Shock to Expectations. The 95% Confidence Region Marked in Grey.
Individual-Level Reactions to Exogenous Expectation Change
Data and Method
As noted, aggregate-level analyses have several advantages, including their consistent, long-term outlook and external validity, but also two potential downsides. First, they could sometimes mask an ecological fallacy, whereby group-level averages do not correspond with individual-level correlations. Second, aggregate changes could be driven by ideological heterogeneity, that is, only voters ideologically predisposed to favour the new expectations would react to it.Footnote 4 In Israel, where partisanship is defined primarily by attitudes toward the conflict, positive (negative) expectation changes may only strengthen the prior preferences of already dovish left-wingers (hawkish right-wingers). If so, prior preferences would remain the root cause of change.
To address these issues, we examine individual-level data using two historical moments with unanticipated exogenous shocks, one positive and one negative, to Jewish-Israeli peace expectations. Our analysis compares individual support for compromise in the weeks before and after these shocks, regarding them as quasi-experimental informational treatments to expectations, assigned as-if randomly between pre-event and post-event respondents (Muñoz, Falcó-Gimeno and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020). The bandwidth of several weeks before and after the shock helps to increase the findings’ generalizability, capture effects that unfold over several days, and improve statistical power. Nevertheless, SA B7 verifies that the results hold with a narrower temporal bandwidth.
This design requires direct and immediate shocks to future expectations independent of preference change. To support this premise, we selected appropriate events based on several criteria. First, the events must be sudden, salient, and at an identified moment in time. Second, to modify expectations, the shocks must communicate new and clear information about the probability of making peace with the adversary. Third, they should not involve violence, which could influence support for compromise directly, regardless of expectations. Fourth, the events should not be adjacent to other conflict-related events of similar magnitude that could exert competing influences. Fifth, their timing should be independent of preceding short-term changes in Israeli-Jewish support for compromise. Finally, we sought events with available and comparable survey data in the weeks before and after.Footnote 5
Two events fit these conditions well. The first negative shock to peace expectations is the surprising victory of the militant Islamist movement Hamas in the Palestinian legislative elections of January 25, 2006. The pre-election polls and media coverage predicted a narrow win for Fatah, the more moderate incumbent party (An-Najah National University 2006; Erlanger and Myre Reference Erlanger and Myre2006). Moreover, Hamas’s 44.5 per cent vote share, in itself a surprising 3-point edge over Fatah, unexpectedly granted 56 per cent of seats due to new electoral rules. Right after its victory, Hamas publicly refused to recognize Israel, respect previous agreements, or restrain its paramilitary activity, sending a salient informational signal about the lower likelihood of resolution. This message was amplified by the international community’s quick opposition to a Hamas-led government that would endorse these positions. At the same time, these events did not include violence against Israeli citizens, which could have affected preferences through other channels.
The data before and after Hamas’s January 25 victory are taken from Peace Index polls. The control group (n = 1,016) was surveyed on November 28-29 and on December 26-27, 2005, while the post-event treatment group (n = 1,005) was polled on February 27-March 1 and on April 3-4, 2006. Preferences for compromise are measured as the level of support for peace negotiations with the Palestinian Authority on a four-point scale.
The second event represents a positive shock to expectations. For a fittingly exogenous treatment, we travel back to the 1970s, when the conflict involved neighbouring Arab countries more prominently. On November 9, 1977, four years after a traumatizing regional war, Egypt’s president, Anwar Sadat, unexpectedly declared his readiness to visit Israel. Sadat’s trip on November 19-21, the first ever by an Arab leader, included multiple symbolic gestures and a public speech in the Israeli parliament calling for peace in return for Israeli territorial withdrawal. The visit was coordinated secretly in small elite circles and caught the Israeli public, parts of its leadership, and other Arab countries by complete surprise (Bar-Siman-Tov Reference Bar-Siman-Tov1994; Berenji Reference Berenji2020).
The data surrounding Sadat’s November 19-21 visit is from the biweekly Continuing Survey series by the Israel Institute for Applied Social Research.Footnote 6 The pre-visit surveys (n = 1,075) are from October 6-9 and October 20-22. The post-visit surveys (n = 1,048) were fielded on November 21-23 and November 30-December 4. Preferences for compromise are gauged as follows: ‘Regarding the territories held by Israel since the Six-Day War, what is the greatest concession that you think should be made for peace with the Arab countries?’ The answers range on a five-point scale from ‘do not relinquish any territory, without exceptions’ to ‘relinquish all territories for a peace agreement’.
In both cases, we estimate the expectation shock’s effect on support for compromise by regressing the latter on a dummy variable indicating post-event respondents. For comparability across both datasets, support for compromise is rescaled to 0-1. We estimate both a bivariate model and a model with controls for gender, age group, education, and religiosity. Then, to examine partisan heterogeneity, we re-estimated the latter using subsamples comprising only right-wing and left-wing respondents, in turn. To avoid post-treatment bias in ideological identification, we measure partisanship based on respondents’ self-reported party vote in the prior election, which we recode into ideological blocs (Left, Centre, and Right). SA B1 elaborates further on covariate operationalizations.
This quasi-experimental method relies on several inferential assumptions (Muñoz, Falcó-Gimeno and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020). First, we assume that the two shocks represent immediate changes in expectations. This assumption is verified by survey data showing that peace expectations changed after both events in the appropriate directions (see SA B3). It is further supported by our aggregate finding that expectations consistently precede preference changes. Second, we assume excludability; that is, no other factors near the events caused a preference change independently of expectations. We support this assumption using our case selection criteria, which rule out parallel events or endogenous event timing, and several placebo tests (see SA B8).Footnote 7 Third, we assume ignorability, that is, respondents were equally likely to be surveyed before and after the events regardless of their views about the conflict. This premise is corroborated by the surveys’ characteristics (see SA B1), balance tests (see SA B3), and a narrower bandwidth test (see SA B7). Finally, we assume compliance, that is, all post-event respondents experienced the events. This premise is supported by the salient nature of both shocks and the high media and national attention they received. SA B2 discusses these assumptions and corroborations in greater detail.
Findings
Figure 3 plots the shocks’ estimated influences on the full sample, with and without controls, and on the partisan subsamples of left-wing and right-wing voters.Footnote 8 We find that both expectation shocks change policy preferences in their respective negative/positive direction, regardless of partisanship. Resembling our aggregate analysis, the effects equal 0.17–0.35 (2006) and 0.24–0.35 (1977) of the outcome’s one standard deviation. Alternative specifications, including entropy balancing and nearest-neighbour matching (see SA B6), and placebo tests (see SA B8), validate the findings’ robustness.
The Effect of Expectation Shock on Support for Compromise, by Model. The Vertical Lines Mark 95 per cent Confidence Intervals.
Importantly, the relative effect is similar across partisan subgroups. While we see a larger point estimate for right-wingers (β R =−0.131) compared to left-wingers (β L =−0.063) in 2006, a Chi-square test for between-model coefficient differences finds this gap insignificant at the 95 per cent level (p=0.061). Moreover, there is no similar difference in 1977. Nonetheless, the noisy gap implies that right-wing voters may react more strongly than left-wingers to negative signals reaffirming their hawkish priors. This potential pattern calls for further research on the complex relationships between information updates and partisanship in conflictual settings, including possible substantive differences between left-wing and right-wing predispositions.
Conclusion
The causal relationship between prospective expectations and policy preferences stands at the heart of a broad disciplinary debate. However, it remains understudied in violent conflicts despite its importance for peacemaking. On the one hand, public support for compromise with adversaries relies on assessments of future scenarios and risks. On the other hand, these preferences often reflect rigid group identities and defensive narratives that can bias future expectations.
This study aimed to empirically disentangle this causal relationship in such contexts. Leveraging uniquely rich data on Jewish-Israeli attitudes from multiple decades and historical junctures, we find robust evidence that changes in expectations about the likelihood of resolution predict subsequent shifts in support for compromise. We find little evidence of null, opposite, or heterogeneous effects. The Israeli-Palestinian conflict’s protracted, deep-rooted nature lends additional credence to these dynamic findings and adds rigour and missing external validity to prior qualitative and micro-behavioural indications in this vein.
Our analysis makes two primary contributions to current debates on public attitudes, preference change, and violent conflicts. First, to the best of our knowledge, this is the first direct test of potential endogeneity between expectations and preferences in violent conflicts. In doing so, we link together research on conflict-related attitudes and broader debates, with mixed results, about belief updating and motivated reasoning. The former tend to concentrate on qualitative case analysis, observational correlations, or one-sided laboratory manipulations of either expectations or preferences without testing their complex interrelations (for example, Cohen-Chen, Halperin, Crisp et al. Reference Cohen-Chen, Halperin, Crisp and Gross2014; Halperin et al. Reference Halperin, Russell, Trzesniewski, Gross and Dweck2011; Hasler et al. Reference Hasler, Leshem, Hasson, Landau, Krayem, Blatansky, Baratz, Friedman, Psaltis, Cakal, Cohen-Chen and Halperin2023; Leshem and Halperin Reference Leshem and Halperin2020; Pruitt Reference Pruitt1997; Rosler et al. Reference Rosler2023; Shamir and Shikaki Reference Shamir and Shikaki2002; Zartman Reference Zartman, Druckman and Stern2000). The latter, meanwhile, focuses on other issue domains, such as economic and domestic policy, where risk assessment and group identities may operate differently (for example, Jerit and Barabas Reference Jerit and Barabas2012; Nadeau, Lewis-Beck and Bélanger Reference Nadeau, Lewis-Beck and Bélanger2012; Taber and Lodge Reference Taber and Lodge2006; Wlezien, Franklin and Twiggs Reference Wlezien, Franklin and Twiggs1997). Our findings, therefore, underscore the need for broader comparative research on this tension and its underlying mechanisms across issues and geopolitical contexts.
Second, our results carry significant implications for conflict resolution. Specifically, they suggest that hawkish attitudes against compromise are not just ideological but can also price in pessimism about its practical prospects. Accordingly, initiatives aimed at improving future expectations can form a key pathway for changing hearts and minds, even in protracted conflicts. Such initiatives could be from the top down, through ‘bold moves’ and conciliatory messages by elites (for example, Berenji Reference Berenji2020; Yakter and Harsgor Reference Yakter and Harsgor2023), or from the bottom up, through signals about the openness of out-group members to compromise (for example, Leshem and Halperin Reference Leshem and Halperin2020; Rosler et al. Reference Rosler2023). Conversely, waiting for public preferences to change before taking such difficult steps may enforce a negative equilibrium and prolong the status quo.
At the same time, our findings do not rule out broader theories about ideological bias in conflicts. While we show that shifts in prospective expectations can move support for compromise across ideological camps, advancing such expectation change in the first place remains a challenge. In-group and partisan predispositions may impede expectation change and require sufficiently large shocks to trigger meaningful shifts. Moreover, partisanship often establishes separate information environments, shaping different expectations and subsequent belief updating. Hence, future research should study the real-world conditions under which exogenous information can pass partisan thresholds, permeate prospective expectations, and build up positive momentum for peacemaking.
Supplementary Material
Supplementary material for this article can be found at https://doi.org/10.1017/S0007123425100550.
Data Availability Statement
Replication data for this paper can be found in Harvard Dataverse at:https://doi.org/10.7910/DVN/R2LLBV.
Acknowledgements
We thank Babak RezaeeDaryakenari, Devorah Manekin, Matthew Simonson, Daniel Smith, James Walsh, Omer Yair, and the BJPS editors and anonymous reviewers for their helpful comments and suggestions. We are also grateful to Mohammad Saad for excellent research assistance. Earlier versions of this paper were presented at the annual meetings of the European Political Science Association (2023), the Jan Tinbergen European Peace Science Conference (2023), the Israeli Association for International Studies (2023), and the Junior Scholars in Quantitative Conflict and International Security Workshop (2023).
Financial Support
This research was supported by the Israel Science Foundation (grant No. 2976/21).
Competing Interests
None.
Open access
