Terrorism and Democracy in Latin America and the Caribbean

Jennifer L. Merolla Claremont Graduate University J. Daniel Montalvo Universidad San Francisco de Quito and Elizabeth J. Zechmeister Vanderbilt University

Working Paper 1/25/12

Abstract. How worried are individuals about violent terrorist attacks? Further, what effects does worry about terrorism have on preferences over democracy itself, its processes, and its core principles? Using survey data from the AmericasBarometer 2010 survey by LAPOP, we first show that concern about terrorist attacks varies greatly across individuals and countries. This variation is predicted by both individual-level and country-level factors, specifically the rule of law and information climate. Second, we provide strong evidence that worry about terrorism is corrosive to democratic attitudes. Specifically, we show that worry about terrorism is negatively related to preferences over democracy, democratic processes, and political tolerance. Political violence is a significant issue for democratic politics in Latin America and the Caribbean. While political violence may have direct effects on democratic processes and policies, we underscore the potential for political violence to indirectly affect democratic politics by elevating anxiety and, consequently, decreasing support for democracy and related core values. A previous version of this paper was presented at the 69th Annual National Conference of the Midwest Political Science Association, March 31-April 3, 2011, Palmer House Hilton, Chicago, IL. Some data collection effort for this project was supported by a National Science Foundation research grant (SES-0851136).

On August 25, 2011, armed men stormed and set on fire a casino in Nuevo Leon, Mexico; 52 civilians were killed in the brutal attack.1 Near midnight on January 16, 2011, a bomb was detonated near a police station in northern Paraguay; a leftist guerilla group, the EPP, claimed responsibility.2 On August 12, 2010, a car bomb exploded outside the studios of a radio station in Bogota, Colombia, injuring at least nine persons; newly inaugurated President Santos condemned the event as an “act of terrorism.”3 While experiences with terrorism across the Americas vary, political violence is a significant issue for democratic politics in the region. Among those concerned about the consequences of political violence in the region, the focus is typically on the extent to which it weakens state capacity, breeds more conflict, or directly disrupts democratic practices and policies (e.g., Wiarda 2010). In contrast, we turn attention to public opinion by assessing the extent and determinants of worry about terrorism across the Americas and the relationship between worry about terrorism and support for democracy, commitment to the democratic rules of the game, and core democratic values. As we will show, concern about terrorist attacks varies greatly across individuals and countries. This variation is predicted by both individual-level and country-level factors; with respect to the latter, context matters: worry about terrorism is higher in countries with a weak rule of law, where past attacks have occurred with greater frequency, and where the media talk more frequently about terrorism. Further, we provide strong evidence that worry about terrorism has a corrosive influence on democratic public opinion. Specifically, we show that worry about terrorism is negatively related to preferences over democracy, democratic processes, and political tolerance. In short, there is reason to be concerned about the extent to which individuals in the Americas are induced to be fearful of terrorist attacks. In the most general sense, terrorism is defined by violent and destructive attacks by nonstate actors for political purposes.4 Terrorism can be committed by domestic or foreign agents

1

http://www.washingtontimes.com/news/2011/aug/29/mexico-offers-reward-in-casinoattack/?utm_source=RSS_Feed&utm_medium=RSS 2 http://news.yahoo.com/s/ap/20110117/ap_on_re_la_am_ca/lt_paraguay_bomb_attacks 3 http://www.businessweek.com/news/2010-08-12/santos-says-colombia-car-bomb-was-terrorist-act-.html 4 On defining terrorism, see Sánchez-Cuenca and de la Calle (2009); their definition recognizes the blurred lines between domestic and international terrorism (on this, see also Crenshaw 2010) and the possibility that terrorists target police and other state actors in such a way that terrorism becomes defined more broadly than including only attacks on civilians; they further note that terrorism is defined by violent actions by “underground” non-state organizations, in order to distinguish between terrorists and armed revolutionary groups occupying territorial bases within a country, but even still there can be overlap between these two concepts, such as when armed guerrillas leave their base to commit violent attacks for political purposes.

1

and for a variety of goals (e.g., to coerce, directly or indirectly, compromise over territory or policy), and across Latin America terrorist acts and plots take many forms. Thus, most concerns in Mexico likely relate to violence perpetrated by narco-terrorists, such as the attack noted above.5 In Chile, worry might center most often on radical anarchist groups.6 In Ecuador and Panama, individuals might be most likely to worry about violence related to extremist groups from Colombia infiltrating these countries’ territories.7 Considered broadly, political terrorism has had a greater presence in some countries in the Americas, such as Colombia and Peru, but terrorist acts have been recorded elsewhere in recent years in countries such as Bolivia, Chile, Mexico, Ecuador, Paraguay, and Venezuela.8 The increased sophistication, scope, and fatalities of modern terrorism leave few corners of the globe immune. Even in countries that have not experienced significant terrorist attacks, citizens may express concern about terrorism out of concern for threats by and conflicts involving various radical groups operating inside and outside the country. Concern about political violence carries significant implications for the extent to which the public supports democratic values and institutions. Merolla and Zechmeister (2009) find that, under conditions in which terrorist threat is elevated, individuals perceive, evaluate, and engage in politics differently (Merolla and Zechmeister 2009). Specifically, conditions in which terrorism is salient have the capacity to increase interpersonal distrust, authoritarian attitudes, preferences for centralization of power in the executive, and willingness to sacrifice civil liberties for stricter security measures. Other scholars have similarly documented a relationship between worry about terrorism and preferences over restricting civic rights (e.g., Davis and Silver 2004; Silver 2007). Another set of findings in the literature relates terrorism to greater preferences for the political right (Berrebi and Klor 2006) and conservative values (Bonanno and 5

See also http://www.proceso.com.mx/rv/modHome/detalleExclusiva/81475 See http://laht.com/article.asp?ArticleId=346791&CategoryId=14094 7 Though scant survey data exists asking about the types of terrorism that worry individuals, data from Panama is illuminating. In Panama, participants in the AmericasBarometer survey in 2010 were asked about their level of worry about attacks on Panamanian soil by the FARC and Islamic fundamentalists, respectively. On a one to seven scale, where higher values indicate more worry, the mean levels of reported worry were 4.3 for the former and 3.9 for the latter. These questions were included on the Panama survey (but not asked in other questions included in the 2010 AmericasBarometer): PANWT4: ¿Qué tan preocupado está usted que la guerrilla colombiana (FARC) lleve acabo un ataque terrorista en Panamá en los proximos 12 meses? PANWT5: ¿Qué tan preocupado está usted que grupos fundamentalistas islamicos lleven acabo un ataque terrorista en Panamá en los próximos 12 meses? The two variables are correlated at 0.78; non-response is less than three percent for both questions. 8 In addition to news sources such as those cited here, on terrorist incidents see the RAND Database of Worldwide Terrorism Incidents (RDWTI), http://www.rand.org/nsrd/projects/terrorism-incidents/ 6

2

Jost 2006; Echebarria-Echabe and Fernández-Guede 2006). But, what effects do worry about terrorism have on preferences over democracy itself, its processes, and its core principles? Is there evidence that concern about terrorism can cause individuals to turn their backs on this broader set of institutions and values? Using data from the 2010 round of the AmericasBarometer survey by the Latin American Public Opinion Project (LAPOP) 9 and hierarchical linear analysis, we will show the answer to this question is “yes,” at least for Latin America and the Caribbean.

Concerns about Terrorism in the Americas We begin by documenting the fact that worry about the possibility of violent terrorist attacks exists and varies across individuals and countries in Latin America and the Caribbean. To assess this subject, we consider responses to a question from the 2010 AmericasBarometer; the question was worded as follows: “(WT1) How worried are you that there will be a violent attack by terrorists in [country] in the next 12 months?”   Respondents answered on a four-point scale from not at all worried to very worried. The question was designed to allow for two important aspects of public opinion on terrorism in the Americas. First, it did not impose a definition of terrorism on respondents in order to capture general concerns about violent attacks by rogue actors, without restricting the individual to only consider certain types of attacks; in other words, the question therefore implicitly recognizes and allows for the varied conceptions of terrorism in the Americas. Second, recognizing that not everyone might have concerns about terrorist attacks, the question allowed an explicit “haven’t thought much about this” option. The percentage of individuals who responded that they “haven’t thought much about this” is 22.99 percent. In the analyses presented in the body of this paper, we omit this group of individuals. However, one interpretation of such a response is that the individual is not worried and, therefore, as a robustness check we have run all the analyses with these individuals coded as not worried and the results are consistent with what we present here. As another robustness check, we do not make any assumptions about these individuals, and run the analyses in the paper treating responses to the question as categorical (“haven’t thought much about this” is included 9

We acknowledge and thank LAPOP and its donors for their important work in making these data available. Funding for the 2010 round mainly came from the United States Agency for International Development (USAID). Important sources of support were also the Inter-American Development Bank (IADB), the United Nations Development Program (UNDP), and Vanderbilt University.

3

as a dummy variable along with dummy variables for the values 1-4 on the scale); conclusions from these results, as well, are consistent with what we report here. We present these alternative set of analyses in appendix material available from the authors on request (see also footnote 21).10 INSERT FIGURE 1 HERE Among those who responded to the terrorism survey question, Figure 1 shows mean responses (with confidence intervals) by country, with those responses recalibrated for the sake of easy comparison from the original 1-4 scale to a 0-100 scale, where 0 means “Not at all worried” and 100 “Very worried”.11 Mean levels of worry about terrorist attacks, as shown in Figure 1, are above the 50-unit mid-point on the scale in ten of the 24 countries. Not surprisingly, Colombia and Peru, two countries that have suffered from numerous terrorist attacks, are found in the top tier of the list of countries, with mean levels of worry of 67.3 and 63.0, respectively which is closest to the category of “worried”. The RAND Database of Worldwide Terrorism Incidents (RDWTI; see footnote 8) reports 46 attempted and/or successful terrorist attacks in Colombia in 2009, most of which are attributed to the FARC. Fewer incidents were reported in Peru, but the country has nonetheless continued to see occasional attacks, typically attributed to the Shining Path, which serve as reminders of much greater levels of terrorism in the 1990s. Interestingly, Ecuador and Paraguay are at the top of the list, suggesting high levels of concern about security in those countries. In Ecuador, this high value may reflect concerns about FARC activity within Ecuador’s borders as well as public anxiety over increasing numbers of bold, public, and often lethal attacks by criminal elements (e.g., sicarios, or hitmen).12 In

10

We did not consider imputing other values for worry about terrorism, which might be appropriate if we thought individuals were systematically censoring or otherwise failing to reveal a quantifiable concern about terrorism. Instead, we believe that those who respond “haven’t thought much about this” are, overall, sincere. Therefore, the available options for how to treat them are the following: leave them aside and study those who have thought about this; treat the response as a separate category; or, alternatively, code them as not worried because thinking about terrorism is a logical antecedent to worrying about terrorism. As noted, we report on the first approach here in the text and the other approaches in appendix material; our findings are robust to all three approaches. 11 The transformation is linear to the 0-100 variable; later we conduct a similar linear transformation of the variable, and others, to a 0-1 scale for the regression analyses. Mean non-response for this question was 5.3%; these individuals are excluded from the analyses in this paper. If we code those who “haven’t thought much about this” as “not worried”, the relative ranking of mean values across the countries remains fairly stable (see Zechmeister, Montalvo, and Merolla 2010). 12 In Ecuador, our personal experience in the country also leads us to note that there is also a tendency for at least some politicians to use the term in passionate rhetoric against political rivals.

4

Paraguay’s case, this may be due to concerns about terrorist cells in the tri-border area13 and could also be due to concerns about domestic guerrilla groups such as the EPP (see the incident mentioned in the paper’s introduction). In direct contrast to these cases, levels of worry are strikingly low in Jamaica, Argentina, and especially Uruguay (whose mean ranking is closest to the category “somewhat worried”). The United States’ placement in the middle of the scale makes sense in light of the fact that economic decline rivaled for the public’s attention in 2010, while the public’s concern about terrorism likely had been dulled by nearly nine years of repeated terror alerts typically followed by little to no sign of terrorist activity. What determines variation in levels of worry about terrorism in the Americas? The above analysis and discussion suggest that country context matters, and it is possible that individual characteristics do as well. In the next section we take a closer look at determinants of worry about terrorism in Latin America and the Caribbean.

Security and Information Climates Affect Worry about Terrorism In predicting worry about terrorism, we consider factors at both the national and individual levels. In the first place, we expect that factors related to the security and information climate in a country will influence levels of worry about terrorism. One factor that should matter is the rule of law. The stronger the rule of law, the more protected citizens should feel from violence by terrorists or others. A weak rule of law could create a climate that foments extremist views and behaviors and/or might act as a proxy for state failures in the realm of the provision of security (e.g., Choi 2010). Second, the extent to which terrorist attacks have occurred in recent years and/or the extent to which terrorism is discussed in public domains, for example by the media, should correlate with concerns about terrorism; such a relationship has been detected in studies focused on 9/11 and its aftermath in the United States (Amsel et al. 2007; Bloch-Elkon, Nacos, and Shapiro 2007; Nacos, Bloch-Elkon, and Shapiro 2011). In considering rule of law and media discussion of terrorism, in neither case are we suggesting a clear-cut causal story. It could very well be that the presence of security concerns influences a country’s rule of law ratings. At the same time, a weak rule of law may undermine individuals’ sense of security, leaving them feeling vulnerable to hostile actors while also creating a more permissive environment for terrorist activity. In the second case, discussion of terrorism in the media may 13

See, for example, the report found at http://www.loc.gov/rr/frd/pdf-files/TerrOrgCrime_TBA.pdf.

5

reflect, but could also fuel, concerns about terrorism among the public. Our use of the third measure, experience with past attacks in the country, helps us achieve at least some greater leverage with respect to causal inferences, and we return to this discussion later. In addition to country context, characteristics of individuals and the neighborhoods where they live should help explain variation in levels of worry about terrorist attacks. There is general agreement in the literature that women tend to perceive a greater risk from and report greater levels of concern about violence in general, and this includes terrorism; this notion receives empirical support from multiple studies across contexts that include Canada, Germany, Israel, and the United States (e.g., Bruck and Muller 2009; Friedland and Merari 1985; Huddy et al. 2005; Klar, Sharvit, and Zakay 2002; Krewski, Lee, Lemyre, and Turner 2006; Nacos, BlochElkon, and Shapiro 2011; Nellis 2009). To the extent that these same researchers examine the relationship between education and worry about terrorism, the results are mixed with, for example, Krewski et al. (2006) finding a positive correlation, Friedland and Merari (1985) finding a negative correlation, and Nacos et al. (2011) finding a changing pattern over time. Findings on age are likewise mixed so that, for example, Klar, Sharvit, and Zakay (2002) find that older individuals are less concerned about terrorism, while Krewski et al. (2006) find the opposite.14 We assess the extent to which such traits correlate with worry about terrorism, but we also expand our analysis of individual level factors in a number of ways. First, we expect worry about crime (perception of neighborhood insecurity) to correlate with worry about terrorist attacks, as the latter may be a plausible subset of the former in some cases. For example, respondents in a country that is plagued by narco-terrorists may interpret activities by these groups as being both crime and terrorism. We also expect that those who have been previously victimized by crime, or who have a household member who was victimized, will be more concerned about the possibility of a violent attack by terrorists. In terms of political ideology, we expect that those on the right will be more worried about violent acts, in part because Latin American guerrillas are usually considered to the left15 and in part because it is consistent with a conservative ideology, broadly speaking, to exhibit greater concerns about law and order; for example, Klar, Sharvit, and Zakay (2002) find that Israelis who voted for the 14

In the U.S. context, Huddy et al (2005) show a relationship between knowing someone who was killed or hurt in the 9/11 attacks and worry about terrorism, as well as one between living near the terrorist attacks and worry. We do not have the ability to test these relationships with the LAPOP’s AmericasBarometer 2010 data. 15 Examples of guerrilla groups considered on the left include the FARC, ELN, Revolutionary Movement TupacAmaru, Shining Path, Tupamaros, and Alfaro Vive Carajo.

6

political right in a recent election were more concerned, at a personal level, about the danger posed by terrorism.16 While we emphasize that the nature of worry about terrorism in Latin America and the Caribbean is likely comprised principally of domestic concerns, it is nonetheless probable that those with relatives abroad will have heighted perceptions of the risk posed by terrorism, given the prevalence of discussions about international terrorism as well as terrorist incidents in both the United States and Europe, where many members of the Latin American and Caribbean diasporas reside. Finally, we expect that individuals residing in smaller cities and rural areas will be more concerned about the likelihood of terrorist attacks. The empirical evidence shows that in countries such as Colombia, Paraguay and Peru, violent groups concentrate in relatively less populated towns where they can plan irregular activities in a clandestine fashion. We are not suggesting however, that the risk of an actual attack is higher in smaller towns and rural areas; what we mean is that people residing in these territories may be more aware of the presence and threats posed by terrorists in a particular nation. In analyses that include measures capturing the above individual-level factors and basic socio-economic and demographic control variables, we include measures of rule of law, experience with past attacks, and the information environment, respectively, as second-level indicators. In the regression analyses, all individual level measures have been re-scaled to run from 0 to 1.17 Our Rule of Law measure is based on data from the Freedom House organization, which provides a measure of the rule of law that is intended to reflect the following questions: (1) Is there an independent judiciary? (2) Does the rule of law prevail in civil and criminal matters? Are police under direct civilian control? (3) Is there protection from political terror, 16

In the U.S. context, Huddy et al (2005) do not find a significant effect for ideology, though they find that authoritarian individuals are more worried about terrorism. The surveys we are using do not have a measure of authoritarianism, so we can only explore the effects of ideology. 17 In addition to the size of town variable [“Size of the City/Town of Residence,” which on the original scale runs from 1 (Rural Area) to 5 (National Capital)], the socio-economic and demographic variables we include are as follows; (a) “Quintiles of Wealth,” which runs from 1 (Least Wealth) to 5 (Most Wealth); (b) “Age,” which on the original scale runs from 18 (16 in the case of Nicaragua) to 98 years of age; (c) “Female,” coded 1 for women and 0 for men; and (d) “Education,” which on the original scale runs from 0 to 18 years of completed education. With respect to the other individual level measures, we include a four-point measure of crime insecurity coded so that higher values indicate more insecurity, as well as dummy variables for whether the individual has relatives living abroad, has been a victim of crime, has a household member who was a victim of crime, identifies on the right ideologically, the center ideologically, or expresses no ideology relative to a baseline of having a left ideology. In addition, as a robustness check on our results, we also included in analyses not shown here a personality measure that reflects the extent to which an individual considers himself an anxious and easily upset person; the variable is significant, but including it requires that we drop Honduras (where the question was not asked) from the analyses. Excluding the variable has no significant effect on our other results and therefore we show analyses here that exclude it, in order to preserve as many cases as possible.

7

unjustified imprisonment, exile, or torture, whether by groups that support or oppose the system? Is there freedom from war and insurgencies? and (4) Do laws, policies, and practices guarantee equal treatment of various segments of the population?18 The variable (prior to the linear transformation to a 0-1 scale) ranges from 5 (in Guatemala and Venezuela) to 15 (in Chile and Uruguay), has a mean of 8.54 across the 22 of countries for which we have data.19 To capture the Information Climate, we created a variable based on a content analysis of major newspapers for as many countries as we were able to search online; given unavailability of data for some countries, our number of country cases is reduced in this analysis to 19.20 The content analysis was conducted by a research assistant who followed precise instructions to create counts of the number of times an article with the term “terrorism” or “terrorist”, or both, appeared in a major national newspaper over a twelve month period surrounding the dates in which the survey data were gathered. The variable (prior to the linear transformation to a 0-1 scale) ranges from 5 (in Belize) to 3817 (in Colombia), and has a mean of 706 articles across the 19 countries for which we have data.21 To capture past attacks (Number of Past Attacks), we use a count of the number of attacks over the last decade, collected from the Global Terrorism Database, which is an open-source database of terrorist events around the globe from 1970 through 2010, housed at the University of Maryland. The database includes domestic, transnational, and international terrorist incidents. The variable ranges from 0 events (in Belize, Costa Rica, Dominican Republic, El Salvador, Nicaragua, Panama, and Surinam) to 712 (in Colombia), and has a mean of 38.7 across the countries in the sample. As with the other variables, for the regression analyses this measure was re-scaled to range from 0 to 1. 18

See: http://www.freedomhouse.org/template.cfm?page=351&ana_page=341&year=2008 Last accessed on 07-2010. 19 These countries are Argentina, Belize, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Guyana, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. 20 Countries for which we were not able to find data are Guyana, Haiti and Panama. 21 We thank Angela Romano for her research assistance on this project. We also note that there are several potential problems with the search. For instance, the count changes from day to day as websites update their search engines. Additionally, some websites did not allow for date parameters to be entered into the search. Thus, some websites only show the past year, others only show a certain number of results, and others do not separate by date. For the websites that contained date parameters or could be separated by year, an exact count was provided; this represents most cases. For several others, we include the number returned by the search (e.g., 1000 in Peru, 500 in El Salvador), even though this represents the cap and the number of actual returns would theoretically be higher if the website did not limit the number of results. More information on the counts is provided in appendix material available from the authors.

8

In analyses aimed at examining the relationships among these country-level variables, we find that the information climate variable and past attacks are highly correlated: where past attacks have occurred with greater frequency in recent times, the media discusses terrorism more often (correlation: 0.89, between our attacks in the 2000s measure and the information climate measure). Therefore, we run two separate models, both of which include the rule of law measure and one of the two other measures. INSERT TABLE 1 HERE Our results are presented in Table 1. Given that our data involve both national and individual level measures, we use statistical techniques that are hierarchical.22 As noted, we fit two multilevel models in order to test the effects of rule of law and information climate on worry about terrorism. Across both models, we include the rule of law variable; the difference in the models is that the first includes our measure of information climate as frequency of media discussion of terrorism and the second includes our measure of past attacks. Across both models, as expected, the results show that those who live in smaller towns and/or more rural areas are more worried about terrorism, as are those who are poorer and less educated. It is possible that this reflects a reality of terrorism: the typical terrorist attack in the Americas inflicts more harm on groups of individuals who are marginalized in such ways; at the same time, it may be that individuals who are marginalized in these ways are simply more likely to express concern about terrorism. We see also that women are far more concerned than men and, as we expected, those with relatives abroad are more worried about terrorism. We suggest this latter result could reflect a transmission of information from places abroad. As expected, individuals who have been the victim of crime, have household members who have been the victim of crime, have higher levels of insecurity, and have an ideology to the right and center of the scale (relative to the left, which is the baseline category in the analysis) are relatively more worried about terrorism. Turning to our first country-level indicator, a country’s level of rule of law is indeed associated with worry about terrorism; this result is consistent across both models. Citizens in countries with lower rankings on the Freedom House Rule of Law scale are more worried about 22

We model these using the XTMIXED command in Stata. This approach helps to avoid two of the most common problems that arise from failing to account for the hierarchical nature of the data: ecological fallacy and atomistic fallacy, where “relationships observed in groups are assumed to hold for individuals” in the former, and “inferences about groups are incorrectly drawn from individual-level information” in the latter (Luke 2004).

9

terrorism than citizens in countries where the rule of law is more secured. Because all of the variables in the analyses are scaled to run from 0-1 and because the method we employ treats the dependent variable as linear and continuous, the predicted substantive effects are easy to discern and compare. The maximum effect of a variable (a change across the range of the variable from its minimum value of 0 to its maximum value of 1) is captured by the regression coefficient. Thus, as Table 1 shows, a person living in a country with the lowest rule of law in the region is 0.21 units more worried about violent terrorist attacks, when compared to a person living in a country with the highest rule of law in the region, all else equal. Turning to the Information Climate variable, we indeed find evidence that the rhetorical environment (as captured by newspaper article counts) is positively related to worry about terrorism. The p-value on the coefficient is 0.113, and we consider this support for our expectation given that we have a very low number of observations (and a uni-directional expectation, which if followed strictly would lead us to evaluate the p-value as 0.05 one-tailed). Substantively, and all else equal, the predicted average level of worry about terrorism is 0.18 points higher on the 0-1 scale in the country for which we recorded the highest value for the information climate variable, Colombia, in comparison to Belize, where we recorded the lowest value for information climate. Finally, when we consider the measure of the Number of Past Attacks we find a maximum substantive effect of 0.19 points. In short, individual level factors matter in predicting worry about terrorism but so, as well, does context. In fact, Table 1 shows that the strongest maximum substantive effects in the model are exerted by the contextual variables. Individuals living in countries with lower levels of rule of law and with a higher number of terrorist events, and greater discussion of terrorism in the media exhibit, all else equal, greater concerns about the likelihood of a terrorist attack in their country over the next twelve months.23 As we have indicated, we do not think that the causal story behind all of these results is a simple one. It is theoretically conceivable that low levels of rule of law induce worry about security in citizens while they also reflect a society in which worry is elevated as a result of experiences with weaknesses in the rule of law. Likewise, the 23

As noted in an earlier footnote, we also ran all the analyses treating each response option as a separate category and find that as a country’s rule of law improves, the odds of being in the “Very worried,” “Somewhat worried,” “A little worried,” and “Haven’t thought much about this” categories decrease relative to being in the “Not at all worried” category. Additionally as the number newspaper articles related to terrorism increases, the odds of being in the “Very worried,” “Somewhat worried,” “A little worried,” categories increase relative to being in the “Not at all worried” category. However, there is no statistical difference in the odds of being in the “Haven’t thought much about this” and “Not at all worried” categories as the number of newspaper articles related to terrorism increases.

10

media environment may stoke concerns about terrorism but may also reflect such concerns.24 At the same time, our assessment of a variable that temporarily precedes the dependent variable and is arguably exogenous to it – frequency of past attacks – provides additional reason to assume that at least some of the causal power runs from information climate to worry about terrorism, and not solely in the reverse fashion. What consequences do these concerns have for democratic public opinion? We address this question next.

Worry about Terrorism Erodes Support for Democracy and Political Tolerance Our basic contention is simple: fear of terrorism affects the ways people think about others and about government. A chief objective of terrorism is to induce anxiety and fear, and extant research shows clearly that terror threat significantly affects political attitudes, evaluations, and behaviors in ways that may place stress on democratic values, processes, and even institutions (Davis 2007; Huddy et al. 2005; Merolla and Zechmeister, 2009). While much of this research has been limited to a focus on specific values and preferences, we argue that worry about terrorism also carries the potential to undermine individuals’ support for democracy and democratic practices more generally, at least within certain country contexts. In this section we present an argument supporting the notion that worry about terrorism can lead to decreased support for democracy, lower support for democratic rules of the game, and, as well, diminished levels of political tolerance. We then test related expectations, once again using data from the 2010 AmericasBarometer survey. Conditions of terror threat heighten individual anxiety and concern about future threats (Huddy, Feldman, Taber, and Lahav 2005; Merolla and Zechmeister 2009). This type of collective threat highlights individuals’ inability to control external circumstances that threaten their financial, psychological, and/or physical well-being. Merolla and Zechmeister (2009) developed the argument that, in attempting to re-establish feelings of control and safety, individuals may adopt one or more of several coping strategies, which in turn affect support for democratic values and institutions. First, with respect to relationships with other individuals, those in a condition of terrorist threat may become less trusting of others, more intolerant of out24

In fact, we carried out two “Hausman Specification Tests” in order to test whether or not there is evidence of simultaneity between the variables worry about terrorism and rule of law on the one hand, and worry about terrorism and information climate on the other. We find confirmatory statistical evidence for simultaneity in both cases.

11

groups, and more supportive of authority, relative to those who are not experiencing a condition of threat. In other words, in order to feel more secure, individuals may become less supportive of democratic values meant to protect civil liberties, at least in the short term, and especially with respect to the civil liberties of out-groups. A second coping strategy involves choosing to turn over control to an external actor, such as a political figure whom one deems capable of solving or handling the crisis. In presidential systems at least, citizens will most often look to the office of the executive to handle the given crisis context. Individuals may come to prefer that the balance of power tilt decidedly in favor of a stronger executive, one who can then pursue a quick, unhindered resolution to the crisis. The expression of this coping strategy thus carries with it some threat to democracy, as individuals prioritize and enable strong executive leadership, potentially at the expense of horizontal accountability (checks and balances). In extreme cases, the desire to empower the leader with the tools to solve the crisis may lead individuals to reject democracy as a regime type. Existing scholarship has found support for a negative relationship between terror threat and support for some democratic values in practice. Using survey and experimental data, scholars have found that citizens who are more worried about terrorism are more willing to trade civil liberties for more security (e.g., Davis and Silver 2004; Davis 2007; Huddy et al. 2005; Merolla and Zechmeister 2009). Bonanno and Jost (2006) documented a general “conservative shift” among a small sample of “high-exposure survivors of the 9/11 terrorist attacks.” Others have demonstrated a relationship between those with high levels of worry about future attacks and willingness to increase surveillance on Arabs and Arab-Americans, increase security checks on Arab visitors, and decrease visas to Arab countries (e.g., Huddy et al. 2005).25 With respect to general measures of political tolerance outside of the U.S., Merolla and Zechmeister (2009) found that subjects exposed to a terror threat condition in Mexico were less willing to provide various rights to their least liked group.26 Yet, while there is some initial support for the expectation that terror threat will lead to a decline in support for democratic values in practice, 25

However, using data from the Pew Center, Kalkan et al (2010) do not find a relationship between concern about terrorism and feelings toward Muslims. In the context of Europe, Das et al. (2009) find that those exposed to terrorism news held more prejudiced attitudes towards Arabs and this was mediated by death-related thoughts. 26 Empirical studies have demonstrated a relationship between other types of external threats, such as globalization and other security crises, and political intolerance in both established and more nascent democracies (Marquart-Pyatt and Paxton 2006; Shamir and Sullivan 1983; Wang and Chang 2006; Sullivan et al. 1993). Others also look at the effects of the perceived threat of a disliked group on levels of political intolerance (Duch and Gibson 1992; McClosky and Brill 1983; Sullivan, Marcus, Feldman and Pierson 1981).

12

only a handful of measures of democratic values have been employed, and scholars have not looked much at whether support for democracy itself and the basic democratic rules of the game shift alongside changes in specific attitudes and values.27 There is some empirical evidence suggesting that, in order to resolve the terror threat, individuals may become more supportive of a stronger executive relative to the legislature. In the U.S. context, in the aftermath of 9/11, the U.S. public became highly supportive of the administration in general and with respect to particular policies such as a unilateral foreign policy approach (Huddy et al. 2005). Furthermore, a majority of Americans supported Bush’s “terrorist surveillance” program, which entailed wiretapping without approval from the FISA court.28 In an experiment that Merolla and Zechmeister (2009) conducted in Mexico in 2006, participants who read a terrorism article expressed preferences for a stronger executive, compared to those who read an article highlighting all of the advances in education, health, and the environment in that country. At the same time, though, Merolla and Zechmeister (2009) did not find a similar effect in the U.S. in a 2007 study; thus, it is not always the case that terror threat will compel citizens to be willing to cede extra institutional power to the executive.29 Considering more anecdotal evidence from Latin America, in the context of both terrorist and economic threats, in the early 1990s Peruvian President Alberto Fujimori temporarily shut down congress and other governmental institutions and, subsequent to these actions, was re-elected to the presidency.30 Finally, another set of scholarship supporting a link between terrorism and democratic values focuses on the links between crime and insecurity, on the one hand, and democratic public opinion, on the other hand; a number of scholars have found links between individuals’ assessments of their safety vis-à-vis crime and their opinions about democratic values, processes, and institutions (e.g., see Fernandez and Kuenzi 2010; Pérez 2003, 2009; Maldonado 2010; Malone 2010; Seligson and Smith 2010; but also see Ceobano, Wood, and Ribeiro 2010).

27

On terror threat and authoritarianism, see also Hetherington and Weiler (2009). FOX News/Opinion Dynamics Poll. 29 We suspect this particular null finding for the U.S. was due to the timing of the survey (2007), when approval for Bush and his policies was low. 30 It is worthwhile noting that the public has not consistently nor universally supported these maneuvers; Fujimori, most obviously, is currently not in office but in jail. Just as in the U.S. case, the issue of duration/timing seems to be key (though outside of the current scope of our study), though other factors likely matter as well (e.g., the particular corruption scandal that led to Fujimori’s quick political downfall). 28

13

Can conditions of terrorist threat undermine support for democratic institutions and practices? To test the extent to which worry about terrorism may erode support for democracy, democratic processes, and core principles, we first identified measures reflecting these attitudes within the AmericasBarometer survey. We consider four sets of measures: support for democracy in the abstract; support for strong, unencumbered leadership; support for military interruptions to the democratic rules of the game; and the basic democratic values of support for public contestation and political tolerance. In the next paragraph we describe how we measure the items in each of these four baskets.31 First, we measure support for democracy using two questions. The Churchillean Agreement item, so-called because of its association with a famous dictum by that leader, asks individuals the extent to which they agree that democracy, despite its flaws, is the best possible political system. The Support for Democracy item asks individuals to indicate whether a democratic system is always preferred, under some cases an authoritarian system can be necessary, or whether it doesn’t matter. Second, we tap a related set of attitudes, concerning support for a strong leader free of potentially cumbersome democratic processes and institutions, with three measures. The Anti-Iron Fist item reflects whether an individual agrees that democracy works best with everyone’s participation or if the country needs a government with an iron fist. The Anti-Strong Leader item asks whether electoral democracy is always best or whether a strong leader might be needed who does not have to be elected. The Pro-Parties item asks whether democracy can, or cannot, exist without political parties. Third, we consider the extent to which individuals are broadly committed to the democratic rules of the game with a series of three items that ask whether a military coup can be justified in cases of high unemployment, high crime, and high corruption, respectively. In analyses not reported here, we found similar results across each of these three separate measures and we further find that the items scale together well (the Cronbach’s alpha for these items is 0.78). Therefore, we combined them here into a single measure, Anti-Military Coups. Finally, we assess two general democratic values: support for Public Contestation and Political Tolerance. The former is based on three items that measure the degree of citizen approval of actions that people can take to achieve their political goals and objectives, such as citizen participation in legal demonstrations, participation 31

The exact wording of these and all survey items reported on in this paper can be found in the AmericasBarometer questionnaires, found online at www.LapopSurveys.org; they can also be found in appendix material available upon request from the authors.

14

in organizations to try to solve community problems, and working on electoral campaigns for a political party or candidate (the Cronbach’s alpha is 0.73). The latter is based on a series of items that ask survey participants to indicate their approval or disapproval of the rights of regime critics to lawfully participate in politics through the act of voting, peaceful demonstrations, running for office, and making speeches (the Cronbach’s alpha for these four items is 0.85). All of the measures are coded such that higher values reflect pro democracy responses. In order to isolate the effects of worry about terrorism, we include in our analyses the following control variables: ideological32 self-placement dummy variables (with left as the baseline); corruption victimization; crime victimization; evaluation of the country’s economic situation; evaluation of one’s personal economic situation; household wealth; gender (female=1); age; education level; size of town of residence; and, country dummy variables. All variables in the analyses are scaled to run from 0 to 1. The results are presented in Table 2. The table also notes the method of regression analysis, which varies by dependent variable.

INSERT TABLE 2 HERE

Across all dependent variables, we find significant effects for worry about terrorism, and in the expected direction. In other words, as fears of terrorism increase, individuals tend to disagree to a greater extent with the idea that democracy is the best system of government. Citizens are also more likely to support the idea that an authoritarian system governed by strong, iron fisted leaders could be necessary when they feel threatened by terrorist activities. In this context of heightened worry about terrorism, individuals are even more willing to support military coups under certain conditions. Support for democratic values such as public contestation and political tolerance also decline when individuals are worried about terrorism.33

INSERT TABLE 4 HERE

32

As in prior analyses in this paper using this variable, the measure is the result of combining left-right (asked in most countries) and liberal-conservative (asked in a few Caribbean countries in place of left-right) self-placement questions. 33 This latter result is consistent with the finding reported by Edwards et al. (2011), who show that worry about terrorism in the Americas predicts a lack of support for allowing regime critics the right to vote; that variable comprises one part of the political tolerance index we examine here.

15

In Table 3, we present evidence of the substantive significance of these relationships. For those dependent variables that we treat as continuous and for which we use OLS analysis, the predicted impact of moving from a minimum to maximum value on worry about terrorism can also be interpreted directly from the coefficients in Table 2. For the multinomial probit and probit analyses, we present the predicted change in the probability of falling into a given category when the worry about terrorism measure is moved from its minimum to maximum value. Considering the continuous measures, worry about terrorism has the strongest substantive relationship with respect to the Anti-Military Coups measure. That is to say that, on average, individuals who are very worried about possible attacks in the next 12 months are 0.11 units more in favor of military intervention in bad times compared to individuals who are not at all worried; that is, the predicted maximum effect moves an individual a distance of just over onetenth of the scale on which the dependent variable is measured. In the models presented in Table 2 (and as depicted in Table 3), worry about terrorism exerts a moderate substantive effect on the Churchillean Agreement measure, as well as the Pro-Parties and Political Tolerance measures. The weakest relationship for those dependent variables analyzed with OLS is found for the Support for Public Contestation variable; while statistically significant, individuals are predicted to shift by only -0.029 units when worry about terrorism shifts from its minimum to maximum value, all else equal. To assess substantive significance with respect to the multinomial probit analysis (for the Support for Democracy measure), we set all variables to the mean and the change worry about terrorism measure from its minimum to maximum value. As Table 3 reports, moving from an individual who has no worry about terrorism to one who has maximum worry decreases the probability of thinking that democracy is preferable (relative to the thinking that an authoritarian system could be necessary) by 5.2 percentage points. For the logit analyses, a comparable shift in worry about terrorism from the minimum to the maximum leads to a 6.9 percentage point drop in the probability of giving the democratic response of anti-iron fist and a 5.8 percentage point decrease in the probability of giving the anti-strong leader response. In general terms, these findings corroborate our expectations that concerns about potential terrorist plots clearly put democratic values at greater risk of decay. Moreover, the fact that the effects for terrorism are found even after controlling for variables related to crime and corruption victimization gives us confidence that we are isolating 16

the effects of worry about terrorism on democratic public opinion. Both of these measures tend to decrease support for democratic institutions and democratic values.34 With respect to the other independent variables, we find that people who place themselves in the right and center category of our ideology measure with respect to those self-identified at the left are generally more supportive of democracy in the abstract and democratic values, though they are less inclined to be opposed to using an iron fist and a military coup under some circumstances. Individuals with higher levels of household wealth and education are generally more supportive of democracy, the rules of the game and democratic values. Assessments of the country’s economic situation, size of town, gender and age have mixed effects across the models. The substantive effects of these other variables are either similar to or less than the effects of terrorism, with the exception of the education measure. While the effects of worry about terrorism on support for the rules of the game and democratic values were modest, the effects of education are stronger. For example, moving from not having an education to having completed at least 18 years of education increases opposition to military coups by 0.13 units and increases tolerance by 0.11 units, effects that are slightly to moderately greater than the maximum effects we found for worry about terrorism for these variables.

Conclusion In this paper we have provided a portrait of concern about terrorism in Latin America and the Caribbean. Worry about the threat of violent attacks by extremist individuals and groups is relatively high in many countries in the Americas, which is consistent with a reality in which both international and domestic terrorism is becoming more lethal, sophisticated, and brazen, and consistent with a reality in which security concerns in general are quite high (when asked about the most serious problem facing the country, the 2010 AmericasBarometer survey shows that 28.3% of respondents in 26 countries in the Americas indicate an issue related to security). Of course, not all individuals are worried about terrorism, and our question was designed to take into account the fact that an important number of respondents would not have thought 34

It is still theoretically plausible that worry about terrorism could be endogenous to support for democratic values if, for example, those who do not support democracy feel generally insecure and specifically insecure about terrorism. To test for endogeneity, we performed a Hausman Specification Error Test on the non-categorical dependent variables. We find no evidence of simultaneity with respect to the Churchillean Measure, Anti-Military Coups, and Support for Public Contestation; we do, however, find evidence of simultaneity with respect to ProParties and Political Tolerance. Therefore, it is possible that – in some cases – there is a reciprocal relationship between one’s concerns about terrorism and democratic values. More details are available from the authors.

17

about terrorism. Therefore, rather than steer respondents to select a level of worry, we allowed the explicit option, “haven’t thought much about it.” As reported earlier, 23 percent of respondents on average selected this option. While we excluded these individuals from the analyses here, as reported, we ran robustness checks in which these individuals were coded into the “not at all worried” category and in which all response options were treated as separate categories, and these achieve similar results (see footnote 24 for example). Our analyses here show that, in statistical terms, countries in which the rule of law is strong are those in which citizens are less worried about terrorism. It could be that such environments are less likely to breed extremists, either because efficient, well-executed, professional security policies act to prevent the growth of terrorist activity, or because such features and a general respect for individuals’ legal rights stymies attempts by extremists to convert others to their cause (see Choi 2010). Alternatively, our rule of law variable may proxy for the general state of security in the country, leaving individuals feeling more secure with respect to terrorism, but likely also other types of crime, in contexts where the rule of law is judged to be comparatively high. Regardless of which interpretation, or whether a mixture of these interpretations, is most accurate, the results at least suggest the utility of a strong rule of law for promoting feelings of security, in this case with respect to concern about terrorist attacks. In addition, information climate matters. Once again, the causal story is likely complex; specifically, we would suggest that the nature of the information environment as captured by the media can affect public concerns about terrorism and, as well, is influenced by those concerns (and the actual experiences and events that may drive those concerns). Finally, we find that past attacks increase citizen insecurity with respect to terrorism; in this case, the causal relationship would appear more straightforward: numerous violent attacks by terrorists induce greater worry in individuals about the likelihood of future attacks. While it is important to understand the sources of worry about terrorism, our paper makes another and important contribution by demonstrating the wide-ranging and significant damaging effects that concerns about terrorism can have on support for democracy in the abstract, the democratic rules of the game, and key democratic values. The effects of terrorism worry on these dependent variables vary from quite substantial to more modest maximum effects. However, it is worth noting that these significant effects obtain just with respect to worry about terrorism; they could arguably be larger in the context of an actual terrorist attack. Are citizens 18

in all countries equally susceptible to the tendencies we have documented here, where terrorist threat undermines support for democracy in the abstract? We would suggest this is not necessarily the case and in future work we intend to examine cross-national differences in the relationship between terrorist threat and democratic public opinion. Nonetheless, and speaking to the general finding we present here, to the degree that terrorist plots and attacks of a variety of types continue to threaten individuals and groups within the Americas, the results presented here are concerning with respect to their implications for public opinion and democracy. To be clear, we are not suggesting that the findings we present here should be read as harbingers of democratic downfall in any of the specific countries that we analyze. Scholarship on politics in the region has noted the resilience of basic democratic practices (e.g., electoral democracy) in the face of various and some severe recent crises (see Hagopian and Mainwaring 1999; Seligson and Smith 2010). Nonetheless, we echo concerns expressed by these and other scholars that decay in support for democracy and basic democratic processes and principles can leave countries vulnerable to political instability and declines in democratic quality. Our work demonstrates one important potential mechanism for such decay: worry about violent terrorist attacks.

19

Works Cited Amsel, Lawrence, Richard A. Bryant, Joan M. Cook, Randall D. Marshall, Yural Neria, Eun Jung Suh. 2007. The Psychology of Ongoing Threat: Relative Risk Appraisal, the September 11 Attacks, and Terrorism-Related Fears. American Psychologist. 62 (4): 304316. The AmericasBarometer by the Latin American Public Opinion Project (LAPOP), www.LapopSurveys.org. Berrebi, Claude, and Esteban F. Klor. 2006. “On Terrorism and Electoral Outcomes: Theory and Evidence from the Israeli-Palestinian Conflict.” The Journal of Conflict Resolution 50(6): 899-925. Bloch-Elkon, Yaeli, Brigitte L. Nacos, Robert Y. Shaprio. 2007. Post-9/11 Terrorism Threats, News Coverage, and Public Perceptions. International Journal of Conflict and Violence. 1 (2). Bonanno, George A., and John T. Jost. 2006. “Conservative Shift Among High-Exposure Survivors of the September 11th Terrorist Attacks.” Basic and Applied Social Psychology 28(4): 311-323. Choi, Seung-Whan. 2010. “Fighting Terrorism through the Rule of Law?” Journal of Conflict Resolution. 54(6): 940-966. Crenshaw, Martha. 2010. The Consequences of Counterterrorism. New York: Russell Sage Foundation Publications. Das, Enny, Brad J. Bushman, Marieke D. Bezemer, Peter Kerkhof, and Ivar E. Vermuelen. 2009. “How Terrorism News Reports Increase Prejudice Against Outgroups: A Terror Management Account.” Journal of Experimental Social Psychology 45(3): 453-459. Davis, Darren W. and Brian D. Silver. 2004. “Civil Liberties vs. Security: Public opinion in the Context of the Terrorist Attacks on America.” American Journal of Political Science 48(1): 28-46. Davis, Darren W. 2007. Negative Liberty: Public Opinion and the Terrorist Attacks on America. New York: Russell Sage Foundation Publications. Duch, Raymond M. and James L. Gibson. 1992. “’Putting Up With’ Fascists in Western Europe: A Comparative, Cross-level Analysis of Political Tolerance.” Western Political Quarterly 45: 237-273. Echebarria-Echabe, Agustin, and Emilia Fernández-Guede. 2006. “Effects of Terrorism on Attitudes and Ideological Orientation.” European Journal of Social Psychology 36, 259– 265. 20

Edwards, Michael, Libby Marden, Judy Wang, and Alex Zarecky, with Mariana Rodríguez. 2011. Political Tolerance in the Americas: Should Regime Critics be Allowed to Vote? Working Paper, Vanderbilt University. Fernandez, Kenneth E. and Michele Kuenzi. 2010. “Crime and Support for Democracy in Latin America.” Political Studies 58: 450-471. Hagopian, Frances, and Scott P. Mainwaring, Eds. The Third Wave of Democratization in Latin America: Advances and Setbacks. New York: Cambridge University Press. Hetherington, Marc J., and Jonathan D. Weiler. 2009 . Authoritarianism and Polarization in American Politics. New York: Cambridge University Press. Huddy, Leonie, Stanley Feldman, Charles Taber, and Gallya Lahav. 2005. Threat, anxiety, and support of antiterrorism policies. American Journal of Political Science 49 (3): 593-608. Kalkan, Kerem Ozan, Geoffrey C. Layman, and Eric M. Uslaner. 2009. “Band of Others? Attitudes Toward Muslims in Contemporary American Society.” Journal of Politics 71: 847-862. Lemyre, Louise, Michelle C. Turner, Jennifer E. C. Lee, and Daniel Krewski. 2006. “Public Perception of Terrorism Threats and Related Information Sources in Canada: Implications for the Management of Terrorism Risks.” Journal of Risk Research 9(7): 755-774. Luke, Douglas A. 2004 . Multilevel Modeling. Thousand Oaks, CA: Sage Publications Inc. Marquart-Pyatt, Sandra and Pamela Paxton. 2006. “In Principle and in Practice: Learning Political Tolerance in Eastern and Western Europe.” Political Behavior 29: 89-113. McClosky, Herbert and Alida Brill. 1983. Dimensions of Tolerance: What Americans Think about Civil Liberties. New York: Russell Sage Foundation Publications. Merolla, Jennifer L., and Elizabeth J. Zechmeister. 2009. Democracy at Risk: How Terrorist Threats affect the Public. Chicago, IL: University of Chicago Press. Nacos, Brigitte L., Yaeli Block-Elkon, and Robert Y. Shapiro. 2011. Selling Fear: Counterterrorism, The Media and Public Opinion. Chicago, IL: University of Chicago Press. Seligson, Mitchell A. and Amy Erica Smith, Eds. 2010. Political Culture of Democracy, 2010: Democratic Consolidation in the Americas in Hard Times: Report on the Americas. Nashville, TN: Vanderbilt University.

21

Sánchez-Cuenca, and Luis de la Calle. 2009. “Domestic Terrorism: The Hidden Side of Political Violence.” Annual Review of Political Science 12:31–49. Shamir, Michal and John Sullivan. 1983. “The Political Context of Tolerance: The United States and Israel.” American Political Science Review 77(4): 911-928. Sullivan, John L., George E. Marcus, Stanley Feldman and James E. Pierson. 1981. “The Sources of Political Tolerance: A Multivariate Analysis.” American Political Science Review 75(1): 92-106. 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: 51-76. Sullivan, John L., James Pierson, and George E. Marcus. 1982. Political Tolerance and American Democracy. Chicago, IL: University of Chicago Press. Wang, T.Y., and G. Andy Chang. 2006. “External Threats and Political Tolerance in Taiwan.” Political Research Quarterly 59: 377-388. Wiarda, Howard J. 2010. “Democracy and Security in Latin America: Underlying Causes of Crime, Violence, and Political Terrorism.” Security and Defense Studies Review 11(FallWinter): 171-177. Zechmeister, Elizabeth J., Daniel Montalvo, and Jennifer L. Merolla. 2010. Citizen Fears of Terrorism in the Americas. AmericasBarometer Insights Report #46, Latin American Public Opinion Project (LAPOP), Vanderbilt University.

22

Figure 1. Fear of Terrorism in the Americas, 2010   Colombia Ecuador Paraguay Peru Venezuela Mexico Panama El Salvador Bolivia Honduras Dominican Republic Suriname Nicaragua United States Costa Rica Belize Guyana Guatemala Chile Brazil Canada Jamaica Argentina Uruguay

67.3 64.6 64.4 63.0 58.9 57.9 55.5 53.7 51.6 50.9 49.9 48.1 48.0 47.7 46.7 46.5 39.3 36.4 35.6 34.3 31.3 30.9 28.5 18.7

0

20

40

60

80

Worried about violent attacks by terrorists 95% C.I. (Design-Effects Based) Source: AmericasBarometer by LAPOP

 

23

 

Table 1. Multilevel Analyses of the Determinants of Fear of Terrorist Attacks, 201035 Worry About Terrorism Model I

Model II

Intercept Individual-Level Size of the Place of Residence Quintiles of Wealth Age Female Education Relatives Living Abroad Ideology: Right Ideology: Center Ideology: Missing Perception of Insecurity Household Member Victimized by Crime Crime Victimization

0.548***

(0.037)

0.526***

(0.045)

-0.060*** -0.047*** -0.007 0.085*** -0.122*** 0.024*** 0.022*** 0.020*** -0.008 0.111*** 0.018*** 0.014**

(0.006) (0.007) (0.013) (0.005) (0.012) (0.006) (0.006) (0.007) (0.007) (0.008) (0.006) (0.006)

-0.062*** -0.061*** 0.003 0.085*** -0.110*** 0.016*** 0.017*** 0.024*** -0.010 0.114*** 0.018*** 0.011*

(0.007) (0.007) (0.013) (0.005) (0.012) (0.006) (0.006) (0.007) (0.008) (0.008) (0.006) (0.006)

Country-Level Means as Outcomes Rule of Law Number of Attacks Information Climate

-0.206*** 0.185* --

(0.077) (0.106) --

-0.213*** -0.181^

(0.079) -(0.113)

Variance Components Random effect Worry About Terrorism, u0 0.135*** Number of Individuals 27,213 Number of Country Cases 22 Note: Standard errors in parentheses. ^=p < .12; * = p < .10; ** = p < .05; *** = p < .01, two-tailed

0.133*** 23,858 19

Data Sources: AmericasBarometer by LAPOP 2010, Freedom House 2010, RAND Database of Worldwide Terrorism Incidents, and various newspaper websites (see text for detail). 35

As robustness checks, we ran less parsimonious models that included other theoretically relevant variables, such as “trust in the military” and “trust in the national government;” however, the coefficients were not statistically significant and our key results remained the same. In one set of analyses, we included a question on personality, and found that those who consider themselves as critical and quarrelsome are more worried about violent attacks by terrorists. However, we do not include it in the model because that variable was not part of the Honduras questionnaire (and therefore that entire country is omitted from that analysis), and its exclusion does not change the significance and direction of the other coefficients.

24

Table 2. Worry about Terrorism as a Predictor of Support for Democracy and Related Measures Support for Democracy (Multinomial, base category: Authoritarian System Could be Necessary)

Churchillean Agreement (Continuous)

Intercept Worry About Terrorism Ideology: Right Ideology: Center Ideology: Missing Crime Victimization Household Member Victimized by Crime Perception of Insecurity Evaluation of Personal Economic Situation Relatives Living Abroad Quintiles of Wealth Female Age Education Size of the Town of Residence

R-squared n

It Doesn't Matter

Source: AmericasBarometer by LAPOP 2010

Anti-Strong Leader (Dichotomous)

Pro-Parties (Continuous)

Anti-Military Coups (Continuous)

Support for Public Contestation (Continuous)

Political Tolerance (Continuous)

0.730***

(0.013)

-0.263*

(1.136)

1.241***

(0.122)

1.067***

(0.120)

2.396***

(0.172)

0.649***

(0.020)

0.596***

(0.019)

0.799***

(0.013)

0.592***

(0.018)

-0.048***

(0.006)

-0.055

(0.053)

-0.268***

(0.044)

-0.349***

(0.048)

-0.520***

(0.063)

-0.029***

(0.007)

-0.113***

(0.008)

-0.026***

(0.005)

-0.067***

(0.006)

0.015***

(0.005)

-0.043

(0.054)

0.032

(0.045)

-0.093**

(0.046)

0.030

(0.060)

-0.005

(0.007)

-0.023***

(0.008)

0.003

(0.004)

0.006

(0.005)

0.015**

(0.006)

0.050

(0.061)

0.125**

(0.052)

0.082

(0.053)

0.312***

(0.069)

0.000

(0.007)

0.001

(0.008)

-0.009*

(0.005)

-0.011*

(0.006)

0.039***

(0.007)

0.088

(0.064)

0.124**

(0.054)

-0.068

(0.054)

0.139*

(0.073)

0.019**

(0.008)

0.011

(0.009)

0.006

(0.006)

0.019***

(0.007)

0.005

(0.005)

-0.058

(0.050)

-0.072*

(0.040)

-0.128***

(0.039)

-0.165***

(0.049)

-0.010

(0.006)

-0.026***

(0.006)

0.060

(0.004)

0.010**

(0.0048)

-0.006

(0.005)

0.003

(0.049)

-0.057

(0.042)

-0.044

(0.042)

-0.033

(0.054)

0.003

(0.006)

-0.041***

(0.007)

0.002

(0.004)

-0.010**

(0.005)

-0.025***

(0.007)

-0.225***

(0.067)

-0.258***

(0.054)

-0.187***

(0.055)

-0.243***

(0.075)

0.010

(0.009)

-0.054***

(0.009)

-0.020***

(0.006)

-0.012*

(0.007)

0.021*

(0.011)

-0.025

(0.067)

0.030

(0.084)

0.101

(0.093)

-0.134

(0.116)

-0.037***

(0.014)

0.051***

(0.015)

-0.021**

(0.009)

-0.053***

(0.010)

-0.006

(0.005)

-0.107**

(0.048)

-0.173***

(0.039)

-0.043

(0.041)

-0.245***

(0.047)

-0.015**

(0.006)

-0.011*

(0.006)

-0.003

(0.004)

-0.007

(0.004)

0.025***

(0.007)

-0.001

(0.065)

0.113**

(0.052)

0.044

(0.052)

0.159**

(0.068)

0.029***

(0.008)

0.023***

(.009)

0.023***

(0.006)

0.026***

(0.006)

-0.011***

(0.004)

0.027

(0.039)

0.023

(0.032)

0.110***

(0.029)

0.051

(0.039)

0.009*

(0.005)

0.004

(0.005)

-0.011***

(0.003)

-0.027***

(0.003)

0.169***

(0.010)

-0.268**

(0.116)

0.888***

(0.087)

-0.331***

(0.086)

0.942***

(0.119)

0.103***

(0.014)

0.210***

(0.014)

0.016*

(0.009)

0.004

(0.010)

0.110***

(0.011)

-0.287***

(0.104)

0.432***

(0.079)

0.927***

(0.078)

1.172***

(0.116)

0.067***

(0.012)

0.131***

(0.014)

0.091***

(0.008)

0.106***

(0.010)

0.007

(0.008)

0.024

(0.070)

-0.037

(0.053)

-0.079

(0.061)

-0.093

().082)

-0.013

(0.009)

0.027***

(0.011)

0.002

(0.008)

0.042***

(0.009)

0.078

--

--

--

--

0.036

0.082

0.095

0.094

25,852

25,491

25,491

26,471

26,110

25,321

26,021

26,858

26,467

Notes: Standard errors in parentheses. Country dummies included but not shown here. * = p < .10; ** = p < .05; *** = p < .01, two-tailed.

Democracy is Preferable

Anti-Iron Fist (Dichotomous)

Table 3. Maximum Effects of Predictors of Support for Democracy and Related Measures (derived from results in Table 2) Churchillean Agreement

Support for Democracy (Base category: Authoritarian System Could be Necessary) It Doesn't Matter

Worry About Terrorism Ideology: Right Ideology: Center Ideology: Missing Crime Victimization Household Member Victimized by Crime Perception of Insecurity Evaluation of Personal Economic Situation Relatives Living Abroad Quintiles of Wealth Female Age Education Size of the Town of Residence

Anti-Iron Fist

Anti-Strong Leader

ProParties

Anti-Military Coups

Support for Public Contestation

Political Tolerance

Democracy is Preferable

-0.048

N.S.

-5.2

-0.069

-5.8

-0.029

-0.113

-0.026

-0.067

0.015

N.S.

N.S.

-0.022

N.S.

N.S.

-0.023

N.S.

N.S.

0.015

N.S.

2.0

N.S.

2.6

N.S.

N.S.

-0.009

-0.011

0.039

N.S.

1.6

-0.014

3.7

0.019

N.S.

N.S.

0.019

N.S.*

N.S.

-1.2

-0.025

-1.8

N.S.

-0.026

N.S.

0.010

N.S.

N.S.

N.S.

N.S.

N.S.

N.S.

-0.041

N.S.

-0.010

-0.025

-0.20

-3.6

-0.033

-2.5

N.S.

-0.054

-0.020

-0.012

0.021

N.S.

N.S.

N.S.

N.S.

-0.037

0.051

-0.021

-0.053

N.S.

0.40

-3.0

N.S.

-2.1

-0.015

-0.011

N.S.

N.S.

0.025

N.S.

2.0

N.S.

1.1

0.029

0.023

0.023

0.026

-0.011

N.S.

N.S.

0.212

N.S.

0.009

N.S.

-0.011

-0.027

0.169

-8.4

.16.9

-0.063

8.2

0.103

0.210

0.016

N.S.

0.110

-7.1

11.6

0.176

12.5

0.067

0.131

0.091

0.106

N.S.

N.S.

N.S.

N.S.

N.S.

N.S.

0.027

N.S.

0.042

*Not Significant p>.1, two-tailed Note: Predicted effects for dependent variables analyzed using multinomial probit (Support for Democracy) or probit (Anti-Strong Leader) are the unit change in probability of falling into the category indicated by the column header. Predicted effects for all the other dependent variables, which were analyzed using OLS, show the maximum effect of each independent variable across its 0-1 range (and therefore match the coefficients in Table 2).

1