The Public and the Military

Civil-military relations constitute a crucial element in democratic transitions and consolidation (Reference Pion-BerlinPion-Berlin 2001). In Latin America, the focus has been on the transition from military to democratic rule (Reference O’Donnell and SchmitterO’Donnell and Schmitter 1986; Reference O’Donnell, Schmitter and WhiteheadO’Donnell, Schmitter, and Whitehead 1986) as well as the subsequent process of democratic consolidation (Reference ZaveruchaZaverucha 1993; Reference Linz and StepanLinz and Stepan 1996; Reference FitchFitch 1998). For instance, what determines whether a newly established democratic regime becomes consolidated, under which circumstances do the armed forces submit to a subordinate role under civilian government, and how do civilian-military interactions have an impact on democratization? Setting and maintaining limits on the military’s power and autonomy is necessary (Reference Cruz and DiamintCruz and Diamint 1998). In general, a democracy is not consolidated until its armed forces are subordinated to the legitimate authority and control of civilian authorities (Reference HuntingtonHuntington 1957; Reference StepanStepan 1988). Maximizing civilian control of the military therefore involves minimizing the scope of military prerogatives and areas of state policy in which the armed forces holds exclusive jurisdiction (Reference Pion-BerlinPion-Berlin 1992; Reference TrinkunasTrinkunas 2005).

Janowitz (Reference Janowitz1960) was one of the first scholars to conceptualize civil-military relations in terms of the relationship between the military and society. Indeed, social institutions like the military require public support as one source of legitimacy. O’Donnell and Schmitter (Reference O’Donnell and Schmitter1986) emphasize how military dictatorships may affect public opinion of the armed forces, which in turn may have an impact on democratic consolidation and subsequent military autonomy in a post-transition context. They argue that military government abuses and a failure to guarantee economic development will make militaries less popular and provide greater impetus for a democratic transition.Footnote ¹

Part of what determines whether the armed forces become integrated into democracy is whether the government and civil society are willing to allow them. Norden (Reference Norden1996) argues that the more discredited the military government, as with Argentina’s Proceso de Reorganización Nacional (National Reorganization Process, NRP) from 1976 to 1983, the less autonomy civilian authorities were willing to delegate. There, the NRP was politically ineffective, lost a war over the Falklands/Malvinas, and largely failed to convince civilians of the legitimacy of their domestic suppression, thus coloring perceptions of the organization and helping shape future policy toward the military. Upon coming to power in 1983, the Raúl Alfonsín government made significant budget and personnel cuts to the armed forces, and passed two laws (23.077 and 23.554) that limited the institution’s political role. Moreover, the number of people training to be officers fell by 50 percent, a consequence of the declining prestige of the armed forces (Reference NordenNorden 1996, 92–94).Footnote ²

Even in places where democracy has endured, public attitudes and civilian control over the military is an ongoing challenge (Reference BattaglinoBattaglino 2015; Reference DiamintDiamint 2015). Armed forces throughout Latin America continue to seek autonomy from civilian oversight through domestic policy initiatives such as food distribution and policing (Reference Pion-Berlin and TrinkunasPion-Berlin and Trinkunas 2005), internal disaster relief and security (Reference Pion-BerlinPion-Berlin 2016), and economic entrepreneurship (Reference ManiMani 2011a), indicating the dynamism of civilian-military relations. However—or perhaps as a consequence—citizens generally profess a high degree of confidence in the institution. In the 2014 Americas Barometer survey, Latin American respondents found the armed forces more trustworthy than any other state institution (60.1 percent) and second to the Catholic Church (62 percent) overall (Reference PérezPérez 2014).

Sources of Trust in Political Institutions

Two dominant theoretical perspectives compete as explanations for the origins of trust in political institutions. Cultural theories hypothesize that trust is exogenous, originating outside the political sphere in long-standing and deeply seated beliefs about people that are rooted in cultural norms and communicated through socialization in early life (Reference BanfieldBanfield 1958; Reference Almond and VerbaAlmond and Verba 1963). They maintain that institutional trust is an extension of inherent individual characteristics and interpersonal trust, learned early in life and, much later, projected onto political institutions.

Micro-level theories of political trust emphasize that socialization nonetheless may allow for substantial variation among individuals according to demographic characteristics like gender, family background, education, and so forth (Reference DaltonDalton 1996; Reference Inglehart, Basañez and MorenoInglehart, Basañez, and Moreno 1998). Many studies of public opinion of the armed forces focus on sociodemographic causes, arguing that men have greater confidence in the armed forces than women, and, in the United States, that minority groups tend to have greater trust in the institution than the majority group does (Reference LealLeal 2005; Reference Rohall, Ender and MatthewsRohall, Ender, and Matthews 2006). Outside the United States, Tiargan-Orr and Eran-Jona (Reference Tiargan-Orr and Eran-Jona2016) found that the Israeli public’s trust in the Israel Defense Forces is generally high across demographic indicators, although it increases during periods of military confrontation. Similarly, in an examination of the Brazilian armed forces, Kaipper Ceratti, Moraes, and Silva (Reference Kaipper Ceratti, de Moraes and Silva Filho2015) found that trust varies by socioeconomic and regional conditions, as well as by individuals’ perceptions about the legitimacy and effectiveness of specific public policies.

Institutional theories, by contrast, hypothesize that political trust is politically endogenous. They argue that it is a consequence of individual evaluations of agency performance. Speaking to this issue, Serra (Reference Serra1995) found that the two most potent shapers of citizens’ satisfaction with bureaucracy are their assessments of both process that the agency went through (promptness, efficiency, and fairness) and the behavior of street-level bureaucrats with whom they personally interacted in terms of knowledgeableness, courtesy, and openness to criticism.

These theories advance the argument that political trust reflects the expected utility of institutions performing satisfactorily (Reference HetheringtonHetherington 1998). Trust is rational and hinges on institutional performance: organizations that perform well generate it, whereas poorly performing ones generate distrust or low confidence. This suggests that an organization like the armed forces should be able to shape public confidence on the basis of how it carries out things like external defense, international peacekeeping, internal defense missions, disaster relief, and social development projects. In essence, public trust in the military is the result of military effectiveness rather than demographics or interpersonal trust.

In a survey of the comparative literature, Malešič and Garb (Reference Malešič, Garb, Caforio and Nuciari2018) found that this explanation holds great explicative power. In some cases, the participation of European armed forces in international peacekeeping missions can increase their legitimacy and prestige, strengthening public trust (Reference Kuhlmann, Callaghan and KernicKuhlmann 2003; Reference van der Meulen, Callaghan and Kernicvan der Meulen 2003). In other cases, poor performance in peacekeeping can reduce trust, as in Canada (Reference Pinch, Callaghan and KernicPinch 2003) and Germany (Reference Collmer, Malešič and KümmelCollmer 2011). In Russia, public trust fell during the failure of the first Chechen war, rebounded after initial success in the second Chechen War, then bottomed out again as the war entered its insurgency phase (Reference Rukavishnikov and VlachováRukavishnikov 2003).

In Latin America, low levels of effectiveness in realms beyond defense also lead to variation in public opinion of the military. This is especially true with regard to performance in social or national development, as in Ecuador and Cuba (Reference ManiMani 2011b), but also to the economic performance and level of repression under military governments (Reference O’Donnell and SchmitterO’Donnell and Schmitter 1986). For instance, military-civilian relations in Argentina and trust in the armed forces reached their nadir toward the end of the state-sponsored terrorism of the Dirty War and in the wake of the disastrous Falklands/Malvinas War under the NRP. However, the military was able to rehabilitate its image: in a break from the past, it stayed in the barracks during the economic and political crisis of 2001–2002, a move that helped democracy endure. This led to the military’s favorability rating reaching its highest level since 1983 (Reference ManiMani 2011a). As public opinion of the military improved in the 2000s, its budget more than doubled (Reference BattaglinoBattaglino 2013).

A second manifestation of the institutional trust hypothesis is Gordon Allport’s (Reference Allport1954) intergroup contact theory, which proposes that political trust is born of personal experiences and interactions with those from different groups. There are numerous examples in the behavioral literature. In the United States, a number of scholars have shown that racial differences in attitudes toward another bureaucratic body, the police, emanate in part from differences in direct experience: people’s good or bad encounters with judicial actors give rise to positive or negative evaluations of police and court officials and of the justice system as a whole (Reference Weitzer and TuchWeitzer and Tuch 2006; Reference Peffley and HurwitzPeffley and Hurwitz 2010). Therefore, individuals’ perceptions of bureaucratic organizations are shaped by the degree of positive intergroup contact versus negative intergroup contact (Reference Paolini, Harwood and RubinPaolini, Harwood, and Rubin 2010).

Politicization and Partisan Leanings

Despite their predictive accuracy in certain contexts, these theories possess decreased explicative power in other environments. Specifically, they are generally drawn from developed countries where agencies have historically possessed high degrees of bureaucratic neutrality—the idea that civil servants and their institutions are not formally aligned with any political party, movement, or ideology, but instead display political impartiality (Reference HuberHuber 2007).

However, across much of the developing world, public agencies are often politicized (Reference Asmerom and ReisAsmerom and Reis 1996; Reference Van de WalleVan de Walle 2006), and key institutions like the armed forces may lack neutrality (Reference MeyerMeyer 1991–1992; Reference Osha, Usuanlele and IbhawohOsha 2017) even if they are professional (Reference Stepan and StepanStepan 1973). Bellin (Reference Bellin2004, Reference Bellin2012) has shown how executives in the Middle East may manipulate their armed forces, and research on neopatrimonialism in Africa discusses how militaries aligned themselves with the executive but then defect under circumstances such as economic crises (Reference Bratton and Van De WalleBratton and Van De Walle 1997).Footnote ³ Likewise, literature on historical Latin American military development is replete with examples of politicized forces, to say nothing of more recent examples from Brazil (Reference StepanStepan 1988), Argentina (Reference Pion-BerlinPion-Berlin 1997), and Venezuela (Reference NordenNorden 2008). In these contexts, existing insight is less useful in understanding the effects that both politics and military behavior have on public support for the armed forces.

There are different types of military involvement in politics. Significantly, being political is not the same as being politicized, as Huntington’s (Reference Huntington1957) old professionalism might have suggested. Instead, following Stepan’s (Reference Stepan and Stepan1973) new professionalism, bureaucratic neutrality is not the same as being professional, and even professional militaries can be politicized. Politicization is “the substitution of political criteria for merit-based criteria in the selection, retention, promotion, rewards, and disciplining of members of the public service” (Reference Peters, Pierre, Guy Peters and PierrePeters and Pierre 2004, 2). This includes when a political principal appoints someone of the same political persuasion to a key post or when key civil servants are appointed on the basis of their loyalty to a particular political party. A politicized military exercises loyalty to a single political party and advocates for and defends partisan political positions and fortunes. The bureaucratic politics literature shows that these actions can negatively affect agency performance and effectiveness when compared to a neutral competence bureaucracy (Reference Peters, Pierre, Guy Peters and PierrePeters and Pierre 2004; Reference LewisLewis 2007; Reference Gallo and LewisGallo and Lewis 2012), as well as the ideological tenor of the implemented policy (Reference Epstein and O’HalloranEpstein and O’Halloran 1999; Reference GailmardGailmard 2002).

As civil-military relations scholars show, neutral competence in the armed forces helps safeguard the peaceful transfer of power between administrations and ensures that voters can exercise their voice free from the threat of coercion (Reference JanowitzJanowitz 1960; Reference O’Donnell and SchmitterO’Donnell and Schmitter 1986; Reference Pion-BerlinPion-Berlin 2001). With a neutral military, decisions about the funding, size, shape, and use of the military are much less likely to be motivated by a desire to defend partisan power and much more likely to be driven by wider strategic, economic, and public values. Moreover, knowing that partisan intentions do not inform professional military advice also enables elected officials to trust the expertise and advice provided by senior officers. In addition to all of this, politicization may lead the public to judge the agency through the prism of their ideological leanings, a condition that undermines accountability and weakens the institution’s mandate.

Instead of evaluating an agency such as the armed forces on its own merits and performance, individuals may associate a politicized bureaucratic body with the governing party or individual and therefore use their evaluations of the government in power as a heuristic for evaluating the agency. In other words, politicization should matter to public agency support and public agency legitimacy insofar as it activates voters’ political leanings, leading individuals to identify the agency as a partisan extension of that particular administration.

The behavioral literature supports this expectation. In The American Voter, Campbell and colleagues (Reference Campbell, Converse, Miller and Stokes1960) argued that partisan commitments play a vital role in shaping attitudes toward political objects. Since then, others have shown how partisanship structures individuals’ assignation of credit or blame to politicians (Reference Malhotra and KuoMalhotra and Kuo 2008), as well as their evaluations of the economy (Reference Evans and AndersonEvans and Anderson 2006; Reference PopescuPopescu 2013), war (Reference Gaines, Kuklinski, Quirk, Peyton and VerkuilenGaines et al. 2007), and other politically sensitive issues (Reference Slothuus and de VreeseSlothuus and de Vreese 2010). This implies that political affiliations, ideological preferences, and biases all have an impact on how individuals perceive policy propositions and government behavior. A loss of neutrality should therefore induce voters to judge an agency as a function of their perception of the government.

This divergence should grow sharper in a polarized political environment. In fact, given the importance of elite cues to the formation of public opinion, scholars have repeatedly shown that elite polarization also splits public opinion (Reference LevenduskyLevendusky 2010; Reference NicholsonNicholson 2012; Reference Druckman, Peterson and SlothuusDruckman, Peterson, and Slothuus 2013). In his examination of public support for war, Berinsky (Reference Berinsky2007) shows that when political elites disagree as to the wisdom of military action, the public tends to divide as well. In contrast, when elites come to a common interpretation of a political reality, the public gives them great latitude to wage war. In other words, attitudes toward the military should be correlated to respondents’ perception of the government—via identification with the governing party, the president, or the respondent’s dominant ideology. In this sense, individuals who view the government favorably will have greater confidence in a politicized armed force than will citizens opposed to the government.

It also follows that the intensity of those ideologies may further color individuals’ perceptions of the armed forces. As such, the more intense a voter’s party identification or sympathy, the more strongly their opinion of the politicized armed forces will reflect their support for the government. Under bureaucratically neutral conditions, where the military serves the political constitution or country through obedience to democratically elected civilian officials without regard for political party or partisan positions, endogenous factors like military effectiveness and exogenous factors like demography and levels of interpersonal trust should all help structure public opinion toward public agencies like the armed forces. However, as this neutrality erodes and politicization increases, public opinion of the government should become a much more salient structuring factor, regardless of the professionalism of the military.

Venezuela as a Laboratory of Study

To test the effects of politicization on institutional trust, I draw on survey data of citizens’ opinions of the Venezuelan Armed Forces. This case is useful since Chavista Venezuela features politicization of the military from the ideological left rather than the right. Under the Southern Cone military dictatorships of the 1970s and 1980s that were allied with the political right, it is difficult to disentangle whether divided public support for the military is a function of the respondent’s ideology and endorsement or criticism of the government, as conservative views are generally more consistent with support for the military. However, in Venezuela, there is no reason a priori to believe that trust in the military is driven by respondents’ ideologies.

Data

To test my hypotheses, I use data from five waves of surveys from the Latin American Public Opinion Project (LAPOP).Footnote ⁸ Each wave consists of 1,500 randomly sampled individuals from across Venezuela.

To operationalize the dependent variable, trust in the FANB, I use the closely related variable confidence in the armed forces, an ordinal response gathered during all five survey waves (1 = no confidence, and 7 = full confidence). Figure 1, which displays the responses weighted by size of the group (circles) and the best-fit linear relationship (dashed line), shows that, on average, confidence has dropped over time. What is more, while in 2006 a plurality of Venezuelans possessed full confidence in the institution, by 2014 a plurality professed no confidence in it.

Figure 1

Confidence in the FANB (2006–2014).

I use a number of variables to capture the opinion of the government and partisan cues. The first is presidential approval, an ordinal variable that measures respondents’ opinion of government performance (1 = very bad, 5 = very good). Because of possible collinearity among other measures of partisan cues or ideology, I instead use them as robustness checks whose results I include in the appendix.Footnote ⁹ These operationalizations are a dummy variable for whether the respondent voted for Chávez (2006, 2008, 2010, 2012) or Maduro (2014); whether the respondent identifies as Chavista; the frequency with which the respondent attends government-run Communal Council meetings (1 = never, 2 = one or two times a year, 3 = one or two times a month, 4 = once a week); and ideology, a ten-point scale that measures the left-right ideological self-placement of the respondent.

To account for the intensity of partisan beliefs, I include an ordinal variable, party meetings, that measures how often the respondent attended meetings (1 = never, 2 = one or two times a year, 3 = one or two times a month, 4 = once a week).

The descriptive data in Figure 2 bear out that partisanship structures beliefs toward the armed forces. The graphs plot respondents’ mean confidence in the FANB over survey wave by evaluation of the government, from “very good” to “very bad.” There is a clear and constant polarization in attitudes toward the military by constituency over time. Respondents who positively evaluate the government display greater confidence in the armed forces, whereas those who negatively evaluate the government have almost no confidence in the institution.

Figure 2

Confidence in the FANB by approval of presidential performance (2006–2014).

To capture military effectiveness—or more precisely, individual evaluations of military effectiveness—as a plausible alternative explanation, I use three ordinal variables. The first, training, measures the extent to which the individual believes that the FANB is well trained and organized (1 = not at all, and 7 = a lot); the second, disasters, asks respondents how the FANB has done dealing with natural disasters (1 = not at all, and 7 = a lot); and the last, human rights, measures the extent to which respondents’ believe the FANB respects human rights (1 = not at all, and 7 = a lot). Of note, training and disasters are available for only the 2012 survey wave, whereas human rights is available for only 2012 and 2014. Additionally, I cannot simultaneously include all three measures in the same specification because of concerns about collinearity.Footnote ¹⁰ As a result, I begin by running a baseline regression that includes all survey waves, and then rotate in different measures of military effectiveness across specifications.

To capture the interaction of civilians with the military, I include the variable bribe, a dummy that measures whether a soldier had solicited a bribe from the respondent in the previous twelve months. For demographics, I include a dummy variable for the sex of the respondent (man = 0, woman = 1), a dummy for whether they live in an urban (0) or rural (1) area, and the number of years of education (an ordinal variable from 0 to 20).

To test for interpersonal trust and demographic sources of public opinion, I include a number of control variables (summary statistics in the online appendix). These include interpersonal confidence, a four-point indicator of a respondent’s trust in others; neighborhood security, a four-point scale of the respondent’s perception of security; victimization, a dummy variable indicating whether the respondent had been the victim of a crime in the preceding twelve months; satisfaction, a four-point indicator of a respondent’s life satisfaction; and aggregate economic situation, a three-point evaluation of the country’s economic situation.

Analytic Strategy

To test the effects of these covariates, I employ a series of ordered logistic regression models (ordered logit) with year fixed effects. Including year dummies allows me to control for time-specific effects, such as the decrease in the percentage of respondents identifying as Chavista from wave to wave. Because the outcome variable is ordered, I turn to an ordered response estimator instead of continuous response one. There is some debate in the literature about which is preferable as the number of response categories grows (Reference Boorooah and Lewis-BeckBoorooah 2001). However, the proportional odds test, which measures whether the relationship between each pair of outcome groups is similar, indicates that the ordered response estimator does not violate the proportional odds assumption (null hypothesis cannot be rejected; p < 0.1853). As a result, I employ ordered logit in the text and report ordinary least squares (OLS) as robustness checks in the online appendix.

Given that the three military-effectiveness variables along with bribe are available for only one or two waves of data, as well as the fact that the three military effectiveness variable appear to be somewhat collinear, I use four specifications in the baseline regression and two in the subsequent one. Of these, the first model uses all survey waves and withholds the military effectiveness and bribe variables; the second, third, and fourth models rotate in each of them. Results are remarkably consistent across specifications and between the two types of estimators.

Results

There is a strong statistical and substantive relationship between partisanship and confidence in the armed forces. As Table 1 shows, presidential approval and the series of proxies for government support are positive and significant across all specifications, as predicted. Venezuelans who support the government are more likely to have a favorable view of the military, whereas those who are supportive of the opposition are more likely to have a negative view. This relationship holds for the four alternative operationalizations of partisan affiliation (presidential vote, party identification, Communal Council meetings, and ideological self-placement; see the appendix).Footnote ¹¹

Table 1

Presidential approval and confidence in FANB (multilevel ordinal logit).

Note: AIC = Akaike information criterion; BIC = Bayesian information criterion.

*** p < 0.001; ** p < 0.01; * p < 0.05.

Moreover, presidential approval retains its sign and significance while controlling for citizen perception of military effectiveness in the form of FANB training (model 2), its response to natural disasters (model 3), and perceptions of its respect for human rights (model 4). On one level, the fact that these variables are positive and significant indicates that military performance—or at least perceptions of performance—does influence people’s opinions of the institution. On another level, it indicates that even when controlling for this perception, respondents’ confidence in the institution is just as strongly driven by political leanings. The variable bribe is not statistically significant in any model.

Although exogenous determinants of demographics and interpersonal trust are statistically significant to varying degrees and under different specifications, their substantive impacts pale in comparison to the effect of political leanings and perceptions of FANB performance. Education is negative and significant in three of the four models, indicating mixed support for the contention that increased levels of education are associated with decreased confidence in the armed forces. Regarding interpersonal trust, neighborhood security and life satisfaction are negative across models but only significant in models 1 and 4; by contrast, sociotropic evaluation is negative and significant for all four models, suggesting that the less satisfied people are with their surroundings, the less likely they are to trust the military. Curiously, interpersonal trust is not statistically significant in any of the models, casting doubt on this specific exogenous source of public opinion toward the FANB.

To illustrate the effects that partisanship have on the politicized military, I calculate the predicted probabilities of different levels of confidence in the FANB across views of presidential performance with all other variables from model 2 held at their means. As Figure 3 displays, the largest observable changes occur at the low and high ends of the spectrum: as a respondent’s evaluation of the president moves from very bad to very good, the percentage of respondents with no confidence in the FANB falls fifteen percentage points from 18 percent to 3 percent, whereas the percentage with “full confidence” rises twenty-four percentage points from 7 percent to 31 percent. The middle-range categories, whose patterns are similarly consistent, probably underreport change, because individuals who drop from, say, a confidence level of 7 to 6 are offset by those who drop from 6 to 5, and so on.

Figure 3

Predicted probabilities of confidence in FANB by presidential approval (model 2).

Next, to examine the effect of partisan intensity on public opinion, I divide the sample into pro- and antigovernment subsamples and include party meetings as a covariate. Voters willing to take the time to attend party meetings are more likely to be party militants or to more intensely share that party’s beliefs. If voters respond to elite cues, party militants should exhibit more intense feelings toward the military (e.g., PSUV supporters who regularly attend party meetings should have more favorable opinions of the armed forces than other PSUV supporters, whereas activists of the opposition should be even less inclined to look upon the military favorably than less militant copartisans).

As earlier, I use an ordered logit estimator with fixed effects. The first two specifications divide the sample into opposition voters (model 5) and Chávez/Maduro voters (model 6) and leave out independent variables for which there are not five full waves of data; models 7 and 8 include the full range of independent variables.

The results in Table 2 only partially support the contention. As expected, more partisan Chavistas exhibit higher levels of support for the military (models 6 and 8). However, party meeting is not statistically significant in the opposition samples (models 5 and 7), which suggests that level of militancy with an opposition political party is not likely to sway citizens’ views of the armed forces. This result, however, should be interpreted with caution: the extreme fragmentation of the opposition party coalition and weak organizational capacity of many opposition political parties have made party militancy much more difficult for members of the opposition than for progovernment supporters. While 21 percent of government supporters attended a PSUV meeting at least once a year and 5 percent attended weekly, those numbers for members of the opposition were only 8 percent and less than 1 percent, respectively. In this same vein, these results may also indicate an asymmetric attachment to political parties between government supporters and members of the opposition.

Table 2

Confidence in FANB by group (ordered logistic regression).

Note: AIC = Akaike information criterion; BIC = Bayesian information criterion.

*** p < 0.001; ** p < 0.01; * p < 0.05.

Beyond this, a belief that the FANB is well trained is associated with an increase in confidence in the institution, even controlling for party militancy and regardless of whether the respondent identifies as a member of the opposition or a Chavista. Likewise, including this proxy for belief in military effectiveness washes out the effects of interpersonal trust and demographic variables, with the sole exception of the negative effect of sociotropic evaluation.

Discussion and Conclusion

Although views of the military are formed by perceptions of military effectiveness, partisan and elite cues are just as meaningful predictors of confidence in Venezuela’s highly politicized armed forces. The most statistically robust and substantively significant determinants of confidence in the armed forces is the respondent’s view of the government, party affiliation, vote choice, and ideology, along with perceptions of the military’s performance in areas of training, disaster relief, and human rights. In fact, unlike the rest of Latin America, where citizens who identify with the ideological right tend to express higher levels of trust in the armed forces (Reference PérezPérez 2014), it is left-of-center Venezuelans who are more likely to have higher confidence in the FANB. The theory presented here suggests that these partisan divisions in attitudes toward the FANB are a consequence of a polarized electorate in combination with a politicized public agency. Neither element should be independently sufficient, but jointly they appear to drive confidence in the military.

This has a range of implications. To begin, public perception of the FANB—and by extension, any politicized public agency—may be orthogonal to that agency’s actual capacity or performance. That is, contrary to findings about popular support for the Israeli Defense Forces (Reference Tiargan-Orr and Eran-JonaTiargan-Orr and Eran-Jona 2016), in Venezuela the public’s perceptions of politicians also shape its perceptions of a nonautonomous military independent of the military’s behavior. Instead, judgments may often simply be driven by political affiliation or ideology, which can severely weaken the military’s mandate. Therefore, to improve their public image and recover legitimacy, bureaucratic agencies should pursue neutrality before seeking to improve performance.

The results are concerning for Venezuela. The loss of FANB neutrality not only erodes the organization’s institutional legitimacy but, if past research is any guide, is likely to result in less public compliance with the law and a lower commitment to the rule of law (Reference Hough, Jackson, Bradford, Myhill and QuintonHough et al. 2010). Moreover, popular trust in political institutions is vital to democracy (Reference Kramer and TylerKramer and Tyler 1996; Reference Braithwaite and LeviBraithwaite and Levi 1998; Reference HetheringtonHetherington 1998; Reference Tyler, Jost and MajorTyler 2001, Reference Tyler2006), and both politicization and polarization appear to detract from it.

This augurs poorly for democratic stability and security. In the Soviet Union, military politicization in a context of divided public loyalties led to the (failed) 1991 military coup (Reference MeyerMeyer 1991–1992). Bureaucratically neutral agencies are better positioned to retain the respect of governments, even when there is political change. In contrast, politicized agencies should be more vulnerable to rapid turnover, an action that may reduce agency expertise. This implies, for instance, that when Maduro or his party finally leave power, the military and other state bodies may experience a process of “de-Chavezation” similar to the de-Baathification that took place in Iraq. In that case, the purge of Baathist military elements led to a reduction in state security capacity and the continuing function of the Iraqi government (Reference PfiffnerPfiffner 2010). In Venezuela, this would mean a loss of expertise as well as agency capacity.

The results matter for other institutions. If similar survey data existed, one would expect to see comparable polarization of public opinion toward other Venezuelan public agencies. In the country’s polarized climate, politicization has helped undermine the FANB’s legitimacy by drawing the opprobrium of nongovernment supporters. By extension, a loss of neutrality of other government agencies, from the National Electoral Council to the Supreme Court, is likely to undermine their legitimacy and weaken their mandates. In all these cases, as government supporters defect and public opinion of the government decreases, confidence in these institutions should fall as well.

Similarly, if politicization and polarization are necessary conditions for structuring attitudes, then there should be similar patterns in other countries with divided politics and similarly politicized public agencies. Peronist Argentina is a case in point. As president, Juan Domingo Perón polarized Argentine society around himself, a division that has endured even after death (Reference LevitskyLevitsky 2003). He also politicized the military, ordering his subordinates to disseminate Peronist doctrine and including this material as part of the curriculum at the Military Academy and the War School (Reference Pion-BerlinPion-Berlin 1997). Lastly, he sought the personal loyalty of military officers rather than institutional obedience through explicit loyalty tests and overtly attempted to incorporate the armed forces in his political movement (Reference NordenNorden 1996, 26).

One contemporary example is Nicaragua under Daniel Ortega. Between 1979 and 1990, the officer corps of the Nicaraguan military, known as the Sandinista Popular Army, was highly politicized and loyal to the Sandinista party’s socialist ideals (Reference Walker and WalkerWalker 1991). However, after the electoral defeat of Ortega in 1990, the administration of Violeta Chamorro made great progress in developing a professional, apolitical institution, with much greater institutional autonomy (Reference RuhlRuhl 2003). Since returning to power in 2007, however, Ortega has repoliticized the military by promoting loyal Sandinista officers and continually intervening in internal institutional matters, including a period when he became the de facto defense minister (Reference PérezPérez 2018). In all cases, a change in government may cause a change in public opinion, but confidence in the institution retains its upper limit because partisan affiliation will continue to structure respondent’s confidence in those institutions.

Although demographic and agency performance may affect public opinion of the armed forces during peacetime (Reference LealLeal 2005; Reference Kaipper Ceratti, de Moraes and Silva FilhoKaipper Ceratti, Moraes, and Silva 2015; Reference Tiargan-Orr and Eran-JonaTiargan-Orr and Eran-Jona 2016), the Venezuelan case demonstrates that the substantive effect of these factors may also pale in comparison to the effect of political leanings. In doing so, the results support the long line of political behavior research beginning with Campbell and colleagues (Reference Campbell, Converse, Miller and Stokes1960) that shows how important partisanship is in structuring political opinions toward a variety of areas, individuals, and institutions. They are also consistent with scholarship showing how elite polarization may also polarize public opinion (Reference LevenduskyLevendusky 2010; Reference NicholsonNicholson 2012; Reference BoothBooth 2013; Reference Druckman, Peterson and SlothuusDruckman, Peterson, and Slothuus 2013). Yet there is more to explore.

The dimensions of polarization and politicization and how they have an impact on public opinion is a potentially rich vein of research. For instance, what happens when the military is politicized but not polarized, or polarized but bureaucratically neutrality? Brazil offers a great laboratory for this latter scenario: it boasts a reasonably neutral military despite high social and political polarization (Reference Bruneau and TollefsonBruneau and Tollefson 2014). Similarly, under which circumstances might polarization and politicization fail to produce divergence in public opinion? Exploring cases like these is necessary to understand whether polarization and politicization are individually sufficient to help structure attitudes toward public agencies or if they are jointly necessary conditions.

Politicization of the bureaucracy can damage agency performance and alter policy outcomes, but the Venezuelan case shows that it can also undermine the agency’s legitimacy in the public’s eye. Ultimately, this is damaging not only to representative institutions but also to democracy itself.