A New Stress-Based Model of Political Extremism

Abstract

Does exposure to terrorism lead to hostility toward minorities? Drawing on theories from clinical and social psychology, we propose a stress-based model of political extremism in which psychological distress—which is largely overlooked in political scholarship—and threat perceptions mediate the relationship between exposure to terrorism and attitudes toward minorities. To test the model, a representative sample of 469 Israeli Jewish respondents was interviewed on three occasions at six-month intervals. Structural Equation Modeling indicated that exposure to terrorism predicted psychological distress (t1), which predicted perceived threat from Palestinian citizens of Israel (t2), which, in turn, predicted exclusionist attitudes toward Palestinian citizens of Israel (t3). These findings provide solid evidence and a mechanism for the hypothesis that terrorism introduces nondemocratic attitudes threatening minority rights. It suggests that psychological distress plays an important role in political decision making and should be incorporated in models drawing upon political psychology.

Ongoing terrorism, especially the indiscriminant kind targeting unarmed civilians, is one of the most severe challenges facing human societies in the twenty-first century. The most direct ramifications of terror are heavy loss of human life and a disproportional reaction characterized by feelings of personal and collective fear, and behavioral responses to that fear. Israeli society has been exposed to a prolonged period of ongoing terrorism during the Al Aqsa Intifada, in which both Israeli and Palestinian societies have suffered thousands of fatalities and injuries. In Israel alone, repeated attacks led to the deaths of about one thousand Jews and Arabs from September 2000 until 2006, mostly as a result of suicide bombings (Israel Ministry of Foreign Affairs 2007). These were especially stressful as they could occur any-where and places with heavy civilian traffic such as buses, restaurants, and night-clubs were targeted.

Alongside the heavy losses and fear, terror creates an enormous challenge to the fabric of democratic societies. In many cases, there is a difficult inner tension between the fundamental need to feel secure and the aspiration to sustain democratic values and preserve democratic culture. More specifically, in times of terrorist threat and severe losses, when direct confrontation with the perpetrators of terrorism is either impossible or does not guarantee public safety, rage is frequently aimed at minority groups and their members. This rage can be easily translated into support for non-democratic practices in dealing with minorities. Hence, one of the key psychosocial–political consequences of terrorism is the development of hostile feelings, attitudes, and behaviors toward minority groups. This process may be exacerbated when these minority groups are believed to be associated with the perpetrators of terrorism (Bar-Tal and Labin 2001; Pettigrew 2003; Skitka, Bauman, and Mullen 2004). In the present investigation, we examine the multimediated effect (Preacher and Hayes 2008) of degree of exposure to terrorism on exclusionist attitudes.

In this article, we propose an infrastructure for understanding the effects of threat perceptions on political extremism in the face of ongoing conflict, a continuous sense of existential insecurity, and growing psychological distress—which is generally unnoticed in political research. We argue that the personal exposure to political violence that results in psychological distress affects political worldviews. This happens when people who are exposed to traumatic events feel threatened and vulnerable as a result of their exposure to terrorism and the climate that surrounds terrorism. Such psychological distress, in turn, exacerbates perceptions of threat, which further invoke “threat buffers” such as political exclusionism.

Exclusionist political attitudes reflect the desire by members of the majority in a certain society to exclude minorities, mainly by using societal and political mechanisms (Coenders and Scheepers 2003). Coenders (2001) and Sniderman, Hagendoorn, and Prior (2004) distinguished between different types of exclusionism, referring to different target out-groups, such as resident ethnic out-group members, immigrants, or political refugees. In the current study, we focus on the political aspects of ethnic exclusionism and mainly on the opposition to the granting of civil and political rights to resident group members—the Palestinian citizens of Israel (PCIs).

A large corpus of literature indicates that perceived threat leads to more exclusionist attitudes (see Canetti-Nisim, Ariely, and Halperin 2008; Shamir and Sagiv-Schifter 2006) and that such attitudes are more prevalent immediately after the occurrence of major terror attacks (Huddy et al. 2002, 2005; Skitka, Bauman, and Mullen 2004). Quite surprisingly, however, the long-term effects of ongoing terror attacks on antidemocratic attitudes have rarely been studied. In addition, only a handful of studies (e.g., Hobfoll, Canetti-Nisim, and Johnson 2006), have focused on the implications of degree of personal exposure to terrorism for exclusionist political attitudes. In this regard, although the influence of terrorism is broader than its effects on direct victims, individuals' degree of exposure may be an important factor in their perception of threat, and in consequent exclusionist attitudes. Such variations in degree of exposure are especially likely in contexts where terrorism is ongoing, because some individuals may be exposed to more than one event and to different events in different ways. A thorough understanding of the processes through which personal exposure to terrorism leads to exclusionist political attitudes may contribute to broader insights regarding the manner in which terrorism and anti-democratic or aggressively militant reactions to terrorism feed each other. We propose that a thorough account of the relationship between exposure to terrorism and exclusionism should integrate contextual, psychological, and political variables. Accordingly, our investigation draws insights from theoretical approaches based in clinical psychology, social psychology, and political science.

Numerous studies have examined the psychological impact of exposure to terrorism (Bleich, Gelkopf, and Solomon 2003; Galea et al. 2002; Miguel-Tobal et al. 2006; Rubin et al. 2007; Schuster et al. 2001; Shalev and Freedman 2005; Silver et al. 2002) and others have investigated terrorism's political consequences (Bar-Tal and Labin 2001; Huddy et al. 2002, 2005; Skitka, Bauman, and Mullen 2004). However, only a few studies, of limited scope and design, have addressed the relationships between psychological consequences of personal exposure—which is more pervasive than would seem—to terrorism and political outcomes (Bonanno and Jost 2006; Hobfoll, Canetti-Nisim, and Johnson 2006).

The aim of the present study is to expand the understanding of these relationships by employing a more sophisticated research design than was previously employed. Specifically, we conducted a three-wave longitudinal study among a representative national sample of Israeli Jews during the period of the Al Aqsa Intifada. This design allows examination of varying levels of exposure to terrorism and enhances the potential generalizability of the findings. The use of a longitudinal design also enables cautious inference of causal relations, as well as an investigation of the long-term implications of ongoing terrorism and the persistence of the effects of exposure to it. A similar approach to the investigation of the long-term effects of exposure to terrorism was used in previous studies (Rubin et al. 2007; Silver et al. 2002). However, these studies were limited to isolated terror events rather than ongoing terrorism and addressed only psychological and not political aspects. The present research design allows us to explore the manner in which psychological processes might mediate the relationship between personal exposure to terror and exclusionist political attitudes toward minority group members.

Literature Review, Research Hypotheses, and Model

Perceived Threat and Exclusionist Attitudes

Perceived threat is considered the single best group-level predictor of exclusionism and intolerance (Quillian 1995; Sullivan, Pierson, and Marcus 1982). It gives expression to individuals' cognitive evaluation regarding the manner in which out-group members interfere with the achievement of individual or group goals (e.g., security, financial welfare, or positive identity). Extensive research in social psychology has demonstrated that threatened individuals and groups commonly cope with threat by adopting hostile attitudes toward out-groups, attempting to reduce their relative power, and supporting actions that potentially harm out-group members in various ways (see e.g., Duckitt and Fisher 2003; Maddux et al. 2008; Morrison and Ybarra 2008; Pettigrew 2003; Stephan and Renfro 2002; Stephan and Stephan 2000; Watts 1996). In politics, such individual tendencies may lead to the initiation of collective exclusionist policies, especially in democratic systems where citizens are capable of affecting public policy, at least to some extent.

One explanation for the increased hostility toward out-groups in response to threat can be found in Terror Management Theory (TMT) (Greenberg, Pyszczynski, and Solomon 1986). The theory suggests that cultural worldviews serve the function of buffering anxiety resulting from awareness of the inevitability of death. The prediction that follows is that mortality salience (MS) motivates affirmation of the cultural worldview and derogation of outsiders who advocate a different worldview. Accordingly, past studies found that MS increased intolerance (Greenberg et al. 1990) and recent studies identified similar patterns among U.S. citizens following the terrorist attacks of September 11, 2001 (Pyszczynski, Solomon, and Greenberg 2003). It should be noted, however, that MS effects in TMT studies are usually observed shortly after the MS-inducing event or experimental manipulation. To the best of our knowledge, there has been no research regarding the degree to which MS effects persist for months or even years following exposure. Concerned with the long-term effects of exposure to terrorism, the present study may offer an extension of TMT research.

The social-psychological theory of integrated threat describes various sources of threat that prompt bigotry (Stephan and Renfro 2002; Stephan and Stephan 2000), and specifically distinguishes between realistic and symbolic threats. Realistic threat refers mainly to potential harm to tangible or concrete objects (e.g., money, land, human life), whereas symbolic threat includes various potential threats to relatively abstract aspects of the collective, such as threats to the in-group's identity, value system, belief system, or worldview (e.g., language, religion, morality) (Duckitt 2001). In addition to this typology, scholars have pointed to the different nature and implications of personal versus national threat in the context of terrorism (Huddy et al. 2002). In the current work, we focus mainly on the second type—the perceived threat to the collective—which was found to have greater influence on questions of restrictions of rights and liberties.

We argue that perceived threat mediates the association between personal exposure to terrorism and exclusionist political attitudes, because perceived threat is at least to some extent a reflection of actual threatening occurrences. Specifically, we posit that individuals who are personally exposed to terrorism will display heightened levels of political exclusionism to the extent that exposure leads to elevated levels of perceived future threat. In the following paragraphs, we delineate the theoretical rationale for this argument.

Exposure to Terrorism, Psychological Distress, and Perceived Threat

More than a decade ago, Chemtob and his colleagues (Chemtob et al. 1997), who studied psychological reactions of combat veterans, identified a sequence of experiences that led individuals who were exposed to horrifying experiences to display extreme forms of negative affect toward other individuals. They defined this syndrome as “the ball of rage.” Similarly, a study of noncombatants recently conducted in Israel found an association between exposure to terrorism and exclusionist attitudes toward minorities (Hobfoll, Canetti-Nisim, and Johnson 2006). In both studies, however, the relations between exposure to the horrifying events and rage or political exclusionism were mediated by high levels of psychological distress. In other words, it was not the exposure itself that set the wheels of rage in motion, but the distress that resulted from the exposure.

Not surprisingly, people who experience serious loss, disruption, injury, or the death of a loved one following a traumatic event tend to show more severe psychological distress than do those who suffer fewer consequences (Hobfoll, Canetti-Nisim, and Johnson 2006; Katz et al. 2002; Pfefferbaum 1997; Yule, Perrin, and Smith 2001). Yet, terrorism is deliberately aimed at inducing psychological distress as well among individuals who were indirectly exposed to the attacks, even if they personally did not suffer any tangible losses. For example, individuals may experience psychological distress upon learning that a significant other or even an acquaintance was exposed to terrorism (Hobfoll et al. 2008). Hence, in the current work, we use a comprehensive concept of personal exposure to terrorism, which includes both direct losses and indirect experiences of exposure through loved ones and acquaintances. The concept of personal exposure precludes exposure to “mere knowledge” of terrorism, as in media coverage.¹

Studies of the psychological implications of terrorist attacks in the United States (Galea et al. 2002; Schuster et al. 2001; Silver et al. 2002), Spain (Miguel-Tobal et al. 2006), and Israel (Bleich, Gelkopf, and Solomon 2003; Shalev and Freedman 2005) have pointed to posttraumatic stress disorder (PTSD) as one of the most prominent and prevalent expressions of psychological distress following such severe attacks. PTSD is a condition that develops in response to witnessing or experiencing a threatening or harmful event that elicits fear, helplessness, or horror. Other than exposure to a traumatic event, the criteria for diagnosis of PTSD include persistent reexperiencing of the trauma (e.g., intrusive memories and nightmares), active avoidance of reminders of the trauma and general numbing, and symptoms of hyperarousal (e.g., anger, sleep disturbance, hypervigilance). Furthermore, at least six symptoms must be present for at least one month and cause clinically significant distress or impairment in functioning (American Psychiatric Association 1994). Because of the multiple criteria for diagnosis, many individuals may not be diagnosed as suffering PTSD, yet still display related symptoms and experience considerable psychological distress. In this study, we use the degree to which individuals experience symptoms of PTSD as an indicator of psychological distress experienced in response to terrorism, regardless of whether they meet the criteria for PTSD diagnosis.

Psychological distress, let alone PTSD, is largely overlooked in traditional political scholarship. The hits for different variants of PTSD in databases of political science are few and far between. Despite the broad attention that psychologists and mental health practitioners have devoted to psychological distress in response to terror and other political traumas, studies in political science and even in political psychology rarely use it as an explanatory factor in their models (for a review, see Koopman 1997). At the same time, however, in addition to democratic values and threat perceptions, traditional political science (Sullivan et al. 1981) did acknowledge the importance of a subjective experience of “psychological insecurity” to inducing intolerance and exclusionist attitudes. Likewise, in the context of terrorism, the work of Huddy et al. (2005) looked at the effects of anxiety resulting from a single terrorist attack on support of antiterrorism policies. We propose that in the face of ongoing terrorist attacks, psychological distress can serve as one of the most suitable proxies for representing such subjective feelings of insecurity. Hence, we accept the challenge presented by Koopman a decade ago (1997) and incorporate psychological distress as manifested in PTSD symptoms into our comprehensive model for explaining exclusionism in the face of terror.

An interesting question, therefore, is what makes individuals that demonstrate symptoms of PTSD more susceptible to developing high levels of political exclusionist attitudes. As previously noted, we suggest that the answer to this question lies with the concept of perceived threat. According to the shattered assumptions approach (Janoff-Bulman 1992), following traumatic events people often face major challenges to their basic assumptions about the world and about themselves. Two assumptions that are most influenced by severe events are the assumption of personal invulnerability and the perception of the world as meaningful, predictable, and benign. Accordingly, psychological distress in response to traumatic events is associated with a perception of the world as malevolent and the self as vulnerable to victimization. Other scholars have suggested that individuals who display psychological resilience (i.e., bounce back quickly or do not develop distress at all) in the face of traumatic events tend to perceive the world as meaningful, predictable, and controllable (Antonovsky 1987; Kobasa 1985; Taylor 1983). Accordingly, we propose that individuals who experience major psychological distress (e.g., symptoms of PTSD) in response to traumatic events will tend to see the world as unpredictable, threatening, and dangerous more than those who are relatively resilient in the face of such events and those who have not been exposed to them.

Empirical support for this contention can be found in several psychological studies conducted in various settings around the world. Magwaza (1999) found that individuals who were traumatized by representatives of the South African apartheid government perceived the world as a more dangerous place. In Israel, Kutz and Dekel (2006) recently found that individuals who were exposed to terror and had PTSD symptoms displayed higher levels of perceived threat from an Iraqi missile attack than those who were not exposed to terror or did not experience such symptoms (see also Foa et al. 1999; Solomon, Iancu, and Tyano 1997). Applying the same rationale, we suggest that psychological distress resulting from terror events will increase perceptions of threat in the world in general, and particularly perceptions of threat posed by minority groups believed to be related to the perpetrators of the traumatic events.

A New Stress-Based Model of Political Extremism

We propose the following model (see Figure 1) for predicting exclusionist political attitudes toward PCIs among Israeli Jews. First, we hypothesize that perceived threat from PCIs would be the most prominent predictor of exclusionist political attitudes toward them. Moreover, threat perception will mediate the influence of other previously studied variables on exclusionist political attitudes. Research indicates that exclusionist political attitudes are affected not only by heightened threat perceptions and psychological distress, but also by demographics and various ideological beliefs. Drawing on the theoretical foundations of Sullivan and colleagues (1981), we included similar factors in our model, particularly those receiving empirical support in subsequent works in various contexts (Canetti-Nisim and Pedahzur 2003; Eisenstein 2006; Halperin, Canetti-Nisim, and Pedahzur 2007; Sullivan et al. 1985). Finally, exposure to terrorism will induce exclusionist political attitudes only through the mediation of psychological distress and perceived threat.

Figure 1.

Stress-Based Model of Political Extremism: Predicting Exclusionist Political Attitudes in the Face of Personal Exposure to Terrorism

The Israeli Context

The Israeli–Palestinian conflict has led to harm to human lives as well as heightened levels of threat perceptions on both sides. Palestinian citizens of Israel, who constitute one-fifth of the Israeli population (Central Bureau of Statistics 2006), are often perceived as a hostile minority with national, religious, and cultural ties to the “enemy,” the arab world; they are often believed to be supportive of subversive activity (Smooha 2004). Although many PCIs support violence against Israel, their direct involvement in such activities has been limited. Nevertheless, their growing numbers increase the perception among many Jewish Israelis that they are threatening the very existence of the state of Israel, as well as its Jewish character (Halperin, Canetti-Nisim and Hirsch-Hoefler 2009; Smooha, 2004).

To a lesser extent, PCIs are perceived as a threat to the democratic values of Israel (Shamir and Sagiv-Schifter 2006) and the potential growth of the PCI population is a source of increasing political dispute (Soffer 1988). They are only part assimilated into Israeli culture and live in separate communities for the most partly. As a consequence, in the minds of many non-Palestinian Israelis, PCIs are seen as a remote out-group (Smooha 2004). With the nature of their inclusion in Israeli society a matter of perpetual debate, they are socially and to some extent formally discriminated against (see Shafir and Peled 2002).

Given these perceived and real tensions, the relations between Jews and PCIs in the face of ongoing terrorism provide a critical test for Israeli democracy. Therefore, Israel provides an excellent setting for the study of the cycle of extremism that leads from individual exposure to terrorism to exclusionist political attitudes.

Method

Sampling and Data Collection

To unveil possible causal relations, data were collected by means of three-wave prospective telephone surveys conducted with a national representative sample of Israeli adults during the Al-Aqsa Intifada. Respondents in the first wave were recruited through a random telephone survey based on stratified samples to ensure representation of both Jews and PCIs, between august 17 and September 8, 2004. Those who agreed and were available were surveyed again approximately six months and one year later (on February 22–April 19 and July 31–October 9, 2005).²

Interviews were conducted by an experienced, computerized survey institute in Israel using trained telephone-survey interviewers.

The response rate among eligible responders was 57 percent. This compared favorably with studies in the United States, especially given that the dialing methods in Israel, unlike the United States, include business phones (approximately 10 percent), which cannot be removed and are treated as a failed attempts, and that the higher rates in U.S. studies typically do not include nonanswered phones (Galea et al. 2002; Stuber et al. 2006). The response rate corrected for business lines was 63 percent.

Wave 1 consisted of 1,613 respondents (1,136 Jews and 477 PCIs). Wave 2 included 840 (609 Jews, 231 PCIs) of the first-wave respondents (52.1 percent reinterview rate). For the third wave, we tried to recontact respondents from the first wave, even if they had not been reached for the second wave. Wave 3 included 716 respondents (573 Jews, 143 PCIs) (44.4 percent reinterview rate). In all, 598 respondents (478 Jews, 120 arabs) participated in all three waves (37.1 percent of the baseline sample).

The main outcome variables tap perceptions toward PCIs, so for the current study we focused on Jewish respondents who participated in all three waves. Nine respondents who were neither Jewish nor PCIs, and who reported a Christian religious affiliation or did not report any religious affiliation, were dropped from the analyses, which left 469 respondents (41.3 percent of the Jewish baseline sample). Studies and reviews suggest that participation rates between 30 percent and 70 percent are, at most, weakly associated with bias, although bias should always be checked by examining the representativeness of the obtained sample (Galea and Tracy 2007).

The sample largely represented the distribution in the adult Jewish Israeli population (Central Bureau of Statistics 2004). One hundred ninety-nine (42.4 percent) of the respondents were men and 270 were women (57.6 percent). Forty-four percent had college education, 22.2 percent had some other post-high school education or were students at the time of the survey, 30.1 percent reported high school as the highest level of educational attainment, and 3.6 percent had less than high school education. Ages ranged from eighteen to eighty-six years with a mean age of 47.36 (SD = 15.4). In terms of religiosity, 67.5 percent defined themselves secular, 20.1 percent considered themselves observant, 7.9 percent considered themselves religious, and 4.5 percent considered themselves very religious. In terms of political stand, 135 respondents (33.5 percent) described themselves as rightists, 151 (37.5 percent) as centrists, and 117 (29 percent) as leftists. We used both ordinary least squares and logistic regression to predict drop-out. Sex and age were consistently predictive of drop-out, and religiosity had borderline effects in some models. However, the significant predictors of drop-out only predicted about 1 percent of the variance for attrition, indicating that attrition was almost entirely random. Central to this article, comparisons of the Jewish respondents who participated in all three waves to those who participated in only one or two waves indicated no significant differences between three-wave responders and one- or two-wave responders in degree of exposure to terrorism or levels of psychological distress (both ps > .30).

Measures

In all three waves, we used an essentially identical, structured thirty-minute interview. To provide support for a cautious suggestion of causality, under the assumption that the antecedent precedes the consequent in time, our model includes measures of recent exposure to terrorism (t1), psychological distress (t1), perceived threat (t2), and exclusionist political attitudes toward PCIs (t2 and t3). We also estimated standard demographics and ideological variables (t1).

Dependent Variable

Exclusionist political attitudes toward PCIs were assessed using a 4-item scale adapted from Scheepers, Gijsberts, and Coenders (2002). Items (e.g., “arabs should not be allowed equal social rights as Jews.”) were answered on a scale ranging from 1 = do not agree at all to 6 = strongly agree, and responses were averaged. The scale has been found to have broad, cross-cultural applicability in fifteen countries (Scheepers, Gijsberts, and Coenders 2002) and was recently translated to Hebrew and validated within the Israeli arena (Canetti-Nisim, Arieli, and Halperin 2008). Cronbach's alphas were adequate, ranging from .74 (t2) to .76 (t3).

Predictor Variables

Recent exposure to terrorism was assessed by a 4-item scale, asking respondents whether they had been exposed directly or indirectly to various terror-related events during the three prior months (answered 0 = no or 1 = yes and summed to a single score, potentially ranging from 0 to 4). Respondents were asked whether (1) they experienced the death of a family member or friend; (2) witnessed a terrorist attack or had been present at a site where there were injuries or fatalities; (3) experienced an injury of oneself, a family member, or a friend; or (4) experienced a period when they did not know if someone close to them was killed or injured, but feared the person might be. Given that one kind of exposure does not necessarily portend another type, internal reliability was not calculated for the four items.

Mediating Variables

PTSD symptoms served as indicators of psychological distress. They were measured using the posttraumatic stress disorder symptom scale, interview format (PSS-I; Foa et al. 1993). The scale contains seventeen items that assess various symptoms based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (American Psychiatric Association 2000) criteria for determining PTSD. Respondents reported symptoms, occurring for at least one month, relating to their exposure to terrorism (which may have been as indirect as television or as direct as being in an attack). Items were answered on a 4-point Likert type scale ranging from 0 = not at all to 4 = extremely. A PTSD symptoms score was calculated by summing the item responses. Cronbach's alpha for the scale in the current study was .87 (t1). As we are not attributing scores to diagnosis, the need for strict adherence to Criterion a (personal exposure to a life-threatening event) is not critical, and some have referred to these scores as trauma-related symptoms (Bleich, Gelkopf, and Solomon 2003).

Perceived threat from PCIs was assessed using a 3-item scale adapted from Sullivan et al. (1985). Items were answered on a scale ranging from 1 = do not agree at all to 6 = strongly agree, and responses were averaged. The items assessed the extent to which the respondent perceived PCIs to endanger the security of Israel, democracy in Israel, and the Jewish character of Israel. With Cronbach's alpha of .86 (t2), internal reliability was very good.

Covariates

Our model controlled for four demographic variables: age (calculated by subtracting reported year of birth from 2004), gender (0 = male, 1 = female), educational attainment (1 = elementary or did not attend school at all, 2 = high school, 3 = post–high school, 4 = college/university student, and 5 = undergraduate education or higher); marital status (single = 1, married =2, divorced =3, widow/er = 4); and income (1 = much lower than the average income in Israel (i.e., $2,200 a month); 2 = slightly below average; 3 = average; 4 = slightly above average; 5 = much above average). Level of religiosity was assessed by a standard self-definition item (1 = secular, 2 = observant, 3 = religious, 4 = very religious).

Also included in the model were two variables representing ideology. Political stand was assessed using the standard self-defined political-stand item (1 = extreme right, to 5 = extreme left). Support for abstract democratic values used three items from Shamir and Sullivan (1985). Items were answered on a scale ranging from 1 = do not agree at all to 6 = strongly agree, and responses were averaged. Sample items are “No matter what a person's political beliefs are, they are entitled to the same legal rights as anyone else” and “I believe in free speech for all, no matter what their views might be.” Cronbach's alpha for the scale was .64 (t1).

Results

Descriptive Statistics

Levels of direct and indirect exposure to terrorism were very high, with 80.2 percent of the respondents experiencing at least one of the four types of exposure assessed, and approximately one-third (32.2 percent) experiencing two or more elements. More than one-fifth (21.1 percent) of respondents experienced death of a family member or a friend as a result of terror attack; 14.1 percent reported witnessing a terrorist attack or being present where there were injuries or fatalities. A large majority of respondents (72.3 percent) experienced a time when they did not know if someone close to them was killed or injured, but feared the person might be.

Not surprisingly, the results show that the wide exposure to terrorism resulted in high levels of psychological distress. For example, 27.2 percent reported at least one reexperiencing item (i.e., repeated, disturbing memories, thoughts, or images of the terrorist attacks and the aftermath of the events).³ A somewhat smaller segment of the sample (21.2 percent) reported at least one type of avoidance behavior (i.e., avoiding activities or situations because they reminded them of the attacks). In addition, 40.2 percent said that they felt hyper alert as a result of the events, and 13.6 percent reported sleeping problems following the attack.

Generally speaking, levels of perceived threat from PCIs were quite high and stable throughout the course of the study. More specifically, paired sample t-tests did not yield any significant differences between the levels of perceived threat in wave 1 (M = 3.37, SD = 1.70), wave 2 (M = 3.25, SD = 1.68), and wave 3 (M = 3.24, SD = 1.67). Levels of exclusionist political attitudes toward PCIs were also high, yet, in that case, the negative attitudes in the third wave (M = 3.50, SD = 1.46) were significantly higher than those measured in the second (M = 3.29, SD = 1.46, t = −4.03, p < .001) and the first (M = 3.39, SD = 1.47, t = −2.14, p < .05) waves. This moderate but significant increase in exclusionist political attitudes emphasizes the importance of discovering the factors that contribute to it, and particularly the role of ongoing terrorism.

Comparison of Respondents Exposed to Terrorism to Those Not Exposed

Table 1 presents descriptive statistics for all the research variables among respondents who reported exposure to at least one of the terrorism items and among those who did not report any terrorism exposure. Interestingly, these two groups differ only with regard to levels of psychological distress. As expected, individuals who had been exposed to terrorism had higher levels of psychological distress than those who had not been exposed. There were no significant differences between individuals exposed to terrorism and those not exposed on any of the control variables or other research variables.

Table 1.

Descriptive Statistics of the Research Variables and Comparison among Those Who Were Exposed to Terrorism and Those Who Were Not

	Entire Sample			Exposed to Terrorism			Not Exposed to Terrorism
Variable	%	M	SD	%	M	SD	%	M	SD	χ 2	t	p
Age	—	47.36	15.43	—	48.88	16.02	—	46.98	15.28	—	1.06	n.s.
Gender
Male	42.4	—	—	40.7	—	—	49.5	—	—	2.35	—	n.s.
Female	57.6	—	—	59.3	—	—	50.5	—	—		—
Education
Elementary	3.6	—	—	3.2	—	—	5.4	—	—	2.60	—	n.s.
High School	30.1	—	—	29.1	—	—	34.4	—	—		—
Some College	22.2	—	—	23.2	—	—	18.3	—	—		—
College Degree	44.0	—	—	44.5	—	—	41.9	—	—		—
Marital Status
Single	18.1	—	—	18.6	—	—	16.1	—	—	2.48	—	n.s.
Married	70.4	—	—	68.9	—	—	76.3	—	—		—
Divorced	7.0	—	—	7.7	—	—	4.3	—	—		—
Widow/er	4.5	—	—	4.8	—	—	3.2	—	—		—
Income Level
Below Average	36.8	—	—	37.4	—	—	34.1	—	—	1.88	—	n.s.
Average	23.7	—	—	24.6	—	—	20.0	—	—		—
Above Average	39.5	—	—	38.0	—	—	45.9	—	—		—
Religiosity
Secular	67.5	—	—	66.1	—	—	72.8	—	—	5.16	—	n.s.
Traditional	20.1	—	—	19.7	—	—	21.7	—	—		—
Religious	7.9	—	—	9.1	—	—	3.3	—	—		—
Very Religious	4.5	—	—	5.1	—	—	2.2	—	—		—
Political left (t1)	—	2.94	0.85	—	2.90	0.86	—	3.09	0.83	—	1.73	n.s.
Democratic values (t1)	—	5.72	0.68	—	5.73	0.66	—	5.68	0.77	—	−0.67	n.s.
Psychological distress (t1)	—	9.97	9.84	—	10.87	9.94	—	6.32	8.55	—	−4.06	<.001
Perceived Threat PCIs (t2)	—	3.37	1.70	—	3.41	1.69	—	3.22	1.75	—	−0.98	n.s.
Exclusionist political attitudes PCIs (t2)	—	3.29	1.46	—	3.27	1.47	—	3.35	1.49	—	0.48	n.s.
Exclusionist political attitudes PCIs (t3)	—	3.50	1.46	—	3.48	1.46	—	3.55	1.46	—	0.36	n.s.

Note: n.s. = not significant.

Bivariate Correlations

As a first step of testing our hypotheses, we examined the bivariate correlations between variables (see Table 2). Our research model assumes a causal chain, in which exposure to terrorism induces psychological distress, and psychological distress and control variables influence the level of perceived threat, which in turn influences exclusionist attitudes. The bivariate correlations among the variables included in the model were generally consistent with our hypotheses. Levels of perceived threat and exclusionist political attitudes were significantly correlated with political stand, religiosity, income, and education level, but not with support for abstract democratic values. Degree of personal exposure to terrorism neither correlated significantly with perceived threat nor with exclusionist political attitudes, but did correlate positively with psychological distress. Moreover, psychological distress (t1) was positively related to perceived threat six months later (t2) and to exclusionist political attitudes a year later (t3). Finally, the correlation between perceived threat (t2) and exclusionist political attitudes (t3)—which are not predictors of the same equation, yet may raise the issue of conceptual equivalence and hence tested—was positive and highly correlated, yet did not cross the multicol-linearity threshold (0.7) (Bagozzi, Youjae, and Phillips 1991).

Table 2.

Bivariate Correlations among the Study Variables

		1	2	3	4	5	6	7	8	9
1.	Exclusionist political attitudes PCIs (wave 3)	—	—	—	—	—	—	—	—	—
2.	Exclusionist political attitudes (wave 2)	.69 ***	—	—	—	—	—	—	—	—
3.	Perceived threat PCIs (wave 2)	.61 ***	.68 ***	—	—	—	—	—	—	—
4.	Psychological distress (wave 1)	.28 ***	.19 ***	.20 ***	—	—	—	—	—	—
5.	Exposure to terrorism (wave 1)	.09	.05	.06	.13 *	—	—	—	—	—
Control Variables
6.	Democratic values (wave 1)	−.07	−.08	−.07	−.05	.03	—	—	—	—
7.	Education	— .17 ***	— .12 **	− .11 **	−.09	.02	.06	—	—	—
8.	Political left	− .58 ***	− .50 ***	− .49 ***	− .16 **	− .18 ***	.14 ***	.09	—	—
9.	Religiosity	.38 ***	.31 ***	.31 ***	.08	.18 ***	− .13 ***	−.04	− .44 ***	—
10.	Income level	— .29 ***	— .26 ***	− .21 ***	− .25 ***	− .12 *	.07	.22 ***	.13 *	− .14 **
11.	Gender (+women)	.11 *	.14 **	.06	.19 ***	.00	.01	.07	.05	.09

^*Significant at the p < .05 level.

^**Significant at the p < .01.

^***Significant at the p < .001 level (two-tailed significance).

Assessment of the Stress-Based Model of Political Extremism

To assess the multimediator model (Preacher and Hayes 2008) we used the aMOS 6 statistical program (Arbuckle 2005) to conduct a Structural equation Modeling (SeM) analysis. SeM enables simultaneous estimation of the relative effects among variable within a given model (Kline 2005). In addition, using various fit indices, it allows comparison of the fit of the hypothesized model to the actual data with the fit of alternative models to the same data. Accordingly, it allows validation of important aspects of the suggested model, such as mediation or direction of causality, by comparing them to their possible alternatives (i.e., inverse causality or direct relations instead of mediation).⁴ These advantages of SeM make it a highly suitable procedure for the assessment of complex models, such as the one proposed in the current work. The combination of a prospective study design with SeM contributes to the validation of the research hypotheses and proposed model.

To affirm the distinctiveness of the scales, we first advanced a measurement model using the variables that were presented in Figure 1. Because of the large number of indicators, to minimize potential pitfalls and to optimize the measurement structure of constructs in SeM procedures, we followed the recommendations of Bandalos (2002) and of Little et al. (2002) and created parcels.⁵ The measurement model consisted of factor-loading paths from the latent constructs (e.g., exposure, psychological distress, perceived threat, and ethnic exclusionism) to their manifest indicators and nondirectional correlations between the latent variables. The measurement model displayed adequate fit to the data (χ²(79, N = 469) = 175.92, p < .00; NFI = .94; IFI = .97; CFI = .97; RMSEA = .05). Correlations between the constructs corresponded with the ones reported in Table 2. Factor loadings on all latent variables were significant and ranged from .71 to .87.

To rule out the likelihood of overlap between perceived threat and exclusionist political attitudes (for a discussion, see Raijman, Semyonov, and Schmidt 2003), we conducted a Confirmatory Factor analysis (see Hu and Bentler 1999). The path between perceived threat and exclusionist political attitudes was first examined without a constraint, then with a constraint of 1 (i.e., making the two variables “identical”). These two models were then tested for χ² difference (see Table 3). The constraint produced a significant change, which indicates that the hypothesized measurement model, in which the two concepts are represented as independent entities, fit the data better than the alternative model. This procedure lends credence to our assumption that perceived threat and exclusionist political attitudes, though related, are two separate concepts.

Table 3.

Fit indices and model comparison

	Comparison of Models: Fit Indices							Model Comparison
Model	N	χ 2	df	CFI	IFI	NFI	RMSea	Δ χ2	Δ df
Measurement Model	469	175.92**	79	.97	.97	.94	.05	—	—
Alternative Measurement	469	200.28**	81	.96	.96	.95	.05	—	—
Model (Threat-Exclusionism constrained to equal)
Structural Model	469	168.46**	81	.97	.97	.95	.05	—	—
Alternative Mode1 1: “Direct Model”	469	575.38**	89	.83	.84	.81	.11	—	—
Alternative Mode1 2: “Integrated Model”	469	157.13**	75	.97	.97	.95	.05	—	—
Measurement Model versus Alternative Measurement Model								24.36**	2
Structural Model versus “Direct Model”								406.92**	8
Structural Model versus “Integrated Model”								11.33	6

^*p < .05.

^**p < .01.

In the next stage, we advanced the structural model, based upon the hypothesized causal relations (see Figure 2).⁶ In more detail, personal exposure to terrorism was assessed as an exogenous variable and exclusionist political attitudes as a dependent endogenous one. Psychological distress and perceived threat were estimated as potential mediators. To assess the net effect of the central components of our model, other variables potentially affecting the endogenous variables were included. All equations control for demographics and ideological beliefs mentioned above, as well as exclusionist political attitudes measured in t2.

Figure 2.

Structural Model (Standardized Regression Weights) for Predicting Exclusionist Political Attitudes in the Face of Personal Exposure to Terrorism Coefficients are Significant at the Level of p < .05

As in the measurement model, all factor loadings in the structural model were significant and ranged from .64 to .90. More importantly, the structural model showed good fit to the data (χ²(81, N = 469) = 168.46, p < .00; NFI = .95; IFI = .97; CFI = .97; RMSEA = .05) and corresponded well with the research hypotheses. As predicted, even when controlling for the long-term aggregated effect of exclusionist political attitudes, perceived threat from PCIs in wave 2 is a pivotal predictor of exclusionist political attitudes six months later (β = .62). Moreover, while some of the control variables (gender, income, and abstract democratic values) had no significant direct or indirect effects on exclusionist political attitudes in wave 3, the effects of other controls (political stand, level of religiosity and level of education) on the outcome variable were fully mediated by the perceived threat and exclusionist political attitudes of wave 2.⁷

Most interestingly, the structural model supports our prediction regarding the chain of factors leading from exposure to exclusionism. On one hand, neither personal exposure to terrorism nor psychological distress directly influenced levels of exclusionist political attitudes in wave 3. On the other hand, the structural model reveals indirect relations between personal exposure to terrorism and exclusionism. Personal exposure is significantly related to psychological distress (β = .12), which in turn has a low but significant effect (β = .11) on perceived threat from PCIs six months later (wave 2).⁸ Eventually, as described earlier, perceived threat in wave 2 strongly influences exclusionism in wave 3. Hence, the findings reaffirm that exposure to terrorism prompts exclusionism for a year or longer following the exposure to the extent that it directly affects psychological distress and indirectly affects perceived threat.

Finally, to rule out plausible alternative interpretations, we compared the hypothesized model with two alternative models. First, we advanced a “direct model” that included merely direct paths by which exposure to terrorism, psychological distress, and the control variables all had direct links to the outcome construct–exclusionist political attitudes. As seen in Table 3, this model yielded poor fit to the data. Second, we advanced an “integrated model,” in which we added the same direct paths to the indirect ones that appeared in the original model. Although the fit measures of this model were similar to the ones of the structural model, none of the added paths in that model was found to be significant. Thus, the statistically tested superiority and general merit of the hypothesized model compared to the rest of the models merit its selection as the best model.

Discussion and Conclusion

The main goal of the current work was to examine a new model for understanding the process that might lead individuals who are exposed to terrorism to support the denial of social and political rights from minority group members. We used a three-wave prospective study conducted in Israel during a period of ongoing, severe terrorist attacks. Analysis of the results by means of structural equation modeling yielded findings consistent with the hypothesized causal chain leading from terrorism exposure to exclusionism through the mediation of psychological distress and perceived threat. These findings suggest that to address the challenge of preserving democratic culture and values in the face of ongoing global terrorism, attention should be given to the psychological distress caused by exposure to terrorism and to the implications of such distress on the degree to which minority groups are subjectively perceived as a threat.

Interestingly, the findings revealed no direct relationship between degree of personal exposure to terrorism and exclusionist attitudes toward PCIs. This indicates that many individuals who were exposed to terrorism did not become more extreme in their exclusionist attitudes toward PCIs. However, the indirect association between exposure and exclusionism suggests an innovative explanation to the pattern of radicalization that may be found among some individuals who are exposed to terrorism. In this manner, the findings suggest that exposure to terrorism will lead to exclusionist political attitudes principally through the multiple mediators of psychological distress and perceived threat. In other words, individuals who are exposed to terrorism may become more exclusionist particularly when they experience psychological distress, which feeds into their perception of threat posed by members of the minority group presumably associated with the source of the psychological distress.

Since psychological distress may be reduced through treatment, political messaging, and public mental health interventions (Hobfoll et al. 2007), the findings regarding its effect on threat perceptions and exclusionist attitudes suggest that psychological interventions, ranging from individual to public health to political, may reduce the negative effects of exposure to terrorism on exclusionism. Contrariwise, this also suggests that political leaders may use fear messaging for political ends (Landau et al. 2004). This further highlights the importance of considering psychological processes when attempting to understand and address the challenges that terrorism poses to the maintenance of democracy.

We developed our stress-based model of political extremism using an interdisciplinary perspective that relied on concepts and theories from the domains of social psychology, clinical psychology, political psychology and political science. The model integrates three important bodies of literature—the literature about the development of political exclusionism and intolerance (Sullivan, Pierson, and Marcus 1982), integrated threat theory (Stephan and Stephan 2001), and the literature on coping with trauma with a particular emphasis on the Shattered Assumptions approach (Janoff-Bulman 1992). The important work of Stephan and Stephan (2001) in the field of social psychology, and of Sullivan and colleagues (1982) in political science, served as theoretical foundations for the emphasis of the pivotal role of perceived threat in the development of negative attitudes toward minorities. The last group of theories explains why exposure to severe events that induce psychological distress leads to perceiving the world in general and the minority group in particular as dangerous or threatening. The present research represents a first attempt to integrate these closely related theories into a comprehensive model.

This model lays the infrastructure for the understanding of the effects of threat perceptions on political extremism in the face of ongoing conflict, a continuous sense of existential insecurity, and growing psychological distress. We are proposing a theory that allows for an analysis of the psychological origins of the desire to limit people's rights and liberties. We argue that personal exposure to political violence that results in psychological distress can affect political worldviews. This happens when people who experience terrorism and the climate surrounding it feel threatened and vulnerable. Such psychological distress exacerbates threat perceptions, which in turn invoke “threat buffers” such as political exclusionism. In other words, people begin to wish to exclude others and perhaps carry out those wishes as a way of lowering their own psychological distress.

In addition to its theoretical innovation, the present study entails some methodological advantages. The first is the above-mentioned use of a prospective design and structural equation modeling, which provide preliminary support to the proposed causal relationships. An additional advantage is the use of a sensitive measure to assess the degree of personal exposure to terrorism. This measure differs significantly from the ones commonly used in previous studies, which pre-assumed exposure of individuals based on geographical proximity or group membership (see Huddy et al. 2002). Finally, in contrast to studies that deal with the effects of onetime terror attacks in the United States (Galea et al. 2002) or Spain (Miguel-Tobal et al. 2006), the current work was conducted in the midst of ongoing attacks on the Israeli public (see also Bleich, Gelkopf, and Solomon 2003). We do note, however, that we did not examine indirect exposure through media, and this might be seen as a limitation of the study.

Other theoretical frameworks might raise alternative interpretations of the presented results. Most notable is the threat-uncertainty model of conservatism (Jost et al. 2003, 2007). This model suggests that conservative ideologies serve the psychological need to reduce uncertainty and threat. Accordingly, it predicts that situational or dispositional factors that increase threat and uncertainty should be associated with increased conservatism. The two core components of conservatism, according to the model, are resistance to change and opposition to equality. Exclusionism may be considered an expression of conservatism, both as a rejection of outsiders who may threaten the status quo, and as a means of maintaining inequality between majority and minority groups. Thus, our finding that perceived threat is the most proximal predictor of exclusionism is consistent with the threat–uncertainty model.

Furthermore, a study based on the threat-uncertainty model of conservatism (Bonanno and Jost 2006) found a positive association between psychological distress following exposure to terrorism and a shift toward conservatism. Although the authors did not examine the mediating role of perceived threat in this relationship, such an interpretation follows from the threat–uncertainty model, which served as their theoretical basis. Our study supports such an interpretation by empirically demonstrating that threat mediated the relationship between distress and exclusionism as an expression of conservatism. The study by Bonanno and Jost (2006) focused on a small, select sample of individuals who had been directly exposed to a terror attack, and therefore could not examine the relationship between degree of exposure (or lack thereof) and psychological distress. Our study makes a further contribution in our use of a general population sample, which allowed us to empirically demonstrate the effect of degree of exposure on psychological distress and through it on perceived threat and exclusionism.

TMT might also be seen as consistent with our findings. Specifically, it could be suggested that exposure to terrorism increases MS, which results in death anxiety that is manifested in psychological distress, which in turn leads to worldview defense in the form of exclusionism. However, as noted above, TMT studies usually observe the effects of MS on worldview defense shortly after the MS-inducing event or experimental manipulation. In contrast, degree of exposure to terrorism in our study reflected accumulated exposure experiences in the three months prior to the first interview, which took place one year before the assessment of exclusionism. Hence, our findings suggest that the effects of exposure to terrorism extend beyond short-term increases in MS.

Furthermore, according to TMT, experiences of anxiety or emotional distress should not mediate the effect of MS on worldview defense, since the worldview functions as a buffer that keeps distress out of consciousness (Pyszczynski, Greenberg, and Solomon 1997). In contrast, we found that exposure to terrorism predicted threat perceptions and exclusionism only to the extent that it was related to self-reported, and hence consciously experienced, psychological distress. Thus, our findings extend the predictions of TMT by demonstrating that exposure to terrorism may induce considerable psychological distress for months and even years after exposure and that such distress mediates the effect of exposure and predicts exclusionism long after the momentary increase in MS would have subsided. Findings from a study by Abdollahi et al. (2008) emphasize the difference between short-term effects of MS manipulations and our findings regarding the long-term effects of exposure to terrorism and the role of psychological distress. Specifically, Abdollahi et al. (2008) found that high levels of PTSD symptoms were associated with less worldview defense following a MS manipulation than lower levels of symptoms, whereas we found that PTSD symptoms were positively associated with exclusionism. This suggests that the long-term effect of exposure to terrorism on exclusionism observed in our study probably reflects a process that is different from the momentary effects of MS manipulations on worldview defense in TMT studies.

The findings showing that psychological distress and threat perceptions mediate the relationship between exposure to terrorism and exclusionism indicate that subjective psychological processes, in addition to situational factors, must be considered when attempting to understand the effects of such powerful events as terror attacks. Appraisal theory of stress (Lazarus and Folkman 1984) suggests that different individuals may experience different levels of stress in response to similar events due to individual differences in the cognitive appraisals of the situation. The complementary theory of Conservation of Resources (Hobfoll 1988) suggests that stress experiences are a function of resource loss and gain resulting from a given event. Accordingly, the level of stress experienced in response to an event may vary depending on the reserves of sustainable resources that are available to different individuals (Hobfoll 2001). Both theories are consistent with the proposition that mere exposure is insufficient to explain the political implications of terrorism, and subjective psychological experiences must be taken into account as well. This idea concurs with the generally accepted approach in social psychology, according to which behaviors are a function of both situational and individual dispositional factors (Ross and Nisbett 1991).

The present work could generate a complementary line of studies that focus on its mirror image—that is, on processes that might lead individuals exposed to terrorist attacks to adopt more moderate ideologies in general and about minority groups in particular. There could be enormous added value in such a line of work. While the insights of the current work can contribute to the understanding and even diminishing of the cycle of extremism, the suggested work could offer ways to delimit and perhaps reverse that cycle while turning resources into more constructive efforts to maintain viable democratic institutions and to end internal and external conflicts.

Taken from a broader perspective, even prior to the recent wave of terrorism, Philip Heymann (1998) argued that one of the most significant threats posed by terrorism is that the interplay between terrorism, public reaction, and governmental response may severely undermine nations' democratic traditions. Nevertheless, the vast writings of political scientists about the challenges that democracies face in the rise of terror have focused mainly on one part of Heymann's equation—i.e., the institutional (non)democratic reactions to terrorism (see, for example, Barber 2003; Weinberg 2007). Our work adds a complementary perspective to this important political discourse by providing an in-depth look into the process by which personal exposure to terror may enhance popular support for nondemocratic regulations and practices. Our premise is that bottom-up processes are likely to translate such support into governmental implementation of nondemocratic policies. We thus strongly believe that the perspective we present in this article is of great importance for those of us who aspire to strengthen democracy in the face of contemporary challenges.