INTRODUCTION
In mid-2016, Rodrigo Duterte came to the presidency of the Philippines as a populist, anti-establishment candidate, promising to scale up the “strongman” rule that he honed during his years as mayor of Davao City on the southern Philippines island of Mindanao. Although relatively unknown outside of Mindanao just months before the election, Duterte quickly became the most popular candidate, obtaining 39 percent of the vote compared to 23 and 21 percent for his closest rivals. In his election campaign, Duterte vowed to rid the country of illegal drugs within six months of his confirmation. In early March 2016, he pronounced that if elected president, he would kill thousands of criminals, the funeral parlors would be packed, and he would dump 100,000 of the slain criminals in Manila Bay where the fish would grow fat.Footnote 1 He promised no mercy, saying “God will weep if I become president” (Miller Reference Miller2018, 14).
He quickly made good on his promises. Within the first month of the program's implementation, police operations resulted in around 330,000 suspected drug users and dealers surrendering, over 9,000 arrests, and 664 deaths. In its most recent report, the Philippines National Police acknowledged that there were 4,279 deaths related to the government's war on drugs from July 2, 2016 to May 21, 2018.Footnote 2 Other organizations estimate that between July 2016 and January 2018 up to 12,000 people were killed by state security forces or by non-state groups working with implicit sanction from the authorities.Footnote 3 In spite of this, the campaign remains extremely popular with Filipinos (Kenny Reference Kenny2019b). Support for the war on illegal drugs has been explained in terms of a loosely defined “penal populism” (Curato Reference Curato2017; McCoy Reference McCoy2017; Pepinsky Reference Pepinsky2017). It remains unclear, however, whether these qualitative observations are reflective of patterns in public opinion. Are populist attitudes and perceptions of Duterte as a charismatic leader associated with support for the campaign against illegal drugs in the Philippines? More generally, do preferences over penal policy have a basis in underlying populist attitudes and charismatic leadership? If so, what might be the specific social psychological pathways linking such beliefs to penal policy?
In criminology, penal populism refers to an understanding of justice in which criminal and anti-social or deviant activity should be harshly punished (Pratt Reference Pratt2007). Penal populism implies that criminal justice should be informed by the views of “ordinary individuals rather than … elite opinion” (Pratt Reference Pratt2007, 5). Penal populism in this sense is characterized by two principal beliefs: The first is that deviant and anti-social activity, such as petty crime and drug use, incurs a steep cost for regular hardworking people who cannot afford to live in economically segregated neighborhoods, pay for the best medical care, or send their children to private schools. The second is that too often the procedure-laden legal system works to the advantage of criminals, and that elites do not do enough to address the concerns of “ordinary people” regarding crime.
These two beliefs map well to onto populism's supposed people-centric and anti-elitist dimensions (Hawkins and Rovira Kaltwasser Reference Hawkins, Kaltwasser, Hawkins, Carlin, Littvay and Kaltwasser2019). However, although it is heavily implied in existing research that penal populism refers to a latent psychological disposition, it has not thus far been measured or tested as such. Rather, empirical research to date has typically equated preferences for harsh penal policies, such as capital punishment, with penal populist attitudes (Jennings, Farrall, et al. Reference Jennings, Farrall, Gray and Hay2017; Newburn Reference Newburn2007). We investigate whether attitudes towards criminal justice are associated with populist attitudes and preferences for populist or charismatic leadership in a more general sense.
Populism itself remains an elusive concept, literally meaning “a practice, system, or doctrine of the people” (Kenny Reference Kenny2019c, 8). Scholars have conceptualized what this people-centric form of politics might mean in a variety of ways, whether a set of redistributive economic policies (Dornbusch and Edwards Reference Dornbusch and Edwards1991; Edwards Reference Edwards2010), a charismatic form of political mobilization (Kenny Reference Kenny2017; Mouzelis Reference Mouzelis1985; Weyland Reference Weyland2001; Reference Weyland, Kaltwasser, Taggart, Mudde and Espejo2017), a lowbrow political style (Moffitt Reference Moffitt2016; Ostiguy Reference Ostiguy2009), a plebiscitary or illiberal form of democracy (Müller Reference Müller2016; Pappas Reference Pappas2015; Urbinati Reference Urbinati1998), or a political ideology that places the people's will over that of the elite (Canovan Reference Canovan1999; Crick Reference Crick2005; Hawkins and Rovira Kaltwasser Reference Hawkins, Kaltwasser, Hawkins, Carlin, Littvay and Kaltwasser2019; Mudde Reference Mudde2004). No definition is necessarily truer than another, and as some authors have argued, it may be that populism in practice combines elements of both substance (e.g., anti-elite ideology) and form (e.g., charismatic leadership) (Albertazzi and McDonnell Reference Albertazzi, McDonnell, Albertazzi and McDonnell2008; Barr Reference Barr2009; Jansen Reference Jansen2011; Roberts Reference Roberts and de la Torre2014). We work within the latter multi-dimensional conceptualization, theorizing and testing the constitutive relationships between individual preferences over penal policy and (a) populist attitudes and/or (b) the attribution of charisma to a leader.
First, we theorize positive relationships between both dimensions of populist attitudes and support for strict penal policies. As noted above, populist attitudes have both people-centric and anti-elitist dimensions. In practice, populists typically draw boundaries between the good “people” and some illegitimate or unworthy “other” (Müller Reference Müller2016; Urbinati Reference Urbinati2019). Drug addicts and criminals fall outside of the community of people; indeed, addicts in the Philippines are often called “zombies”—literally, the walking dead. Because drug-related crime leads people to fear for their property, and for their own safety and that of their children, addicts and dealers are perceived to be harmful to “ordinary people,” and are deserving of exclusion or punishment. The anti-elite component of populist attitudes should be associated with support for the drug war, less in the sense of the people wanting to punish the elite (as per some versions of populism, e.g., Mudde Reference Mudde2004), than in the sense of not trusting the elite to fulfil their duty to protect the people. Thus, to the extent that people distrust the elite establishment, we would expect them to favor swift, aprocedural, or even vigilante justice over common legal procedure.
Second, charismatic authority by its nature is unbounded by tradition or the law (Weber Reference Weber1978; Willner Reference Willner1984), yet at the same time it is also deeply tied to the desire for order (Shils Reference Shils1965). We would expect those inclined to view a leader as possessing charisma to be more likely to be willing to delegate authority to the leader directly. A strong leader is perceived to have the will and the capacity to circumvent supposedly corrupt legal and political institutions that might inhibit the people's desire for “justice.” That is, charismatic authority substitutes for legal-rational or traditional authority. Note, however, this hypothesized relationship does not necessarily imply that people who perceive a leader as being charismatic are generally illiberal (in the sense of being anti-pluralist) or that they are willing to cede total authority to him in the manner of a dictator (in the sense of having an authoritarian personality); rather, it is consistent with the populist and democratic notion that the people want a strong, but ultimately accountable, leader (O'Donnell Reference O'Donnell1994).
Using evidence from survey modules fielded in the Philippines in 2016 and 2017, we estimate the extent of populist attitudes and the attribution of charisma to Duterte among Filipinos, and then test their relationships with attitudes toward penal policy. We find evidence of a positive relationship between populist attitudes and support for the campaign against illegal drugs and of penal policy in general. We also find support for a relationship between populist attitudes and support for the extra-judicial killing (EJK) of suspected drug users and dealers who resist arrest. Additionally, we find evidence of a relationship between the attribution of charisma to Duterte and support for the campaign against illegal drugs. Our evidence also indicates that support for the war on drugs is not simply a proxy for a preference for authoritarian rule or of distrust in institutions in general. We find no evidence of a positive association between populist attitudes or belief in Duterte's charisma and a either a belief that Martial Law might be necessary or a general distrust of democratic institutions such as the Supreme Court, the Senate, and the House of Representatives. In short, we find a specific association between populist attitudes and the attribution of charisma to Duterte and support for the drug war. We draw out the implications of these findings for research on populism and penal populism in the final section.
THEORETICAL CONSIDERATIONS
As we noted above, we take a multi-dimensional approach to the conceptualization of populism that incorporates both elements of substance (i.e., ideology) and form (i.e., organization). We take from the ideational approach the idea that populism can be understood as a set of beliefs about how the political world should be ordered. As one proponent of this approach puts it, populism in this sense is a “thin-centered ideology that considers society to be ultimately separated into two homogeneous and antagonistic groups, ‘the pure people’ versus the ‘corrupt elite,’ and which argues that politics should be an expression of the volonté générale (general will) of the people” (Mudde Reference Mudde2004, 543). A growing body of research demonstrates that this ideology is measurable as a set of populist attitudes (Akkerman, Mudde, and Zaslove Reference Akkerman, Mudde and Zaslove2013; Bakker, Rooduijn, and Schumacher Reference Bakker, Rooduijn and Schumacher2016; Elchardus and Spruyt Reference Elchardus and Spruyt2014; Hawkins, Riding, and Mudde Reference Hawkins, Riding and Mudde2012; Kenny and Bizumic Reference Kenny and Bizumic2019; Spruyt, Keppens, and Van Droogenbroeck Reference Spruyt, Keppens and Droogenbroeck2016; Schimpf and Schoen Reference Schimpf and Schoen2019; Stavrakakis, Andreadis, and Katsambekis Reference Stavrakakis, Andreadis and Katsambekis2016; Van Hauwaert and Van Kessel Reference Van Hauwaert and Kessel2018). Populism in this sense has two main components: people-centrism and anti-elitism. People centrism refers to the view that “the people” are the repository of positive values in society. Although not explicitly classist or nativist in orientation, it reflects a belief in folk values rather than in learned expertise, the spontaneous judgment of the crowd over abstract procedure (Canovan Reference Canovan1999; Taggart Reference Taggart2000). Anti-elitism asserts that the problems with society are due to the machinations of elites, whether the latter are defined in political, economic, or cultural terms (Mudde Reference Mudde2004).Footnote 4 Even as both people-centric and anti-elite attitudes are consonant with a strong commitment to democratic government, we argue that this world view is also compatible with preferences for harsh penal policies, at least when crime is highly salient (Wiesehomeier Reference Wiesehomeier, Hawkins, Carlin, Littvay and Kaltwasser2019).
Populism, as a broader set of attitudes, has two key components that relate to penal populism as a narrower set of beliefs. First, populism is people-centric. For populists, the will of society's (imagined) majority, the people, should take priority over that of the other, the deviant minority to which it is opposed. Crime—drug-related crime in particular—represents a particular challenge to the social order; across a wide range of historical and political contexts, drug addiction and the criminality associated with its distribution, have given rise to marked social anxieties—even “moral panics”—and punitive law and order policies (Ben-Yehuda Reference Ben-Yehuda1986; Cohen Reference Cohen2011; Forman Reference Forman2018; Hari Reference Hari2015). The pervasive petty crime committed by addicts attempting to feed their habits is associated with widespread fear and anxiety, not only over property, but over personal safety and that of family members (Curato Reference Curato2017). Although individual freedoms can still be important to those holding a populist worldview, criminality and anti-social behavior are offenses against the social order that need to be punished severely. Thus, even though, as in the Philippines case, many of the “victims” of Duterte's penal policy come from poor (i.e., non-elite) backgrounds, their deviant social behavior nevertheless puts them beyond “the people” and worthy of punishment. Drug addicts—typically those on methamphetamines—are frequently likened to zombies in the Philippines, Duterte's rhetoric reflecting a common trope among Filipinos. Dealers and gangs, for their part, represent threats to the family and social order more generally. We thus posit that those holding populist attitudes should also support stern penal policies when directed towards those deemed to be engaging in deviant social behavior.
Second, populism is also anti-elitist. According to the penal populism thesis, regular people have a better sense of justice than elites or politicians; legal procedures and rules work to the benefit of criminals, not least because elites (especially lawyers) manipulate the rules to protect perpetrators rather than victims (Pratt Reference Pratt2007). For populists, elites cannot be trusted to deal with the problems facing society, not least with respect to crime. As Ryan (Reference Ryan2004, 9) puts it, the “people are less and less prepared to leave questions, including difficult penal questions, to their masters.” The point is less that regular people want the elite to face punishment but that they want to cut the elite out of the justice system. This dimension of populism thus similarly leads to the inference that those holding populist attitudes should be more likely to support swift, even street justice, rather than risk that the people's preferences for order be undermined by the elite. As Curato (Reference Curato2017, 94) writes of the Philippines case, “Duterte's appeal lies in his promise to overcome the corrupt bureaucracy in the justice system and deliver peace and order in a swift and decisive manner.”
In practice, when we speak of populism, we typically also speak of populists. That is, although populism as an ideology may refer to a belief that the people's will should prevail over that of the elite, in practice this ideology often coalesces into the form of a charismatically led mass movement (Albertazzi and McDonnell Reference Albertazzi, McDonnell, Albertazzi and McDonnell2008; Barr Reference Barr2009; Jansen Reference Jansen2011; Roberts Reference Roberts and de la Torre2014). A great deal of disagreement and confusion arises with respect to the use of charisma as a political science concept. First, in common language, charisma is often thought of as a personality trait or as some quality that an individual objectively possesses. Even Max Weber, with whom the concept is most closely associated, confusingly stated that charisma refers to “a certain quality of an individual personality by which he is set apart and treated as endowed with supernatural, superhuman, or … exceptional powers or qualities” (Weber Reference Weber1978, 241). However, a more careful interpretation of Weber elucidates that charisma describes a relationship. An individual is charismatic only to the extent that his followers treat him as endowed as such. As Weber argued: “It is recognition on the part of those subject to authority which is decisive for the validity of charisma” (Weber Reference Weber1978, 242). Charisma is thus “an attribute of the belief of the followers and not of the quality of the leader” (Bensman and Givant Reference Bensman and Givant1975, 578). Whether a leader possess charisma, in other words, is in the eye of the beholder, although there is little consensus on how to operationalize this notion of charisma in public opinion (Antonakis, Avolio, and Sivasubramaniam Reference Antonakis, Avolio and Sivasubramaniam2003; Davies Reference Davies1954; Merolla, Ramos, and Zechmeister Reference Merolla, Ramos and Zechmeister2007).
For our purposes, what is critical about charismatic leadership in the Weberian sense is that it is distinct from a traditional or bureaucratic rules-based order. People who believe that a leader is endowed with charisma should have less of an attachment to bureaucratic procedures per se. The belief that Duterte is a charismatic leader—i.e., that he possesses an extraordinary ability to understand the popular mood and to effectively channel it into action—should be associated a greater willingness to delegate authority to him to act outside of regular institutions and norms. This should be particularly the case under extraordinary or emergency conditions, when charismatic leadership is most salient (Weber Reference Weber1978; Willner Reference Willner1984). It is in such times of crisis that the supremacy of authority over the law is revealed (Schmitt Reference Schmitt1985). Thus, although the inclination towards charismatic leadership betrays a deep suspicion of institutional and legal processes, as Edward Shils argues, it is in sympathy with a desire for social order; he writes, “The need for order and the fascination of disorder persist, and the charismatic propensity is a function of the need for order” (Shils Reference Shils1965, 203). Echoing the elements of populist attitudes noted above, from a psychological perspective we propose a constitutive relationship between the fear of disorder and crime (threats to “the people”) and the desire for a strong leader to stand above the corrupt elite to restore order.
Another dimension of Weber's conceptualization of charisma is worth drawing out. Although Weber wrote about the possibility of plebiscitary democracy turning into authoritarianism, he was careful not to equate charismatic leadership with dictatorship or authoritarianism (Weber Reference Weber, Gerth and Wright Mills1946). Indeed, charismatic leaders may depend on popular support to a greater degree than other types of leaders who can draw on tradition and law to legitimate their rule. Weber (Reference Weber1978, 1114–1115) wrote that the “genuinely charismatic ruler” is “responsible to the ruled—responsible, that is, to prove that he himself is indeed the master willed by God … If the people withdraw their recognition, the master becomes a mere private person.” The authority of the charismatic leader derives from popular adoration alone in contrast to other merely personalist rulers. Charismatic leadership “rests on the faith of the ruled” (Weber Reference Weber1978, 1125). The attribution of charisma is consistent with the delegation of authority to a strong leader, but that leader should also be accountable. It is thus in the joint senses that people who attribute charisma to a leader should be willing to delegate authority to him and that this authority should be directed to the imposition of order, at least as the majority understands it, that leads to the connection between charisma and penal policy.
Last, it is conceivable that support for harsh punishment of deviant social behavior could also be associated with authoritarian beliefs, and authoritarian beliefs with populist attitudes and belief in charismatic leadership. The definition and operationalization of authoritarianism as a psychological construct is no less contested than that of populism (Adorno Reference Adorno1950; Altemeyer Reference Altemeyer1996; Stenner Reference Stenner2005). Taking Stenner's (Reference Stenner2005) parsimonious conceptualization of authoritarianism as an enduring predisposition to intolerance of difference, there is some reason to believe that populist attitudes and authoritarian beliefs are related. Moreover, to the extent that populist attitudes refer to an intermediary psychological construct, they may in turn be caused by some of the same deeper psychological dispositions associated with authoritarianism, such as neuroticism and a lack of agreeableness (Kenny and Bizumic Reference Kenny and Bizumic2019). However, we agree with theorists of populist ideology that populism and authoritarianism are distinct. The items included on most populist attitudes scales are consistent with a deep attachment to democratic norms; most items include some reference to the importance of the people, a key principle of democracy. Moreover, as we argued above, the attribution of charismatic qualities to a leader is also consistent with democratic norms, in the sense that the authority of a leader who relies on charisma is dependent on popular endorsement. As we detail below, our measure of the degree to which an individual perceives a leader as possessing charismatic qualities is empirically quite distinct from submission to authority, or social dominance orientation, per se. Our surveys do not include direct measures of personality or of latent authoritarian beliefs. However, they do include a measure of preference for authoritarian government. Although this question could be subject to some social desirability bias, it would seem to be a reasonable proxy of authoritarian dispositions, at least for those at the extremes of the scale.
Our hypotheses are that populist attitudes and the attribution of charismatic leadership to Duterte should be associated with support for the campaign against illegal drugs in the Philippines, approval of criminal policy in general, and support for EJKs specifically.
DATA
To test the relationships between populism, charismatic leadership, and support for the Duterte administration's violent campaign against illegal drug dealers and users, we rely primarily on survey modules embedded in nationally representative surveys conducted by Pulse Asia Research Inc. in September 2016 and September 2017. Each survey was based on a national sample of face-to-face interviews with 1,200 adults (over 18). To adequately cover the regional diversity of the Philippines, 300 respondents were selected from four “study areas”: the National Capital Region (NCR), Luzon (excluding the capital), Visayas, and Mindanao. Multi-stage probability sampling was used to select sixty sample barangays, which is the smallest administrative division in the Philippines, equivalent to a district or ward, from each of the study areas, yielding 240 Barangays in total with ten respondents per Barangay. The procedure for selecting Barangays is described in the Appendix.
MEASURING POPULIST ATTITUDES
In this section we describe how our measure of populist attitudes is constructed. In recent years, several survey items have been proposed to capture populist attitudes in the sense outlined above (Akkerman, Mudde, and Zaslove Reference Akkerman, Mudde and Zaslove2013; Castanho Silva, et al. Reference Silva, Ioannis Andreadis, Anduiza, Blanuša, Corti, Delfino, Rico, Ruth, Spruyt, Steenbergen, Littvay, Hawkins, Carlin, Littvay and Kaltwasser2018; Hawkins, Riding, and Mudde Reference Hawkins, Riding and Mudde2012; Schulz, et al. Reference Schulz, Müller, Schemer, Wirz, Wettstein and Wirth2017). We implemented the six-item battery of Akkerman, Mudde, and Zaslove (Reference Akkerman, Mudde and Zaslove2013), which was the most widely used instrument at the time of our study, and is the scale deployed in the Comparative Study of Electoral Systems (CSES) Module 5 (2016–21). The questions were designed to capture the degree to which individuals prioritize the will of the people over the views of the elite. Respondents were asked to rate their agreement with the following statements (translated into Filipino) on a 5-point Likert scale ranging from 1 (I very much disagree) to 5 (I very much agree):
1. The politicians in Congress need to follow the will of the people.
2. The people, not the politicians, should make our most important policy decisions.
3. The political differences between the people and the elite are larger than the differences among the people.
4. What people call “compromise” in politics is really just selling out on one's principles.
5. I'd rather be represented by an ordinary citizen than an experienced politician.
6. Politicians talk too much and take too little action.
We normalize responses between 0 and 1 and construct a populist attitude scale based on the average of individuals’ responses to all six items. The mean populist attitude score is 0.635 with a standard deviation of 0.122 (n = 2,400).Footnote 5 The inter-item correlation of responses to these items in the Philippines case is modest but comparable to that observed in other non-European cases (Crombach's α = 0.60) (Andreadis and Ruth Reference Andreadis, Ruth, Hawkins, Carlin, Littvay and Kaltwasser2018). One limitation of the Akkerman, Mudde, and Zaslove (Reference Akkerman, Mudde and Zaslove2013) scale is its tendency to capture only a fairly narrow range of the construct. In other words, it poorly captures those holding more extreme positive or negative values on the populist attitudes scale (Van Hauwaert, Schimpf, and Azevedo Reference Van Hauwaert, Schimpf, Azevedo, Hawkins, Carlin, Littvay and Kaltwasser2019). Another limitation is that it does not allow researchers to easily distinguish populism's people-centric and anti-elitist dimensions (Castanho Silva, et al. Reference Silva, Ioannis Andreadis, Anduiza, Blanuša, Corti, Delfino, Rico, Ruth, Spruyt, Steenbergen, Littvay, Hawkins, Carlin, Littvay and Kaltwasser2018).
We investigated whether populist attitudes were correlated with respondent characteristics, namely the region of the respondent, sex, and social class. In the Appendix, Table A2 compares mean populist attitudes across the four survey regions, and Table A3 tests whether populist attitudes in Duterte's home area of Mindanao are different from those elsewhere. We find no evidence that this is the case. We also examine populist attitudes by social class. Table A4 in the Appendix shows that there is no statistically significant difference in populist attitudes across the main middle and upper (ABC) and lower middle and lower (DE) social classes. Table A5 does, however, show that populist attitudes differ by sex. Women have lower populist attitudes than do men. Our data also shows that in the aggregate populist attitudes were marginally higher in 2017 than in 2016 and that the difference is statistically significant (Table A6). Although we cannot infer from this that populist attitudes are variable at the individual level over time, we can say that at the population level, populist attitudes do not appear to be a constant, but are likely partly responsive to changes in political, economic, and social context.
MEASURING CHARISMA
Given persistent disagreement over its conceptualization, charisma has proven notoriously difficult to measure. Management studies have typically operationalized charisma as a set of personality traits that respondents believe a leader to have. They include having vision, pride, selflessness, optimism, enthusiasm, confidence, respect, power, morals, values and beliefs, and a sense of purpose or mission (Antonakis, Avolio, and Sivasubramaniam Reference Antonakis, Avolio and Sivasubramaniam2003). This approach has been replicated in political science (Merolla, Ramos, and Zechmeister Reference Merolla, Ramos and Zechmeister2007). Problematically, however, Weber (Reference Weber1978) himself used charisma in a very different sense. Weber meant it to distinguish charisma as a source of legitimate authority that was wholly distinct from the bureaucratic or traditional forms. While the traits listed in existing studies are largely positive, and while they may correlate into a particular personality type, it is not clear how they distinguish a truly “charismatic” leader from a “regular” one.
We rely on the fact that, for Weberians, charismatic leadership refers to the beliefs of supporters rather than the specific traits of the leader herself (Bensman and Givant Reference Bensman and Givant1975, 578). Of course, whether people believe a leader to be charismatic in the Weberian sense is difficult to measure and the existing literature provides little explicit guidance (Davies Reference Davies1954; Willner Reference Willner1984). To measure this sense of charisma, rather than pre-determine a set of charismatic traits and then measure the extent to which respondents ascribe these traits to Duterte, we instead recovered estimates of the perception of charisma from respondents open-ended descriptions of him in a word or sentence (in Filipino). To code whether respondents were describing Duterte as either a charismatic or regular leader, rather than using an expert or machine coding approach, which would require the detailed specification of a set of key words in advance, we instead used crowd-sourced non-expert text analysis (Benoit, et al. Reference Benoit, Conway, Lauderdale, Laver and Mikhaylov2016).Footnote 6 This approach allowed us to leave the interpretation of a respondent's description open within the confines of Weber's broad understanding of what makes people perceive a leader as charismatic.
We instructed our Filipino-speaking coders as follows:
Each of the following words or sentences was used by respondents in a recent survey to describe some political leaders. We would like you to say whether you think the respondent believes the leader in question to be “charismatic.” A charismatic leader is one who people believe to be gifted with extraordinary personal qualities. Respondents who believe that a leader is charismatic may describe him or her as heroic, infallible, or superlative in some other way. This is in contrast to “regular” leaders who people typically describe in terms of professional competences, policies, or achievements. Do these listed words/phases suggest that the leader being described is charismatic?
For quality control, coders were also required to answer test questions, in which we provided descriptions of our own, which in our view clearly corresponded to charismatic or regular leadership respectively. Test descriptions were randomly presented to coders along with the original respondents’ descriptions. Coders who failed to answer a minimum of 70 percent of the test questions correctly were blocked from coding further responses. For a list of our test questions and their coding see Table A1 in the Appendix. Coders were able to identify charismatic and regular leadership with a good degree of accuracy. Seventy-three percent of the judgments of the charismatic leadership test descriptions were accurate (i.e., matching with our own judgment), while 78 percent of the judgments of regular leadership test descriptions were accurate (see Figure A2 Appendix). No individual coders failed to meet the 70 percent accuracy standard. Forty-two coders (with unique identities) made between 10 and 350 judgments (10 percent of which were test questions). We obtained a minimum of five judgments per description. For our main models, we settled on a threshold of at least 80 percent of judgments (i.e., a minimum of four out of five) in agreement to code a description as charismatic. According to this threshold, 953 respondents’ descriptions were coded to be of Charismatic leader, while 1,447 were not. Lower or higher agreement thresholds are possible. A 60 percent agreement threshold codes 1,570 respondents out of 2,400 as attributing charisma to Duterte, a 70 percent threshold, 1,051 respondents, a 90 percent threshold, 328 respondents, and a 100 percent threshold, 294 respondents. Given that some disagreement in judgment is likely (as observed even the relatively clear-cut test descriptions), we argue that a threshold below 100 percent agreement is preferable, but that it should be more than three out of five (60 percent). Hence, 80 percent would appear to be the most reasonable cut-off. The attribution of charisma to Duterte is higher in Mindanao (see Table A2 and Table A3 in the Appendix) but it does not differ by social class (see Table A4 in the Appendix). Unlike populist attitudes, the attribution of charismatic traits to Duterte does not differ by sex (see Table A5 in the Appendix). The attribution of charisma to Duterte was lower in 2017 than in 2016 (recall that populist attitudes were higher in 2017) (see Table A6 in the Appendix). Populist attitudes and the attribution of charisma to Duterte are not correlated with one another (see Table A7 in the Appendix).
CONTROL VARIABLES
We include four individual level controls in our main models. Social class is measured as a dummy variable, taking on the value of 1 if a respondent is from the upper A, B, or C classes, and 0 if she is from the lower D or E classes.Footnote 7 We also include ordinal categorical controls for Age cohort (by decade) and Education (no or elementary education, some high school, completed high school, vocational, some college, completed college). We include a dummy variable for sex, with a value of 1 denoting Woman. We include study Area fixed effects to account for possible cross-unit heterogeneity in the prevalence of factors such as drug-related criminality or support for Duterte (NCR, Luzon, Visayas, and Mindanao). Additional robustness tests replace Area fixed effects with Barangay fixed effects. We include survey fixed effects where responses are pooled across more than one survey.
DEPENDENT VARIABLE
Our main dependent variable is whether respondents support the following statement, with responses ranging from 1 (I truly do not support) to 5 (I truly support):
Do you support or not support the campaign against illegal drugs of the Duterte administration?
Support for the campaign is high, with 786 of 1,789 respondents truly supporting it, and 787 respondents supporting it. Only 64 do not support the campaign. The distribution of responses over 2016 and 2017 is shown in Figure 1.
FIGURE 1 Support for the Campaign against Illegal Drugs of the Duterte Administration
We also ask respondents to express the degree to which they approve of the government's performance in “Fighting Criminality” and “Enforcing the law on all, whether influential or ordinary people.” Responses range from 1 = truly not approve to 5 = truly approve. Although related, each dependent variable captures approval of a distinct element of penal policy. The partial correlations between each item are modest as indicated in Table A8, ranging from 0.169 between support for the war on drugs and rating on enforcing the law, up to 0.531 between rating on crime policy and rating on enforcing the law. As an additional robustness check, we create an “approval of penal policy” variable that is a simple mean of responses to the three questions. The distribution of responses to each item (with ordinal scales normalized between 0 and 1) and the combined scale is shown in Figure A3 in the Appendix.
Last, we attempted to capture more precisely the relationship between populist attitudes and support for the most extreme penal policies, namely EJKs. To get at this, in our September 2017 survey, we asked a randomly selected subset of 600 respondents whether they supported “the killing by the police of alleged drug addicts or pushers who resist arrest.” The meaning of the phrase is clear to Philippine respondents as it is the commonly deployed official euphemism for targeted assassinations by police. 552 (92 percent) respondents do not support EJKs; 47 (8 percent) respondents do support them (with 1 non-response).
ANALYSIS
We now examine the relationship between populist attitudes, the attribution of charisma to Duterte, and support for the administration's penal policies. Our principal approach is an ordered logistic regression model (over the range of support for the anti-drug campaign (1 to 5). Standard errors are clustered by Barangay in all models. Models 1 and 2 of Table 1 show that both populist attitudes (Populist attitudes) and the attribution of charisma to Duterte (Charismatic leader) are associated with increased support for the anti-drug campaign. Model 1 includes only Populist attitudes, Charismatic leader, and year and area fixed effects; model 2 introduces controls for respondent education, class, age, and gender. In both models, coefficients for Populist attitudes and Charismatic leader are statistically significant.
TABLE 1 Ordered Logit Models of Support for Anti-Drug Campaign
Clustered standard errors in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01
There would seem to be little ambiguity about the violent implications of the campaign against illegal drugs. However, to get at whether our results may have been sensitive to respondents’ exposure to the campaign's violent effects, in our second survey in 2017, we asked respondents to state the degree to which he/she feared that he/she or a member of his/her family could be a victim of the campaign (Fear). 77.9 percent of respondents said that they believe that extra judicial killings were occurring, while 72.6 percent of respondents were at least somewhat concerned that they themselves might be killed as a result of the campaign against illegal drugs. Models 3 includes a binary Fear variable for those who responded positively to this question. Results remain robust.
Table A9 in the Appendix replicates models 1 and 2 in Table 1 using Barangay fixed effects in place of Area fixed effects. Results remain broadly similar, but the coefficient on Populist attitudes just falls below the conventional threshold for statistical significance (p = 0.069). It is possible that there could be some omitted contextual factors at the very local level, such as a recent shooting, which are correlated with both approval of the war on drugs and populist attitudes and belief in Duterte's charisma. However, these results should be interpreted with caution, as with only seven to eight observations per Barangay, thirty-one observations are completely determined, making the reported standard errors unreliable and no overall estimate of fit obtainable for models 3 and 4. Table A10 replicates the models in Table 1 using OLS in place of ordered logistic models (Angrist and Pischke Reference Angrist and Pischke2009). Results are consistent, but in model 2, which includes the full set of control variables, the coefficient on populist attitudes falls just outside conventional levels of statistical significance (p = 0.064). Given the highly skewed distribution of the dependent variable (Figure A3 in the Appendix), with the vast majority of respondents supporting or strongly supporting the campaign, we tend to give greater credibility to the ordered logistic regression models. However, we are able to run additional OLS models on the more normally distributed composite penal policy dependent variable described below.
Table A11 in the Appendix uses alternative thresholds for the attribution of charismatic leadership to Duterte. Recall that in our main models, we code respondents as attributing charisma to Duterte if at least 80 percent of judgments were in agreement that the description provided was of a “charismatic” rather than “regular” leader. In model 1 of Table A11, we code respondents as attributing charisma to Duterte if at least 60 percent of judgments were in agreement (1,570 respondents out of 2,400), in model 2, 70 percent (1,051 respondents), in model 3, 90 percent (328 respondents), and in model 4, 100 percent (294 respondents). Results are robust except when we restrict coding a respondent as attributing charisma to Duterte to 100 percent agreement among coders.
Table 2 presents the results of OLS models in which the dependent variables are approval ratings of the government's performance Enforcing the law against all (1), Fighting criminality (2), and of the mean of all three dependent variables, which we call Penal policy (3). Each dependent variable has been rescaled to take a value between 0 and 1. Populist attitudes are positively associated with greater approval in both domains and with the constructed penal policy variable. The attribution of charismatic leadership traits to Duterte is similarly associated with Fighting criminality and Penal policy, but not with Enforcing the law against all.
TABLE 2 OLS Models of Approval of Penal Policy
Clustered standard errors in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01
We next test the relationship between populist attitudes and the attribution of charisma to Duterte with support for EJKs. Support for EJKs is not common, most likely reflecting the belief that the killing of suspects who resist arrest is an extreme measure that may be illegal and is perhaps immoral (recall that just 47 out of 599 respondents asked this question admitted support for the strategy). We find that those with more populist attitudes have a higher probability of supporting the killing of suspects who resist arrest (model 1 in Table 3). Notably, however, we find no evidence for a relationship between the attribution of charismatic leadership to Duterte and support for the killing of suspects who resist arrest. It thus may be unsurprising that those who attribute positive leadership traits to Duterte do not necessarily associate him personally with EJKs. Duterte has distanced himself from especially egregious EJKs.
TABLE 3 Logit Models of Support for Extra Judicial Killings
Clustered standard errors in parentheses: * p < 0.05, ** p < 0.05, *** p < 0.01
To investigate further the pathway through which populist attitudes and the attribution of charisma to Duterte may be related to attitudes towards penal policy, we also look at the relationships between these variables and attitudes towards authoritarian rule and institutional trust more broadly. First, Table A12 in the Appendix shows that neither populist attitudes nor belief in Duterte's charisma are associated with belief in the necessity of Martial Law. Second, we construct a measure of “institutional trust,” which comprises trust ratings of the Supreme Court, the armed services, the Philippines National Police, the Senate, and the House of Representatives. The distribution of responses is shown in Figure A4 in the Appendix. Contrary to the argument that populism reflects a distrust of institutions in general (Algan, et al. Reference Algan, Guriev, Papaioannou and Passari2017), we in fact find a positive association between populist attitudes and institutional trust. See Table A13 in the Appendix. In sum, we find that support for the campaign against illegal drugs in a broad sense is strongly correlated with populist attitudes in general and with the attribution of charismatic leadership traits to Duterte.
DISCUSSION AND CONCLUSION
Building on recent conceptual and measurement advances in the study of populism, this article interrogates the penal populism thesis. It asks whether there is a constitutive relationship between individual preferences over penal policy and populist ideological attitudes and/or belief in a leader's charisma. The results indicate that penal policy preferences do indeed share a constitutive relationship with populist beliefs and attitudes towards leadership more generally. We acknowledge that some caution is warranted in drawing out general implications from this evidence. First, we lack precise estimates of related psychological concepts such as authoritarian values and nativism, which may confound the relationship between populist attitudes and support for populist parties. Evidence in from the United States, for instance, indicates that support for Donald Trump is correlated with authoritarian personality traits.Footnote 8 Second, the results are likely to be context dependent. Crime was a highly salient issue for large numbers of voters in the Philippines at the time of Duterte's election in mid-2016. It could be that the association between populist attitudes and penal policy are dependent on this underlying salience. Jair Bolsonaro's rise in Brazil on a law and order and anti-corruption platform has obvious parallels, but without comparative data, we cannot answer that question in this article. These issues noted, we argue that this article advances both the penal populism and populism research agendas in at least three ways.
First, our findings indicate a constitutive association between populist beliefs and certain salient policy preferences. While scholars of populism operating in both the ideational and political-strategic paradigms have tended to parse populist attitudes and mobilization practices from any particular set of policies or interests (Pappas Reference Pappas2016; Moffitt Reference Moffitt2016; Mudde Reference Mudde2004; Weyland Reference Weyland, Kaltwasser, Taggart, Mudde and Espejo2017), our findings instead indicate that populist attitudes may be related to specific policy preferences such as law and order, at least when they are highly salient to voters (Wiesehomeier Reference Wiesehomeier, Hawkins, Carlin, Littvay and Kaltwasser2019). The results presented here also suggest a way of explaining the affinity between populist and nativist attitudes, which has been observed in previous research (Van Hauwaert and Van Kessel Reference Van Hauwaert and Kessel2018). Anti-immigrant policies, unlike, say, pro-worker policies or protectionism, which have more diverse economic impacts, can be easily framed as being in the interests of “the people.” “Illegal” immigrants, like criminals, are especially prone to being excluded from “the people,” justifying harsh policies that target them. The conclusion, to be clear, is not that populist attitudes are causal. Recent research suggests that populist attitudes are themselves intermediary psychological constructs that sit between deeper attitudinal dispositions and political behavior (Kenny and Bizumic Reference Kenny and Bizumic2019; Schimpf and Schoen Reference Schimpf and Schoen2019). Rather, our findings imply a constitutive connection at the attitudinal level between populist attitudes—in the sense of a broader vision of how politics should be ordered—and some of the particular policies, at least with respect to order, common to populist parties and candidates. Future research could elaborate further on exactly how and why individuals come to hold this combination of beliefs.
Second, although existing research has long claimed that populism is associated with top-down, charismatic forms of leadership (Albertazzi and McDonnell Reference Albertazzi, McDonnell, Albertazzi and McDonnell2008; Barr Reference Barr2009; Kenny Reference Kenny2017; Mouzelis Reference Mouzelis1985; Roberts Reference Roberts1995; Reference Roberts and de la Torre2014; Weyland Reference Weyland2001), this dimension of populism has not been as systematically empirically investigated from a public opinion perspective. We provide a novel and replicable approach to the measurement of charismatic leadership and find that populist attitudes and belief in a leader's charisma are not correlated with each other, yet are each correlated with preferences over penal policy. Attitudes towards charismatic leadership are an important set of beliefs in their own right that are distinct from populist attitudes more broadly. If the attribution of charismatic leadership is associated with specific illiberal policy preferences (e.g., on crime or immigration), this again may partially explain the increasingly well-established relationship between populist rule and the erosion of liberal democratic institutions and norms (Foa and Mounk Reference Foa and Mounk2017; Houle and Kenny Reference Houle and Kenny2018; Huber and Schimpf Reference Huber and Schimpf2017; Kenny Reference Kenny2019a; Ruth Reference Ruth2018).
Third, contrary to a large body of theoretical research, which equates populism with “democratic illiberalism” (Pappas Reference Pappas2019) or “anti-pluralism” (Galston Reference Galston2018), we do not find that those holding populist attitudes are illiberal in general. Even though particular populist leaders, such as Duterte, may have illiberal or even authoritarian ambitions, supporters of populists do not necessarily share those same views. Rather, supporters of populists can be attached to liberal democracy and liberal democratic institutions, but still favor some particular illiberal policies, for example, in the domains of law and order or immigration. Paradoxically then, instances of democratic regression may result from the voting preferences of pro-democratic voters, where illiberal leaders nevertheless deliver on policy issues most salient to them. This suggests the possibility of an indirect route to democratic rollback, with populist attitudes leading voters to support candidates who deliver on salient policy preferences, and those leaders in turn using popular support to roll back institutional or civil society constraints on their authority (Kenny Reference Kenny2019a). In the Philippines case, Duterte has clearly utilized the overwhelming popularity of his war on drugs to assault already severely damaged political institutions, from the legislature to the judiciary to the media. We suggest that it is not that Filipinos directly desire this erosion of democracy. Rather, they tolerate it because they support the war on drugs. The implications of this line of thinking go well beyond the Philippines, as research shows that populist attitudes are widely held across democracies in the West, not least in the United States (Hawkins, Riding, and Mudde Reference Hawkins, Riding and Mudde2012). Leaders like Donald Trump may come to power promising a resolution to a highly salient policy demand, in his case on immigration, and then use their popularity to erode democratic norms and institutions. Further research could attempt to better target these chains of causality through the longitudinal analysis of public opinion data.
SUPPLEMENTARY MATERIAL
The supplementary material for this article can be found at https://doi.org/10.1017/jea.2020.8.
CONFLICTS OF INTEREST
The authors have no conflicts of interest to declare.
