Tolerance and Threat Revisited: The Dynamics of Political Tolerance Under Persistent Terrorism

Michal Shamir, Marc L. Hutchison, Mark Peffley, and Yu Ouyang

One of the most consistent findings of political tolerance studies is that threat is a powerful force in pushing people away from tolerance, defined classically by Sullivan et al. (1982, 2) as a “willingness to permit the expression of ideas or interests one opposes.” In the first systematic study of tolerance, Stouffer (1955, 13) recognized that Communism, Conformity, and Civil Liberties was about the “reactions [of Americans] to the Communist threat.” But it was not until a quarter century later that John Sullivan and his colleagues, in Political Tolerance and American Democracy (1982), established a post-Stouffer paradigm for explaining political tolerance and what has become a routine finding of tolerance research—that citizens’ perceptions of threat from highly disliked groups strongly undermine tolerance and thereby put liberal democracies at risk.

Sullivan et al.’s work on political tolerance inspired a generation of research on a core democratic value. Certainly, John Sullivan’s legacy lives on in our own work on political tolerance. Shamir and Peffley both worked with Sullivan as graduate students at University of Minnesota. In fact, Shamir co-authored one of the classic works on tolerance in Israel with Sullivan in 1983. Hutchison worked with Peffley at University of Kentucky, and after meeting Shamir, the three of them made early plans to work eventually on the Israeli terrorism and political tolerance project, of which this chapter is a part. Yu Ouyang, who also received his PhD at Kentucky, joined the project later.

Our chapter re-examines the substantial work on tolerance by Sullivan and others to first clarify the role of both subjective and objective threat in motivating political intolerance and then to show empirically how the two types of threat influence tolerance. Our study is unique in its ability to disentangle the connections between objective threat, subjective threat, and political tolerance toward disliked groups. We do so by examining the role of terror attacks in shaping perceptions of threat and undermining political tolerance in Israel during an extended period when all three variables were in flux.

Despite the strong and consistent role of threat in motivating political intolerance, conundrums abound on the how, why, and when of the relationship between threat and tolerance. In his essay on “Enigmas of Intolerance,” Gibson (2006, 21) discussed “Threat as an Unexplained Variable” first, noting that “(t]he single most important predictor of intolerance—perceptions of the threat posed by one’s domestic political enemies—is an unexplained variable in nearly all studies of intolerance. Thus, the key determinant of intolerance is itself poorly understood.” In other words, studies using Sullivan et al.’s least-liked tolerance measure regularly find that threat perceptions are essentially exogenous—i.e., unexplained by other individual-level variables.

A related conundrum concerning threat is posed by Sullivan and Hendricks (2009, 379) in their review of political tolerance research before and after 9/11: “... the role that political threat plays in shaping levels of political tolerance appears to be largely a subjective one, as perceived levels of threat often do not correspond well with more objective measures, and it is increases in this subjectively felt level of threat that seem primarily to diminish levels of tolerance...”. This view of threat as subjective and enigmatic prevails in much of the political tolerance literature, despite the authors’ acknowledgment (in the same paragraph) that objective levels of national threat to “economic and political security (such as during open warfare)” clearly influence tolerance also.

Our chapter takes aim at both conundrums—i.e., the inability to predict perceived threat and the ambiguity of the relationship between two types of threat—objective and subjective, and political tolerance. We do so by studying the dynamics of terrorism, perceived threat, and political tolerance toward least-liked groups in Israel over a 30-year period.

Context and Threat in Political Tolerance Studies

The seminal study of political tolerance by Sullivan et al. (1982) emerged from the political context of the late 1970s in the United States, which was very different from the McCarthy Red Scare of the 1950s when Stouffer studied tolerance. Sullivan et al. argued that the political context of their study was one of pluralistic intolerance, where intolerance is high and target groups selection is diverse, whereas the McCarthy era was best described as focused intolerance, where more consensus exists on the selection of highly disliked groups, like American Communists and fellow travelers. Sullivan and others (e.g., Marcus, Gibson, and Sniderman) expanded this pioneering research in two directions. One direction extended the study of tolerance cross-nationally; the other manipulated contextual conditions using survey experiments.

Our work falls squarely into the first camp. In fact, Michal Shamir and John Sullivan (1983) undertook the first large-scale comparative study using the least-liked approach in Israel in 1980, two years after Sullivan et al.’s first survey in the United States. In the three democracies that constituted Political Tolerance in Context (1985), which added New Zealand to the mix, Israel was selected in large part because of its exceptional threat context, where “a high level of political threat, both internal and external... to the regime, to the values of most Israeli citizens, and to their very existence is far more objective and immediate than is the case for Americans” (Sullivan et al. 1984, 321; see also Sullivan et al. 1985). Over the next three decades, Shamir and other Israeli scholars conducted a multitude of tolerance surveys using the least-liked measure, from which we draw in this study.¹

While Political Tolerance in Context produced important insights, the ability to study systemic differences in the level of objective threat was understandably limited by the use of cross-sectional surveys in three democracies. Several other studies expanded tolerance research to more countries with a wider range of objective threat levels, including the former Soviet Union, Russia, South Africa, Denmark, and Canada (for a partial review, see Gibson 1998a, 1998b; Gibson and Gouws 2003; Sniderman et al. 1996,2014).² In addition, Hutchison (2014) and Hutchison and Gibler (2007) undertook a more systematic cross-national analysis using the 1995—1997 World Values Survey, which included a least-liked measure of political tolerance, and found that political tolerance was much lower in countries that experienced higher levels of state-level conflict. Collectively, these cross-national studies shed important light on the types of objective threat that reduce tolerance among mass publics, but their cross-sectional nature makes it difficult to unravel various causal influences. Theoretically, both political tolerance and threat should be assessed over time to allow for a dynamic analysis of how variation in the level of threat explains fluctuations in political tolerance over time (Peffley et al. 2015; Gibson and Gouws 2003).

The use of experiments—both in the lab and embedded in surveys, opened a second line of investigation for studying the power of different types of threat to lower forbearance. In Malice Toward Some (1995), Marcus et al. manipulated threat, as well as a variety of other conditions (democratic norms, persuasion, source credibility) to enhance our understanding of tolerance. Survey and laboratory experiments have also been employed outside the United States (US) using a variety of “treatment” conditions (e.g., information, counter-arguments, local contexts, and characteristics of groups) (e.g., Gibson 1998a; Gibson and Gouws 2003; Sniderman 1996, 2014). Following 9/11, several US studies manipulated various aspects of terrorist threat (e.g., its salience, imagery, and association with particular groups) to demonstrate that, indeed, increasing threat invariably erodes tolerance, typically measured as support for restrictions on civil liberties (e.g., Merolla and Zechmeister 2009; Albertson and Gadarian 2015,117—124; Malhotra and Popp 2012).

Experimental studies have certainly enriched our understanding of how threatening conditions influence political tolerance. Nevertheless, while experiments doubtless enhance causal inference, manipulations of threatening conditions are unlikely to produce the kind of real, palpable existential threat experienced in countries like South Africa and Israel, or in the United States after 9/11. As Gibson and Gouws (2003) found, for example, manipulating the situational characteristics of civil liberties disputes in South Africa had almost no impact on political tolerance levels because the high levels of objective threat in the country created entrenched views on political tolerance that were not subject to experimental manipulation. If high levels of systemic threat operating in the background in some contexts can overpower experimental manipulations of threat and other conditions, it seems premature to conclude that subjective threat is generally more powerful than objective threat (e.g., Sullivan & Hendricks 2009).

In summary, despite advances in research on political tolerance, few studies have examined the impact of large changes in the objective threat environment on either individual perceptions of threat or the influence of both types of threat on political tolerance. We take up this question in a specific context: Israel. We know from our prior study that fluctuations in the number of terrorist attacks are strongly associated with the rise and fall of political tolerance assessed with the least-liked method over a 30-year period from 1980 to 2011 (Peffley et al. 2015). We also found that much of the downturn in tolerance occurring after an upsurge in terrorist attacks in Israel was due to the reactions of individuals on the Right moving toward intolerance.

Peffley et al. (2015) did not examine individual-level threat perceptions, however, since several of the surveys did not include a measure of perceived threat from least-liked groups (LLGs). Here, we examine the 13 surveys³ that include such a measure of perceived threat to investigate the relationship between fluctuating levels of terror attacks (objective threat), subjective threat perceptions, and political tolerance toward least-liked groups.

Terrorism, Perceptions of Threat, and Tolerance: Hypotheses

Our study combines repeated cross-sections of political tolerance surveys with data on terror attacks that fluctuate dramatically over time, allowing us to investigate the following questions: To what extent do fluctuations in terrorism influence citizens’ perceptions of threat from their least-liked groups? Are shifts in perceived threat driven by reality or do they reside largely in the eye of the beholder? Relatedly, to what extent do individual-level predispositions moderate the translation of objective threat to subjective threat? And finally, how do objective and subjective threat interact to influence political tolerance?

As such, our study is uniquely situated to disentangle the connections between objective threat, subjective threat, and political tolerance. Much of the work on political tolerance is based on single cross-sectional surveys, which naturally focus on individual-level differences. Indeed, such studies find substantial variation in subjective threat perceptions—under presumably fixed levels of systemic threat.

It is therefore not surprising that perceived threat emerges as a strong predictor of intolerance, even though both theory and research acknowledge the importance of objective threat conditions. The advantage of studying a single country where objective threat levels vary significantly over time is clear enough.

Moreover, several developments in Israel since the 1980 study by Shamir and Sullivan (1983) make it likely that the connection between terrorism and intolerance in Israel is strongest among Jewish citizens on the political Right. As was true in 1980, the major cleavage dimension separating the Left and the Right in Israel is defined primarily by the Israeli-Palestinian conflict. After 1980 and up through our final tolerance survey in 2011, Israel faced chronic and persistent terrorism, with over 1500 attacks and two organized terror campaigns⁴ perpetrated mostly by Palestinians in the occupied territories. Since 1980, political identification with the Left and Right has become heavily embedded in different conceptions of collective identity (Shamir and Arian 1999), with the Right more entrenched in its unwillingness to negotiate with the Palestinians in the occupied territories, and viewing Israeli Arabs'¹ as a threatening “fifth column,” while those on the Left attach a higher priority to the values of democracy, including minority rights, and peace. Furthermore, political identities are now more aligned with religious identities, with orthodox and ultra-orthodox Jews consistently identifying with the Right, while Jews on the Left are more often secular.

We are not aware of any study that has examined empirically the joint impact of systemic threat and individual threat perceptions on political tolerance over a prolonged period of time. Two small-scale studies were conducted in Israel during the second (Al-Aqsa) Intifada. Shamir and Sagiv-Schifter’s (2006) case study investigated the impact of the Intifada on tolerance using 5 of our 13 surveys, immediately before and after the onset of atrocities (2000—2002). And Canetti-Nisim and colleagues (2009) used a three-wave panel in 2004—2005 to explore the impact of exposure to terrorism on threat perceptions and intolerance (i.e., what they term, “exclusionary attitudes”) toward Israeli minority groups. In their model, exposure to terrorism significantly (but not strongly) influences threat perceptions, although the effects of exposure largely vanish a year later. By contrast, our research design exploits the chronic, yet fluctuating reality of terrorism in Israel over 30 years by combining surveys assessing perceived threat and tolerance toward LLGs with terrorism data assessed three months prior to each of the 13 surveys.

Consistent with the discussion above, we test the following hypotheses:

Hp Terror attacks should increase individuals’ threat perceptions from LLGs, and for reasons outlined above, this effect should be stronger among Israeli Jews on the Right than those in the Center or Left. Terrorism should have its maximum impact on perceived threat when individuals on the Right select Arab groups as least-liked.

H₂: Both objective threat (terror attacks) and subjective threat perceptions of LLGs are expected to increase political intolerance. But terror attacks can increase intolerance both indirectly, by elevating subjective threat perceptions, and directly, independent of perceptions of threat. In the latter case, elites’ rhetoric and policy enforcement in response to terrorism is a likely source of this link (e.g., Nacos et al. 2011).

H₃: Because individuals on the Kight tend to be more sensitive to social and political threats from a range of outgroups, due to political, personality, and value orientations (e.g., Johnston et al. 2017; Stenner 2005; Feldman 2003), the impact of both objective and perceived threat on tolerance is likely to be stronger for Jews on the Right versus the Left and Center.

Measures and Methods

Method

As noted, we test our hypotheses using 13 national surveys representative of Jewish citizens from 1980 to 2011/’ Table 6.Al in the Appendix lists the 13 surveys (for more details on the surveys, see Peffley et al. 2015). Given that individuals are nested within surveys conducted at different points in time in this study, we test our hypotheses using a linear multilevel model to account for the “nested” nature of our data (Snijders and Bosker 2012).

Measures

Perceptions of Threat. Given the nature of threat in Israel, respondents in various surveys were asked various questions assessing the level of threat their LLG posed to the state, the regime, security, or democracy. For the Israeli case, this more direct measure is an improvement over Sullivan et al.’s proxy measure of perceived threat toward LLGs based on ratings of six semantic differentials (e.g., good-bad, safe-dangerous). Although the wording of the items varies slightly, all emphasize perceptions of socio-tropic or national threat, which has consistently been found to be a much stronger predictor of intolerant responses than egocentric threats to the personal safety of oneself or one’s family (e.g., Gibson 2006; Davis and Silver 2004; Huddy et al. 2005). Responses were coded on a 5-point scale ranging from “not very dangerous at all” (0) to “very dangerous” (4).⁷

Political Tolerance. Political tolerance is measured using Sullivan et al.’s least-liked approach, where respondents are first asked to select the group they like the least from a list of groups (or to suggest a group not on the list), and then asked two political tolerance questions to indicate their willingness to allow the group to “give a speech on TV” and to “demonstrate,” each assessed on a Likert scale ranging from strongly disagree (0) to strongly agree (4), with the resulting additive scale ranging from 0 (low tolerance) to 8 (high).

Political Identification. Left-Right political identification remains one of the most important predictors and moderators of political tolerance in Israel, with individuals on the Left being more willing to extend civil liberties to their least-liked group than those on the Right (e.g., Shamir and Sagiv-Schifter 2006; Peffley et al. 2015). Political identification is coded on a 5-point scale: Right (0), Moderate Right (1), Center (2) Moderate Left (3), and Left (4). For further details on this and other measures, see the online appendix (Section 1) in Peffley et al. (2015).

We also include several socio-demographic variables in the analyses as controls: Religiosity, education, age, female, and income.⁸

Terrorism. Our measure of objective or systemic threat is the total number of terror attacks perpetrated within the internationally recognized borders of Israel three months prior to the survey. The terrorism data are from the Global Terrorism Database (GTD) provided by the National Consortium for the Study of Terrorism and Responses to Terrorism (2016). The GTD data are the most widely used data of their kind, which allows for greater comparability across studies of terrorism. Terrorist Attacks is the number of attacks that occurred within 3 months prior to the start of the survey, which for the surveys in the analysis ranges from 0 to 27, with a mean of 12 attacks.

Other Macrolevel Control Variables. Because past research shows that political tolerance is higher in countries with longer uninterrupted democratic governance, where mass publics experience more democratic learning (e.g., see Peffley and Rohrschneider 2003; Hutchison and Gibler 2007; Peffley et al. 2015), we include first- and second-order quadratic terms to capture the non-monotonic effects of continuous democracy, measured as the number of months Israel had been a democratic state by the time of the survey, along with adjustments to make the coefficient more easily interpreted.⁹ Also, to control for the possibility that the performance of the economy could influence either perceived threat or political tolerance, we include a measure of economic growth, the percentage change in gross domestic product (GDP) per capita one year prior to the survey, based on annual World Development Indicators (World Bank 2014), which range from 0% to 9.2%.

Analysis

Our prior analysis of the entire set of 18 surveys found that as the number of terror attacks rose, support for political tolerance toward least-liked groups among Israeli Jews declined (Peffley et al. 2015). To get a rough visual sense of the relationship between terrorism (objective threat) and tolerance over the 30 years of our study, Figure 6.1 from the 2015 article charts the average level of tolerance across the surveys (1980—2011) versus the total number of terror attacks per year (1979—2010, the years prior to the first and last surveys). We see that terror attacks

FIGURE 6.1 Fluctuations in terrorist attacks and political tolerance (1980-2011) (From Peffley, M. et al.. Am. Pol. Sci. Rev., 109, 823, 2015.)

in Israel have been chronic, persistent, and quite variable over time, fluctuating between 6 and 131 yearly attacks, and that fluctuations in tolerance appear to coincide roughly with the rise and fall of terrorism.

Terrorism and Subjective Threat Perceptions

The purpose of our analysis is twofold—first, to determine how fluctuations in terror attacks influence perceived threat from individuals’ least-liked groups, and, second, to investigate how objective and subjective threat jointly influence political tolerance. Models 1 through 5 in Table 6.1 present the results of several multilevel models testing hypotheses 1 through 3.

We begin by examining the impact of the number of terror attacks on perceptions of threat from least-liked groups in models 1 and 2. Model 1 regresses perceived threat on several micro-level variables (political identity, religiosity, education, income, age, and female) and the three macro-level predictors. Of special interest are the effects of terrorist attacks and the cross-level interaction between political identity (coded as Right to Left) and the number of terror attacks (political identity X terrorist attacks'). The significant coefficient for terror attacks gives the impact of attacks among Jews on the Right (where political identity = 0).^ln The negative coefficient for the interaction term indicates that the impact of terror attacks on perceived threat is significantly stronger for those on the Right versus the Center or Left, consistent with H,.

TABLE 6.1 Impact of Terrorism on Least-Liked Group Threat Perception and Tolerance in Israel, 1980-2011

(Continued)

TABLE 6.1 (Continued)

All coefficients in the table have p < 0.05 except those listed in bold. ' indicates coefficient is significant at the 0.10 level.

Note. Entries are maximum likelihood coefficients estimated for the linear multilevel model with a random coefficient specification using Stata 14, with standard errors in parentheses. Higher values on the following variables indicate: Greater political tolerance, political identity with the left, female, orthodox religiosity higher threat perception, GDP growth (1 year lag), number of terrorist attacks in the three months prior to the survey, and continuous yeats of democracy.

The full test of H| is provided in model 2, where we add a dummy variable to distinguish whether the respondent selected an Arab or a non-Arab (i.e., Jewish) LLG, as well as a three-way interaction between political identity, terror attacks, and the Arab LLG dummy variable in addition to all lower-order terms. The significant coefficient associated with the Arab LLG dummy indicates that among right-wing Jews, Arab LLGs are viewed as more threatening than non-Arab groups even when the number of terror attacks prior to the survey equals 0.

Because coefficients associated with three-way interactions are difficult to interpret, we plot the effects of the interactions between political identity and terror attacks separately for Arab (right-hand panel) and non-Arab LLGs (lefthand side) in Figure 6.2, based on the coefficients from model 2. Focusing on the right-hand panel, terrorism clearly has the greatest impact in increasing perceptions of threat toward Arab LLGs amongjews on the Kight. Calculating the slopes for terrorism, the coefficients decrease from the Right (0.016, p = 0.001) to the Center (0.009, p = 0.04) and become small and insignificant for those on the Left (0.002, p = 0.75). Thus, consistent with H_b terrorism has its greatest impact in elevating perceptions of threat amongjews on the Right who select Arab political groups as most disliked.

FIGURE 6.2 Effect of terrorism on threat perceptions of Arab and non-Arab LLGs across political identity. Predicted values are based on the coefficients of model 2 in Table 6.1 (Courtesy of Author.)

Also, judging from the graph on the left-hand side of Figure 6.2, even among individuals who selected Jewish (i.e., non-Arab) LLGs, terrorism again has its greatest impact in elevating perceptions of threat among right-wing Israelis." Comparing the effects of terrorism among right-wingers across the two panels of Figure 6.2, terrorism elevates perceived threat toward least-liked Arab groups more than Jewish groups, and this difference is significant. What is most telling is that among right-wingers, terrorism also significantly raises perceptions of threat toward Jewish least-liked groups, most of whom are leftist groups that are ideological “fellow travelers” of Arab groups because they support equal political rights for Israeli Arabs and negotiation with the Palestinians in the occupied territories (Peffley et al. 2016).

To summarize, in line with our first hypothesis, subjective perceptions of threat are clearly a function of the objective threat context (i.e., terror attacks), but the power of terrorism to move threat perceptions is moderated by individuals’ political identities as well as the types of domestic groups they view as enemies. Thus, subjective threat perceptions are not completely exogenous, nor are they impervious to real-life threat conditions. Fluctuations in the number of terror attacks help drive the ups and downs in perceived threat from LLGs, at least among the Right in Israel.

On the other hand, our results should not be interpreted as a repudiation of the oft-noted finding that individual-level characteristics are not strong predictors of perceived threat from LLGs. For example, if we regress perceived threat on the six micro-level predictors for each of the 13 surveys, few predictors emerge as significant in either a statistical or substantive sense (results not shown).Thus, while perceptions of threat from LLGs are shaped by fluctuations in terrorism and this impact is moderated by political identity, a large portion of the variance in subjective threat remains unexplained, consistent with Sullivan’s work.

Terrorism, Subjective Threat Perceptions, and Political Tolerance

A second question is whether threat—either objective or subjective, helps explain fluctuations in political tolerance over time, particularly among right-wingers.To test hypotheses 2 and 3, we estimated multilevel models 3, 4, and 5 in Table 6.1, where least-liked political tolerance is a function of all the macro and micro predictors in model 2, including the interaction between terrorism and Left-Right identity, which we know from prior research is significant (Peffley et al. 2015). Model 4 adds the 5-point scale assessing perceived threat from LLGs. By comparing the coefficient for terrorism in model 3 with that of model 4, after perceived threat is added, we can make a rough assessment of whether subjective threat mediates the effects of terrorism on political tolerance.

The results for model 3 confirm that terrorism has a greater impact on political tolerance for Israeli Jews on the Right than those in the Center or on the Left. Also, as predicted, in model 4, the impact of perceived threat on political tolerance is substantial (—0.46, p = 0.001); moving one full point on the 5-point threat scale is associated with a half-point reduction in the 9-point tolerance scale. More importantly for current purposes, once we add perceived threat to model 4, the coefficient for terrorism among right-wingers (—0.035), while still significant, is about 20% smaller than in model 3 (—0.044), before perceived threat is added. This means that while some of the impact of terrorism on tolerance is mediated by perceptions of threat from LLGs, much of the power of terrorism to move people toward intolerance is independent of perceived threat, consistent with H₂.

To test H₃, which predicts that the impact of both terrorism and perceived threat on political tolerance should be greater among the Kight, we estimate model 5, which includes a three-way interaction between perceived threat, terrorism, and political identity (LLG Threat Perception X Terrorist Attacks X Political ID), plus all lower-order terms. Perceived threat in this model is measured as a dummy variable indicating low and high threat (split at 2.5) to simplify the interpretation of its effects. In model 5, as expected, the impact of the perceived threat dummy on political intolerance is relatively strong. The coefficient for threat (—1.09) indicates that even when terrorism equals 0, those on the Kight are a full point less tolerant toward their LLG if they think their group poses a high versus a low threat to the state.

To gain a better sense of how the effects of terrorism vary across political identity and perceived threat. Figure 6.3 plots predicted values of tolerance across terror attacks for individuals who perceive low versus high threat toward their LLG among three political identity groups: The Right, Center, and Left (based on model 5). Several results stand out in the Figure. First, confirming prior analysis, terrorism takes its greatest toll in eroding tolerance

FIGURE 6.3 Effect of terrorism and threat perception on political tolerance across political identity. Predicted values and 95% confidence intervals are based on the coefficients of model 5 in Table 6.1 (Courtesy of Author.)

toward LLGs among right-wingers compared with centrists and left-wingers. The slope coefficients for terrorism are about half the size among centrists as they are for right-wingers, and among leftists the effect of terrorism on tolerance is not significantly different from 0. However, unlike realistic threat, the impact of perceived threat does not vary significantly across Left-Right groups, which runs counter to H_t. Moreover, the slope for terrorism does not vary across levels of perceived threat among any of the three Left-Right groups. Although perceiving higher threat clearly reduces forbearance toward LLGs, the effect of terrorism is approximately the same at high and low levels of subjective threat.¹²

To summarize our findings, terrorism affects threat perceptions, although there is far from a one-to-one correspondence between rising terrorism levels and increasing perceptions of threat. Thus, we can say that perceived threat is “exogenous” only in the sense that individual characteristics do not move it. We also find that both terrorism and perceived threat influence tolerance. While some of the impact of terrorism on tolerance is mediated by threat perceptions, most of its influence in driving tolerance appears independent of subjective threat perceptions, a result we discuss further in the conclusions.

For now, we note that our findings show that political identity moderates the impact of terrorism on both perceived threat and political tolerance (Figures 6.2 and 6.3). Those on the Right are far more likely to shift their views in response to terrorism than centrists or leftists. Part of the reason that right-wingers move more in response to terrorism is that they are more likely to select Arab than Jewish LLGs. But this is not the whole story. Neither centrists nor leftists who select Arab LLGs respond to terrorism as strongly as right-wingers do in increasing their perceptions of threat (Figure 6.2) or lowering their levels of tolerance toward Arab groups (not shown here). Moreover, Israelis on the Right also downgrade their tolerance toward Jewish LLGs in response to terrorism, whereas centrists and leftists do not.

On the other hand, political identity does not moderate the effects of perceived threat on tolerance (Figure 6.3). Although perceived threat toward LLGs is responsive to fluctuations in terrorism, a substantial portion of the variance in subjective threat does not appear to move in response to exogenous forces like terrorism or endogenous forces like political identity.

Conclusions

The major goal of this chapter is to contribute to our understanding of the seemingly intuitive yet empirically enigmatic relationship between threat and intolerance. In many ways, our study has followed in John Sullivan’s footsteps, both theoretically and empirically. Theoretically, we build on his conceptualinnovation of defining tolerance as a willingness to extend basic political freedoms to groups one opposes, on the pivotal role of threat in explaining variation in tolerance, and on the deep appreciation of context for understanding tolerance judgments. Empirically, we use his least-liked approach to measure tolerance, and, to the extent that our data allowed, his now classic individuallevel model.

At the same time, our study departs from those of Sullivan and others by focusing on differences in systemic threat within a single country over time instead of cross nationally. Moreover, the variation in the threat environment in our study stems from a particular source—i.e., terrorist attacks that fluctuate over time. Naturally, the implications of our findings for understanding tolerance must be viewed in light of the specific context and purpose of our study, which is to understand the influence of changes in the objective threat environment—i.e., terrorism.

Moreover, although we rely on Sullivan et al.’s least-liked group method, we depart from its traditional use and apply it in a way that suits our research question: How a very specific type of threat, the real-world threat of terrorism, affects threat perceptions and political tolerance. For this type of research question, groups—both ingroups and outgroups—matter. It was thus important to examine relationships separately for particular types of LLGs (i.e., Israeli Arab and Jewish LLGs) associated with the violence by particular ingroups (i.e., Jews on the Right).¹³

Our study challenges the prevailing wisdom of tolerance studies using the least-liked method that threat perceptions are entirely exogenous. We found that perceptions of threat from least-liked groups are moved significantly by fluctuations in the systemic (objective) threat environment. Moreover, the degree to which threat perceptions move in response to terrorism hinges on individuals’ predispositions, which critically shape the way they perceive the environment. Specifically, when terrorism escalates, Israeli Jews who identify with the Right were more likely than leftists or centrists to elevate their perceptions of threat from least-liked groups. Perceptions of threat also vary across the characteristics of the least-liked groups people select. Because most of the attacks during the 30 years of our study were perpetrated by Palestinian terrorists from the occupied territories, least-liked Arab domestic groups were viewed more threateningly when terrorism escalated. In addition, increased terror attacks prompted right-wingers to elevate their perceptions of threat toward Jewish (leftist) least-liked groups, rendering them “fellow travelers,” similar to the 1950s in the United States, when the threat from Communism spread to socialists, atheists, and other groups on the left (Stouffer 1955).

As for political tolerance, our analyses here reinforce previous findings (Peffley et al. 2015, 2016) that prolonged terrorist campaigns reduce tolerance. But the major contribution of this chapter is in demonstrating that the impact of terrorism is only partly mediated by an increase in perceived threat from LLGs. Although terrorism has an indirect effect on tolerance through threat perceptions, this is only a minor part of the story, since the impact of terrorism is only slightly diminished once threat perceptions are taken into account, a finding that is contrary to our expectations. Most of the impact of terrorism on tolerance is direct, and importantly, is significantly moderated by political identification, with the strongest impact being among right-wingers.

Putting these two results together—i.e., small mediation and substantial moderation, our interpretation is that social and political processes outside individual threat perceptions play an important role in shaping public support for tolerance. Political, cultural, and media elites likely play a substantial role in this process. They may target and vilify domestic groups when threat levels rise, activate pre-existing threat perceptions and connect them to tolerance judgments, and raise and legitimize intolerant policy initiatives (or not). Elites’ public messages, legislative efforts, and policy enforcement in response to terrorism likely have a crucial impact on the climate of opinion that influences citizens’ concerns about terrorism and forbearance. Contrast, for example, the calming response of Danish elites to the potentially inflammatory events surrounding the “Cartoon Crisis” (Sniderman et al. 2014) to the incendiary rhetoric of Republican candidates during the 2016 presidential primaries or right-wing elites in Israel before and after the 2015 Knesset election.

Taken together, our results undermine the claim that it is only subjective threat that matters and not objective threat. Dramatic shifts in real-world threat conditions, such as an increase in terror attacks, impact citizens’ support for political tolerance. The fact that this influence is only partially mediated by perceived threat from LLGs clearly warrants further research by political scientists.

Finally, we wish to emphasize the value of studies such as ours in research fields with shifting threat contexts, including existential threat. Notwithstanding the value of experimentally manipulated situations, one cannot lose sight of the primary importance of a country’s context. While our study focuses on only one country, Israel offers a real-life laboratory for the study of political tolerance, and, in particular, how terror attacks influence those attitudes. It is no coincidence that Israel was one of the first countries outside the United States where political tolerance was studied using Sullivan’s methodology and remains an important site of recent innovative work in this area. Moving forward, further research along these lines within the framework of the Sullivan paradigm is very much needed to provide a clear picture of the complex interplay between objective threat, threat perception, and political tolerance.

Appendix

TABLE 6.Al Survey Information

Source: Peffley, M. et al., Am. Pol. Sri. Rev.. 109, 4,2015.

Notes

• 1 Caspi and Seligson conducted the very first, but small-scale study of political tolerance in Israel utilizing the least-liked group methodology in 1979 (1983). We are grateful to Yochanan Peres, Eppie Yuchtman-Yaar, and Noah Lewin-Epstein for sharing their survey data with us. We are indebted to Yasmin Alkalai, Yael Proaktor, and Jason Kehrberg for their technical assistance in the preparation of the data.
• 2 American cross-sectional survey studies conducted following 9/11 provide a similar extension of tolerance research to varying conditions of threat (e.g., Huddy et al. 2005; Davis and Silver 2004; Sullivan and' Hendricks 2009).
• 3 See Stegmueller (2013) who demonstrates that the use of as few as 12 units in multilevel analysis still produces robust results.
• 4 The first and second Intifadas, between 1987-1993 and 2000-2005.
• 5 We use here the terms “Palestinians” to refer to the inhabitants of the occupied territories and “Israeli Arabs" for Palestinians who are citizens of Israel, making up a minority of about 20% of Israel's population. Both groups share kinship, national identity, and most are Muslims.
• 6 Our analysis relies on Jewish respondents because several surveys in our data set included only Jews and because when studying terrorism in Israel and the threat it entails, it is appropriate to study separately Jewish and Arab citizens, as is commonly done.
• 7 In the seven surveys that asked respondents whether their LLG posed a threat to the security of the country and to democracy, responses to the two items were combined (r = 0.47) and coded to a 0 to 4 scale.
• 8 We note that due to data limitations, we do not include controls for several individual-level predictors of tolerance, such as democratic norms and psychological security. Based on prior studies using the least-liked approach, we are confident that our results do not suffer from omitted variable bias; however, we acknowledge that the omission limits our ability to identify important micro-level causal mechanisms (see Peffley et al. 2015).
• 9 Specifically, we convert the months to years by dividing by 12, and set the first survey in 1980 to 0, which means the last survey in 2011 equals 30.4 years. We set the first survey to 0 to give the constant in the multilevel models a meaningful interpretation (i.e., when all predictors equal 0).
• 10 The coefficient of 0.018 indicates that moving from the minimum to the maximum number of attacks (i.e., from 0 to 27) shifts perceptions of threat from LLGs among right-wingers about one half of one point on the 5-point threat scale.
• 11 Among centrists, the flat slope (b = 0.000) in Figure 6.2 shows that terrorism has no real effect on perceived threat from Jewish LLGs, and the slope for left-wing Jews is negative. Leftists select mostly right-wingjews as least-liked, which they do not associate with terrorism. Centrists select Jewish LLGs on both the Left and the Right. For the distribution of LLGs by political identity, see Table A3 in the online appendix in Peffley et al. (2015).
• 12 We obtain essentially the same results when we estimate model 5 only for the 3211 respondents who selected Arab LLGs.
• 13 Sullivan et al. (1993) used a similar strategy to examine politicians who selected Kach vs. other groups (in Israel) and Sinn Fein vs. other groups (in Britain) to explore the sources of political intolerance among political elites.

References

Albertson, Bethany, and Shana Kushner Gadarian. 2015. Anxious Politics. New York: Cambridge University Press.

Canetti-Nisim, Daphna, Eran Halperin, Keren Sharvit, and Stevan E. Hobfoil. 2009. “A New Stress-Based Model of Political Extremism: Personal Exposure to Terrorism, Psychological Distress, and Exclusionist Political Attitudes.” Journal of Conflict Resolution 53 (3): 363-389.

Davis, Darren W., and Brian D. Silver. 2004. “Civil Liberties vs. Security in the Context of the Terrorist Attacks on America.” American Journal of Political Science 48 (1): 28-46.

Feldman, Stanley. 2003. “Enforcing Social Conformity: A Theory of Authoritarianism.” Political Psychology 24 (1): 41-74.

Gibson, James L. 1998a. “A Sober Second Thought: An Experiment in Persuading Russians to Tolerate.” American Journal of Political Science 42 (3): 819—850.

Gibson,James L. 1998b. “Putting Up with Fellow Russians: An Analysis of Political Tolerance in the Fledgling Russian Democracy.” Political Research Quarterly 51 (1): 37—68.

Gibson, James L. 2006. “Enigmas of Intolerance: Fifty Years after Stouffer’s Communism, Conformity, and Civil Liberties.” Perspectives on Politics 4 (1): 21—34.

Gibson, James L., and Amanda Gouws. 2003. Overcoming Intolerance in South Africa. Cambridge: Cambridge University Press.

Huddy, Leonie, Stanley Feldman, Charles Taber, and Galiya Lahav. 2005. “Threat, Anxiety, and Support for Antiterrorism Policies." American Journal of Political Science 49 (3): 593-608.

Hutchison, Marc L. 2014. “Tolerating Threat? The Independent Effects of Civil Conflict on Domestic Political Tolerance." Journal of Conflict Resolution 58 (5): 796—824.

Hutchison, Marc L., and Douglas M. Gibler. 2007. “Political Tolerance and Territorial Threat: A Cross-National Study.” Journal of Politics 69 (1): 128—142.

Johnston, Christopher D., Howard G. Lavine, and Christopher M. Federico. 2017. Open Versus Closed: Personality, Identity, and the Politics of Redistribution. Cambridge: Cambridge University Press.

Malhotra, Neil, and Elizabeth Popp. 2012. “Bridging Partisan Divisions over Antiterrorism Policies:The Role ofThreat Perceptions.” Political Research Quarterly 65 (1): 34—47.

Marcus, George E., John L. Sullivan. Elizabeth Theiss-Morse, and Sandra L. Wood. 1995. With Malice Toward Some. Cambridge: Cambridge University Press.

Merolla, Jennifer L., and Elizabeth J. Zechmeister. 2009. Democracy at Risk. Chicago: University of Chicago Press.

Nacos, Brigitte L., Yaeli Bloch-Elkon, and Robert Y. Shapiro. 2011. Selling Fear. Chicago: University of Chicago Press.

National Consortium for the Study of Terrorism and Responses to Terrorism (START). 2016. Global Terrorism Database. Available at: http://www.start.umd.edu/gtd [Accessed October 2, 2016].

Peffley, Mark, and Robert Rohrschneider. 2003. “Democratization and Political Tolerance in Seventeen Countries: A Multi-level Model of Democratic Learning.” Political Research Quarterly 56 (3): 243-257.

Peffley, Mark, Marc L. Hutchison, and Michal Shamir. 2015. “The Impact of Persistent Terrorism on Political Tolerance: Israel, 1980 to 2011.” American Political Science Review 109 (4): 817-832.

Peffley, Mark, Marc L. Hutchison, and Michal Shamir. 2016.“Tolerating ‘FellowTravelers’: Terrorism and Political Tolerance toward Israeli Arabs, 1980—2011." Paper Presented at the Annual Meeting of the American Political Science Association, September 1—4, Israel.

Shamir. Michal, and Asher Arian. 1999. “Collective Identity and Electoral Competition in Israel.” American Political Science Review 93 (2): 265—277.

Shamir, Michal, and John Sullivan. 1983. “The Political Context of Tolerance: The United States and Israel." American Political Science Review 77 (4): 911-928.

Shamir, Michal, and Tammy Sagiv-Schifter. 2006. “Conflict, Identity and Tolerance: Israel in the Al-Aqsa Intifada.” Political Psychology 27 (3): 569-596.

Sniderman, Paul M., Joseph F. Fletcher, Peter H. Russell, and Philip E. Tetlock. 1996. The Clash of Rights. New Haven, CT: Yale University Press.

Sniderman, Paul M., Michael Bang Petersen, Rune Slothuus, and Rune Stubager. 2014. Paradoxes of Liberal Democracy. Princeton: Princeton University Press.

Snijders, Tom A.B., and RoelJ. Bosker. 2012. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling, 2nd ed. London: Sage Publishers.

Stegmueller, Daniel. 2013. “How Many Countries for Multilevel Modeling? A Comparison of Frequentist and Bayesian Approaches.” American Journal of Political Science 57 (3): 748-761.

Stenner, Karen. 2005. The Authoritarian Dynamic. Cambridge: Cambridge University Press.

Stouffer, Samuel. 1955. Communism, Conformity, and Civil Liberties. New York: Doubleday.

Sullivan, John L., and Henriet Hendricks. 2009. “Public Support for Civil Liberties Pre-and Post-9/11.” Annual Review of Law and Social Science 5 (1): 375—391.

Sullivan, John L., James Piereson, and George E. Marcus. 1982. Political Tolerance and American Democracy. Chicago: University of Chicago Press.

Sullivan, John L., Michal Shamir, Nigel S. Roberts, and Patrick Walsh. 1984. “Political Intolerance and the Structure of Mass Attitudes: A Study of the United States, Israel, and New Zealand.” Comparative Political Studies 17 (3): 319-344.

Sullivan, John L., Michal Shamir, Patrick Walsh, and Nigel S. Roberts. 1985. Political Tolerance in Context. Boulder: Westview Press.

Sullivan, John L., Pat Walsh, Michal Shamir, David G. Barnum, and James L. Gibson. 1993. “Why Politicians Are More Tolerant: Selective Recruitment and Socialization among Political Elites in Britain, Israel, New Zealand and the United States." British Journal of Political Science 23 (1): 51-76.

World Bank. 2014. World Development Indicators. Available at: http://data.worldbank. org/data-catalog/world-development-indicators [Accessed May 1, 2014|.