Populism, voters, and parties

The comparative literature on populism has focused primarily on explaining the populist offer among political elites through the analysis of partisan manifestos, campaign speeches, and the media (^{Jagers and Walgrave, 2007}; ^{Hawkins, 2009}; ^{Rooduijn and Pauwels, 2011}; ^{Rooduijn, 2014}; ^{Rooduijn, De Lange y Van der Brug, 2014}). The growing electoral success of populist parties in various regions of the world has prompted a broader debate about what populism entails. In general, there are four different definitions of populism: structuralist, economic, institutional-political or strategic, and ideational (^{Hawkins and Rovira, 2017}). Both the structuralist and the economic approaches define populism in terms of the actions and interests of political leaders towards the implementation of a macroeconomic model that seeks short-term growth through economic policies such as import substitution industrialization, with the populist candidate aiming to mobilize his or her voters (^{Cardoso and Faletto, 1979}; ^{Oxhorn, 1998}; ^{Vilas, 1992}; ^{Dornbusch and Edwards, 1991}). According to the strategic definition, populism is built through charismatic leadership with an anti-elite discourse that appeals to direct democracy or a style of government more closely aligned to the best interest of the people (^{Barr, 2009}; ^{Weyland, 2001}). Thus, these two approaches to populism focus mainly on political discourse and supply, limiting the possibility of understanding the diffusion and prevalence of populist attitudes in the electorate.

In contrast, the ideational approach has become the most widely used theoretical framework in political science to understand the growing number of populist movements, parties, and candidacies in various regions of the world (^{Laclau, 2005}; ^{Mudde and Rovira, 2012}). The ideational approach defines populism as a unique set of ideas, in which politics is understood as a Manichaean struggle between the people’s good- will and a conspiratorial elite (^{Hawkins, 2009}, ²⁰¹⁰; ^{Mudde and Rovira, 2012}). Following the ideational logic, there are three factors that make up populism: 1) a Manichaean and moral cosmology; 2) a proclamation of the “people” as a homogeneous and virtuous community; and 3) a representation of “the elite” as corrupt, selfish, and self-serving. Contrary to structuralist or strategic definitions, according to the ideational approach, the reference to the “people” as a superior entity for the identification of populism is not enough; instead, these three characteristics must be jointly present. One of the main advantages of the ideational approach is that it makes possible to identify the populist elements present in the speeches and attitudes of political leaders, as well as their manifestation and prevalence among the electorate. In this way, going beyond support for populist leaders, the most recent literature has examined the factors behind populist attitudes among voters, following the ideational view and thus complementing the dimension of populist demand (^{Aguilar and Carlin, 2017}; ^{Hawkins et al., 2018}; ^{Meléndez and Rovira, 2017, among others}).

Previous studies on populist attitudes have identified, in general terms, three conditions for their activation: context, framing, and mobilization. It should be noted that, although these three different elements have been considered as crucial for the activation of populist attitudes, empirical studies in this regard usually focus on a particular element —context, framing, or anger— without analyzing them as a whole or referring to the processes by which they are activated during campaigns. Furthermore, it is important to also emphasize that most of the works do not usually use survey data and their geographical coverage is usually limited to European countries (^{Spruyt et al., 2016}; ^{Tsatsanis et al., 2018}). Therefore, we still know little about the logic of its emergence and prevalence in Latin American countries. The limited evidence in the region has been concentrated in the Chilean case (^{Aguilar and Carlin, 2017}; ^{Hawkins et al., 2018}), along with general reviews of support for populist leaders in Latin America (^{Doyle, 2011}), although without an understanding at the individual level.

The next section details our theory of populist activation of the electorate. As we explain, unlike previous studies that have analyzed populist attitudes, the present study focuses on the conditions that allow the activation of such attitudes among voters. We argue that populist attitudes are not relevant in themselves, but require both an activation under a given context and the work of ambitious politicians with the capacity to effectively politicize the grievances of the electorate. Likewise, our theory argues that the voters’ response to populist framing depends not only on the populist attitudes prevailing among voters, but also on the political predispositions of the voters. Specifically, we argue that partisanship can hinder or facilitate voters’ responses to this populist framing during political campaigns, depending on whether or not that identity agrees with the candidate of the party that represents it.

Populist activation of the electorate

The growing academic consensus around the ideational approach argues that populism characterizes the public sphere as divided between the “people” and a type of elite or political establishment (^{Laclau, 2005}; ^{Mudde and Rovira, 2012}). Populism unites the demands and grievances around the “people”, which can only be successfully mobilized if there is a favorable context for populist rhetoric (^{Borschier, 2010}; ^{Roberts, 2012}), which varies in each country or region. Some studies suggest that perceived socioeconomic vulnerability (^{Spruyt et al., 2016}) is associated with a higher prevalence of populist attitudes among individuals, probably because the perception of economic failure weakens the democratic legitimacy of the political class. For example, in the cases of Europe and the United States, this context is constructed from the growing relevance of cultural and identity cleavages (“cultural backlash”, ^{Mudde, 2007}; ^{Kriesi, 2010}; ^{Inglehart and Norris, 2018}), as well as the effects of globalization (“losers of globalization”, ^{Bornschier, 2010}; ^{Kriesi et al., 2012}; ^{Teney et al., 2014}; ^{Rama and Cordero, 2018}), both of which are triggers of populist demand among voters. Populist candidates in such contexts have been able to take advantage of both significant representation deficits —policies that voters support but have not been successfully channeled by parties and elites— and valence deficits —economic prosperity, good governance, security, etc.— that political elites have been unable to address (^{Hawkins et al., 2017}; ^{Roberts, 2012}).

It is important to mention, however, that not just any context constitutes an opportunity for populist mobilization. As explained by ^{Hawkins et al. (2018)}, fertile grounds for populism are the scandals that show deeply rooted behavior such as, for example, systemic corruption (^{Hanley and Sikk, 2014}; ^{Hawkins, 2010}; ^{De la Torre, 2010}), which generates citizen dissatisfaction with democracy (^{Kriesi, 2014}). Large-scale scandals can vary in each party system, but what is relevant is the generation of a perception of the political elite as a group colluded against the “people”. This context weakens the democratic legitimacy of the parties and the political class and makes a populist candidacy an option in response to that crisis of legitimacy.

It should be noted that populism studies tend to assume that voters effectively perceive the failures that the populist candidate denounces in the country, which are in turn translated into negative evaluations of the national context. However, these negative evaluations must be verified empirically —for example, through opinion polls— to identify whether such a context conducive to populist mobilization is indeed recognized among the electorate. Furthermore, it is necessary to consider that it is possible that the electorate’s populist attitudes (^{Akkerman et al., 2014}; ^{Hawking et al., 2018}) exacerbate the negative evaluation of the national situation, so that voters with a lower degree of populist attitudes see this context as less negative. This means that this context is limited not only by objective conditions of the economy, security, or corruption in each country, but also by the perceptions with which voters view that national context. Thus, the first hypothesis of this study is the following:

H1 (context). The greater the degree of populist attitudes among voters, the greater the probability of reporting negative evaluations of the national context.

It is important to acknowledge, however, that the political context and the corresponding evaluations do not in themselves generate populist mobilization. Many voters may indeed possess populist attitudes, but these can remain latent (^{Hawkins and Rovira, 2017}) and not manifest themselves. For this reason, the capacity of ambitious politicians is necessary to politicize the country’s grievances and make them salient. To do this, consistent with the ideational approach of populism, populist politicians often present the country’s problems as a struggle between the people and a corrupt elite. Such populist rhetoric or framing has several functions. First, it fosters a common ingroup identity that makes citizens feel identified as part of the “people” (^{Hawkins and Rovira, 2017}), beyond their particular interests (for example, class consciousness or religion). Second, populist rhetoric or framing conveys the perception of the existence of a collusion among the corrupt elite against the interests of the “people” (^{Hawkins et al., 2018}). And third, this rhetoric also identifies people as honest and virtuous and, therefore, as victims of the corrupt elites that act in detriment to their best interests. As ^{Hameleers et al. (2016)} explains, populism inherently blames elites for negative events and completely absolves the people.

Various studies have additionally found that a populist framing that refers to an antiestablishment identity represents fertile ground for the emergence of populist attitudes (^{Melendez and Rovira, 2017}). It is important to note that the literature of populism tends to assume that the electorate effectively believes in the existence of that corrupt elite that conspires against the will of the people, but few studies have empirically verified such belief regarding the elite. In this sense, in order to study populist activation of the electorate, it is necessary to identify that said framing effectively permeates the electorate and, in turn, interacts with the populist attitudes of voters. Therefore, the second hypothesis of this study is the following:

H2 (populist framing). The greater the degree of populist attitudes among voters, the greater the probability of believing in a corrupt elite.

Up to this point, we have exclusively referred to the perceptual effect of populist attitudes, without necessarily leading to political behavior. The context and framing through which political reality is interpreted are necessary but not sufficient conditions to effectively mobilize populist attitudes electorally. For this purpose, it is necessary that there be anger among the electorate so that these populist grievances and attitudes can be activated (^{Hawkins et al., 2018}). It is possible that there are voters who have a negative evaluation of the country’s situation and even consider that the elite is colluding, but that do not necessarily have a motivation to translate these variables into electoral mobilization. In this sense, as previously suggested and consistent with the literature in participation, anger represents a central variable that allows translating populist attitudes into political behavior.

Anger in politics tends to increase the political participation of individuals (^{Valentino et al., 2011}; ^{Weber, 2013}), including protest (^{Van Troost et al., 2013}), while other types of emotions, such as fear or anxiety, makes individuals more risk-averse and open to compromise (^{Mackuen et al., 2010}). In this case, populism uses emotions to assign blame and anger motivates action against the elites responsible for failures in a country. As ^{Rico et al. (2017)} explain, anger is also accompanied by a normative judgment that encourages a response from those who feel aggravated. And more-over, the populist inclination to divide society into two antagonistic groups necessarily makes anger polarizing and facilitates responsibility attribution, which also motivates action against elites (^{Arceneaux, 2003}; ^{Javeline, 2003}). In this sense, the third set of hypothesis of this study is the following:

Finally, on the side of political demand, there is a variable that will allow the translation of populist rhetoric and anger into electoral behavior, but which has been scantly studied in the literature on populism: partisanship. The most important variable to understand voting behavior is partisanship, as it constitutes the filter through which voters give meaning to the political world (^{Green, Palmquist and Schickler, 2004}; ^{Lewis-Beck et al., 2008}). Moreover, partisanship is an information filter that makes voters absorb information in a biased way: voters often accept information that is consistent with their political predispositions and reject information that is contrary to their way of understanding the political world ^{Lodge and Hamill, 1986}; ^{Kuklinski and Hurley, 1994}; ^{Zaller, 1992}).

Following recent studies in Latin America that find that partisanship is stronger than previous analyses have assumed (^{Baker and Renno, 2019}; ^{Castro Cornejo, 2019}; ^{Lupu, 2015}), this study argues that voters’ partisan attachments constitute a moderator that will allow or will reduce the likelihood that voters will accept populist rhetoric. This means that even if voters have a similar level of populist attitudes, the activation of those attitudes will be conditioned by partisanship. For the same reason, voters who share partisanship with the populist candidate participating in a given election are more likely to accept a populist framing that considers that there is a corrupt elite and that the situation in the country is very serious. Likewise, these voters will be more likely to mobilize electorally, increasing their anger and turnout in the election. In contrast, voters who do not share the partisanship of the populist candidate —even those with a high degree of populist attitudes— will reject a populist framing and will not mobilize electorally based on populist rhetoric or attitudes.

In summary, we argue that populist attitudes translate into electoral mobilization when there is an ideal context to politicize the grievances suffered by voters, which in turn is exploited through an effective rhetoric that persuades an angry electorate to mobilize against the political establishment. However, depending on their partisan-ship, voters will have a different response to populist framing. For the evaluation of this argument, we focus on the 2018 presidential election in Mexico, where the winning candidate relied on populist framing consistent with the ideational approach, apparently benefiting electorally from the electorate’s activation of populist attitudes.

The “power mafia” and populist activation in Mexico

Before the 2018 presidential election, the party system in Mexico was one of the most stable in Latin America (^{Mainwaring, 2018}).² Since the transition to democracy in 1997, the PRI, the PAN, and the PRD had remained the main parties in Mexico. However, the 2018 presidential election represents a break with the traditional party system. Morena and its candidate, López Obrador, managed to win the country›s presidency with 53 per cent of the votes and the legislative majority together with its partisan allies in the Chamber of Deputies and Senate.

López Obrador had already been a presidential candidate in 2006 as a PRD candidate when he lost the election to Felipe Calderón, the candidate for the National Action Party (PAN), by less than one per cent of the vote share. At the time, López Obrador argued that a corrupt elite, the so-called “mafia del poder” (power mafia), had swindled the presidency away from him. This mafia, which, according to López Obrador’s perspective, is composed of PAN and PRI politicians and businessmen, is the cause of poverty in Mexico and the state of “national disaster” resulting from rampant corruption and unbridled neoliberalism in the nation during the last 30 years. In 2012, when López Obrador lost by just over five points to the PRI candidate —Enrique Peña Nieto— the former denounced the electoral result again as a fraud, this time, organized by the power mafia to buy votes, on a massive scale, in support of the PRI campaign.

After the 2012 presidential elections, relations between López Obrador and his party deteriorated markedly after the PRD’S decision to join the “Pact for Mexico” (“Pacto por México”, in Spanish) —with the participation of the PAN and the PRI— which sought to create an understanding between political forces to approve various structural reforms in Congress. López Obrador denounced the PRD for betraying its militants by joining the same “power mafia” as the “PRIAN”, the term he uses colloquially to conflate the PRI and the PAN. Following his resignation from the PRD, López Obrador founded, along with his political allies, a personalist party —the National Regeneration Movement or Morena— which backed his third bid for the presidency. In 2018, his campaign focused primarily on denouncing the corruption of the PRI and PAN governments, energizing the internal market, and repealing the neoliberal structural reforms approved by the “Pact for Mexico” during the six-year term of Enrique Peña Nieto.

Consistent with the ideational approach of populism, López Obrador seems to divide society in two. On the one hand, in López Obrador’s view, “the people” are virtuous as evidenced by his comments that “the greatest wealth of Mexico is the honesty of its people”³ and that his movement is built on “the conviction that the people are good —they are honest” (^{Páramo, 2020}). At the same time, the “power mafia”⁴ is the framing that López Obrador has used to label the corrupt elites who, from his perspective, “believe they own the country” and have conspired among themselves since 1982, when the country passed through a neoliberal phase, which reached its peak during the presidency of Carlos Salinas de Gortari, beginning in1988. From his perspective, “Salinas handed over the nation’s wealth” to a small number of entrepreneurs who have become real masters of the country, as he asserted in his first presidential campaign in 2006:

Let’s bring up those on the bottom, and bring down those on top. What we have to do is unite the people; this is a fight that must be taken up by all Mexicans to defend our interests, against a group that has perpetuated its stranglehold on power and has ruined Mexico. Those at the top do not want to give up power. They are not satisfied. They want to continue devouring the country, but enough is enough. Now it is the people’s turn. It is time for the people to rule this country in a way that benefits the people. Money and power will never win over the dignity and moral character of our people, and we will demonstrate this on July 2 (^{Bruhn, 2012}).

Similar to his rhetoric from the 2006 presidential campaign (^{Bruhn, 2012}), López Obrador in 2018 denounced the PAN, the PRI, and the political establishment as part of a “power mafia” that has impoverished the country with its neoliberal policies and its corruption. It is important to note that the belief that there is a corrupt elite ruling Mexico is not limited only to López Obrador’s rhetoric but has also spread to a significant proportion of the electorate. In fact, according to the CIDE-CSES, 2018 National Electoral Study, 38 per cent of voters in Mexico believe that it is “very true” that there is a “power mafia made up of the PRI, PAN, and businessmen”, while an additional 32 per cent consider this perception to be “somewhat true” (^{Beltrán, Ley and Castro Cornejo, 2020}). The prevalence of this perception then becomes fertile ground for populist mobilization.

Likewise, unlike 2006, the 2018 situation in Mexico represented an ideal context to mobilize voters through populist framing. According to the CIDE-CSES, 2018 National Electoral Study, the Mexican electorate was quite critical of the situation in the country, registering the most negative results since the study was first conducted in 1997: two thirds considered that the economic situation of the country was worse than in the previous government ( see Figure 1). Likewise, the president’s approval ratings are the lowest recorded by the CSES: only 18 per cent of voters approved of the way Enrique Peña Nieto governed, well below previous presidents such as Vicente Fox (67%) and Felipe Calderón (54%; cses, 2018). At the same time, most voters reported being angry with the country’s situation. On a scale of 0 to 10, where 0 means “not angry” and 10 “very angry”, the average is 7.1 (^{7.5 among independent voters; 7.2 among voters who identify with Morena; 6.8 among PAN members; 6.7 among PRI members; Beltrán, Ley and Castro Cornejo, 2020}).

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}).

Figure 1 Evaluation of the country’s economic situation (2000-2018)  

The Mexican electorate is also increasingly critical of the party system resulting from the transition to democracy. According to the same National Electoral Study (CIDE-CSES), at the beginning of the 2018 presidential campaign, 52 per cent of voters expressed that they would never vote for the PRI and 23 per cent that they would never vote for the PAN, while only 11 per cent stated that they would never vote for Morena. Moreover, 46 per cent of voters considered that the PAN, the PRI, and the PRD represented “the same thing”.⁵ Therefore, López Obrador had an ideal context to politicize and electorally mobilize the grievances suffered by voters, which he took advantage of through effective framing that rhetorically denounced corrupt elites as culpable of all the country’s ills.

Populist attitudes of the electorate in Mexico: empirical strategy

To measure populist demand among voters, this work is based on the 2018 CIDE- CSES National Electoral Study, which is part of the Comparative Study of Electoral Systems. The study was carried out as a panel survey with four waves of the same sample of voters. This work focuses on the questions that were measured in the first wave⁶ that had 2 600 interviews with a representativeness at the national level. Module 5 of the cses that was conducted in this edition included a battery of questions consistent with the ideational approach to measure the populist attitudes of voters in various countries of the world. Based on a factor analysis, the questions that registered a common latent dimension of populism were evaluated (Table 1 includes the questions that were part of the index).⁷ Subsequently, an additive index was constructed with those that were part of the populism dimension. This index reports a high degree of reliability based on Cronbach’s alpha (0.72). Subsequently, the index was rescaled from 0 to 1 to facilitate its interpretation.

Figure 2 highlights that the level of populist attitudes among the electorate is quite high: 0.70 on a scale that goes from 0 to 1.⁸ Likewise, as Table A1 in the Appendix shows, the level of populism is not significantly different among voters across partisan groups: not only PAN and PRI partisans (henceforth panistas and priistas, respectively), but also Morena voters (henceforth morenistas) and independents have a high level of populist attitudes. Panistas and independent voters tend to report slightly more populist attitudes, but the magnitude is neither statistically (p> 0.05) nor substantively significant (Figure 2, Table A1 in the Appendix). As explained earlier, this study argues that populist attitudes do not in themselves translate into electoral behavior. As discussed further below, voters, depending on their partisanship, will have a different response to populist rhetoric as we discuss in the next section.

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: dependent variable = populism index (from 0 to 1).

Figure 2 Populist attitudes of the electorate by partisanship  

The analysis also highlights that no socioeconomic variable is a significant predictor of a high level of populist attitudes. Neither the gender of the respondent, nor the type of electoral area they live in (rural or urban) or their civil status are variables associated with their level of populist attitudes. Only the educational level and employment status of voters reaches a statistically significant relationship, but the magnitude is not particularly large: university-educated voters report 4 percentage points more than those with only an elementary school education (p<0.05), while unemployed voters report 6 percentage points less than employed voters (p<0.05). This finding is consistent with recent works that show that the electoral bases of populist parties are not concentrated in a specific socioeconomic group (^{Rooduijn, 2018}) and that, in general, socioeconomic variables tend to have little explanatory power in the variance of populist attitudes (^{Ivarsflaten, 2008}).

To evaluate the three conditions for populist activation of the electorate, the analysis relies on various questions included in the National Electoral Study ( see Table 2). First, to analyze voters’ evaluations of the situation in Mexico (context), we based our work on the questions that measure evaluation of the country’s situation regarding the economy, security, and corruption. To analyze the populist framing, we inquired whether voters consider the existence of the power mafia to be true or false. Finally, to analyze the populist mobilization, we measured the levels of voters’ anger over the situation in the country as well as the probability said voters would participate on election day (turnout). Table A2 in the Appendix reports the descriptive statistics of the variables included in this study. To measure partisanship, we rely on the following question from the National Electoral Study:

“Regardless of the party you have voted for or plan to vote for, do you normally consider yourself panista, priista, perredista (PRD constituent), or do you identify with Morena or some other party?” In the first wave of the panel survey, partisanship is made up as follows: Morena (22%), PAN (17%), PRI (14%), other parties (6%), independent voters (39%), do not know / did not answer (1%). Therefore, we have enough observations to separate the models across partisan groups.

The models that we present in the following section include the control variables that we list below, in order to ensure the robustness of our analysis, as well as to show that our results are not derived from omitted variables. We include socio-economic variables such as age, gender, level of education, employment status, and the type of electoral area where the interviewee lives (rural, urban, and mixed) since it is possible that it is not their populist attitudes but a position of vulnerability what motivates the respondent to develop a negative evaluation of the economy, security, and corruption and consider that there is an elite that colludes against the people. Likewise, the models contain political variables such as presidential approval, the opinion of the interviewees about the PAN, PRI, PRD, and Morena (favorability from 0 to 10), as well as the strength of partisanship (weak/strong partisan),⁹ to be sure that populism is the factor that motivates negative evaluations or belief in the power mafia and that it is not the result of the perception about the party system or the government of president Enrique Peña Nieto.

Results

In what follows, we present the results of our ordinary least squares (OLS) models, which, consistent with the argument of this research, are displayed in aggregate terms and across partisan groups. As previously discussed, populist attitudes are expected to have a different effect on each partisan group. First, we assess the context that makes populist activation of the electorate possible based on the assessment that voters make of the country’s situation, in particular on the state of the economy, public security, and corruption. In the case of the economy (Figure 3, Table A3 in the Appendix), a statistically significant relationship is observed between populist attitudes and the evaluation of the economy in the aggregate (p< 0.01). However, when the analysis separates voters by partisanship, there is a relevant substantive variation, which the aggregate analysis tends to hide. For example, among panistas, morenistas, and independents, evaluations of the economy are more negative as populist attitudes increase. The relationship is statistically and substantially significant (p<0.01) among morenistas, for whom the probability of reporting a negative evaluation of the economy increases by about 15 per cent (from 0.67 to 0.81) when the populist attitudes index increases from 0 to 1. Among independents, the magnitude is quite similar (from 0.73 to 0.87, p<0.01), while in the case of the PAN it increases from 0.58 to 0.73 (p< 0.05). On the contrary, among priistas, there is no relation between the evaluation of the economy and its populist attitudes (p>0.05). These results are robust even when socio-economic and political variables are included in the models as seen in Table A3 in the Appendix. In the case of public security, quite similar results are recorded (Figure 4, Table A4 in the Appendix). While the populist attitudes of panistas (p< 0.05), morenistas (p< 0.01), and independents (p< 0.01) are associated with a negative evaluation of the security situation, among priistas no statistically significant relationship is reported between the two variables (p>0.05). Only in the case of the perception of corruption (Figure 5, Table A5 in the Appendix), there is a significant relationship between populist attitudes and evaluations of the country’s situation both in the aggregate and across partisan groups.

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A3 in the Appendix we report the OLS models including control variables.

Figure 3 Negative evaluations of the economy and populist attitudes  

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A4 in the Appendix, we report the OLS models including control variables.

Figure 4 Evaluations of security and populist attitudes  

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A5 in the Appendix, we report the OLS models including control variables.

Figure 5 Evaluations of corruption and populist attitudes  

Overall, there are differences in the perception of the country’s situation according to voters’ partisanship, which is consistent with the argument of this study. Among priistas, except for the perception of corruption, populist attitudes do not exacerbate a negative evaluation of the country’s situation. In the case of the panistas, morenistas, or independents, the first necessary condition for populist mobilization is present: very critical evaluations of the situation in the country and which are exacerbated by populist attitudes (Hypothesis 1). It is important to emphasize that this difference is found despite the fact that the priistas, panistas, morenistas, and independents have a very similar level of populist attitudes (Figure 2). However, consistent with Hypothesis 4, populist attitudes have a differentiated role in this first condition between priistas and other partisan groups.

A second condition that the literature argues occurs in the process of activating populist attitudes refers to the voters’ reception of a populist framing regarding the existence of a corrupt elite, which has rarely been measured in studies on populist demand among the electorate. In particular, this study analyzes the perception of the existence of a “power mafia” within the electorate. The aggregate results show that there is a significant relationship between the populist attitudes of the voters and the belief that the “power mafia” exists (p<0.01) (Figure 6, Table A6 in the Appendix), but again we see some degree of variation across partisan groups. As the level of populist attitudes among panistas, morenistas, and independents increases, the belief in the “power mafia” increases substantially as well. For example, among morenistas, it increases from 0.38 to 0.85, a change of almost 50 percentage points (p<0.01). This result is robust even when controlling with political variables such as presidential approval and the opinion of the interviewees on the political parties, as well as strength of partisanship (Table A6 in the Appendix). The result is particularly interesting among panistas, among whom belief in a “power mafia” increases from 0.37 to 0.76 as their populist attitudes increase (p < 0.01). Among panistas, populist framing also seems to be successful despite the fact that, according to López Obrador, various pan politicians are part of the “power mafia” that has impoverished the country. However, it is possible that panistas have their own definition of this “power mafia”, distinct from López Obrador’s interpretation, which might explain why a significant proportion of panistas believes that there is such a mafia in Mexico.¹⁰ Among priistas, again, there is no relationship between populist attitudes and belief in the existence of a corrupt elite. In summary, we find that the second requirement for populist mobilization is registered among panistas, morenistas, and independents, but not among priistas (Hypothesis 2), which again highlights the important role of partisanship as a moderator of the relationship between populist and political attitudes (Hypothesis 4), in this particular case, of belief in a corrupt elite.

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A6 in the Appendix, we report the OLS models including control variables.

Figure 6 Power mafia and populist attitudes  

So far, there are three groups of voters reporting two of the three conditions necessary for populist activation: panistas, morenistas, and independents. Next is the third necessary condition for the electoral activation of populist attitudes and which refers to the anger of voters and their subsequent electoral participation. In other words, not only do voters need to assess the country’s situation (context) poorly or belief in a corrupt elite (framing); these perceptions must also translate into anger and turnout (mobilization). As reported in Figure 7, in the aggregate, populist attitudes are significantly associated with voters’ anger, but this result is only recorded among morenistas and independents (p<0.01, Figure 7, Table A7 in the Appendix) and does not hold true among panistas and priistas, for whom there is no significant relationship (p>0.05). Moreover, between morenistas and independents there is a statistically significant relationship between populist attitudes and the probability of voter turnout. Again, populist attitudes do not make the voting among panistas and priistas more likely (p>0.05, Hypothesis 3b, Figure 8). In this sense, only among morenistas and independents are the three conditions for populist activation present and in accordance with the hypotheses of this study (see summary in Table 3). Only these two partisan groups translate their populist attitudes into mobilization and are not limited to a negative perception of the situation in the country and the existence of a corrupt elite.

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A7 in the Appendix, we report the OLS models including control variables.

Figure 7 Anger and populist attitudes  

Source: National Electoral Study, CIDE-CSES, 2018 (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: In Table A8 in the Appendix, we report the OLS models including control variables.

Figure 8 Participation and populist attitudes  

The previous results corroborate the hypotheses of this study: among voters who identify with Morena and independents, the populist attitudes of the electorate exacerbate a negative evaluation of the situation in the country, the belief in a corrupt elite, and the anger over the country’s situation, which constitute the three conditions necessary for the populist activation of the electorate. This activation is not registered among priistas who do not meet any conditions, despite showing a similar level of populist attitudes as morenistas and independents. Among panistas, only two of the three conditions are met, which are limited to the negative perception of the situation in the country and the belief in corrupt elites forming a “power mafia”, but do not display anger or electoral participation (Table 3). Consistent with the literature on political behavior, partisanship represents a moderator, as it is morenistas who accept López Obrador’s populist framing, which is consistent with their political predispositions. One of the successes of López Obrador’s candidacy is that his rhetoric also succeeded among independent voters who also met all three conditions and were activated by López Obrador’s populist framing.¹¹ Therefore, it is not entirely surprising that, also according to the National Electoral Study, regarding voting intention, the majority of independent voters supported López Obrador during the presidential campaign.¹² The results that we reported in the previous paragraphs are based on the first wave of the panel survey that was conducted before election day. If, as the argument of this article holds, populist attitudes are activated by the context and by ambitious elites using a populist framing, such attitudes should exhibit an increase during the campaign, followed by a decrease after the elections, since the context changed radically after the victory of Morena’s candidate, López Obrador. In other words, while populist attitudes may be stable among voters, the connection those attitudes have to other variables —for example, the evaluation of the country’s situation, the belief in a corrupt elite, or anger— should be stronger during the campaign, given the role of political elites (López Obrador’s populist framing) that would reinforce this connection. As ^{Zaller (1992)} maintains, political opinions are a marriage of political predispositions of voters and the signals sent by political elites, making political campaigns a key moment for the candidates’ message. This process is very similar to that described by ^{Gelman and King (1992)} and which activates partisanship during political campaigns. As election day approaches, the connection between partisanship and voting intention grows stronger, so that by the end of the campaign the vast majority of partisans vote for their party’s candidate.

Although we do not have pre-campaign data to estimate the stability of the connection between populist attitudes and political attitudes before the electoral campaign, Table 4 reports the statistical significance of the same models presented in the preceding paragraphs, but with data that measures the context, rhetoric, and mobilization during the third wave of the panel survey that was conducted after the election. Unlike the previous results, after the election in which López Obrador won the presidency, the populist attitudes of voters are a less important predictor than before election day. Among voters that identify with Morena, populist attitudes practically lose predictive power in each item studied in this article—they only maintain a predictive power in relation to the belief in the power mafia (see Table 4). It should be noted that this change is not due to change in partisanship between the first and third waves of the survey, since the sample is restricted to respondents who self-identify as morenistas during the first wave. Although the effect is not as marked among independents (for context and framing, the statistical significance remains), populist attitudes are also less predictive of the different items analyzed in this study.

These data suggest that populist attitudes seem to have been deactivated once the context changes after the historic result of the presidential election, and this effect is particularly important among people who share partisanship with López Obrador. In fact, the connection between these variables is not the only variable that tends to decrease; even among morenistas, the belief in a power mafia and anger over the country’s situation tends to decrease with respect to the start of the campaign and even the evaluations of the situation in Mexico are somewhat more favorable. For example, the percentage of voters who believe it is “very true” that there is a power mafia fell from 44 to 28 per cent, while anger over the situation in the country decreased from 7.2 to 6.8 (on a scale from 0 to 10) between the first and third waves of the poll among voters who identified with Morena in the first wave. As discussed earlier, the context that makes populist activation of the electorate possible is not permanent, but the situation of the country and the perception that citizens have about it are essential in this process.

Discussion

This work contributes to an understanding of the success of Andrés Manuel López Obrador’s campaign in 2018, his third bid for the Mexican presidency. The populist framing that divides Mexican society in two —the “people” vs. the “power mafia”— had moderate success in 2006 and 2012, when López Obrador finished in second place. It was not until 2018 that conditions were particularly conducive to the success of his candidacy and, specifically, to López Obrador’s populist framing. Unlike the candidate’s first two presidential campaigns, this time the electorate found itself aggrieved and angry at the country’s situation, and he successfully mobilized voters to take their grievances to the polls, securing a victory that represents the breakdown of a party system installed at the beginning of the transition to democracy in Mexico.

This work also contributes to the literature on populism, particularly on the demand side. Firstly, unlike previous studies that tend to focus on the conditions that allow populist activation of the electorate in isolation, our article analyzes three conditions that make populist activation of the electorate possible. In this article, we find that populism requires an enabling context for populist activation of voters to be possible. It is also necessary that voters feel aggrieved by their country’s situation. And given that context, the capacity of ambitious politicians is equally necessary to make such grievance salient. In this case, López Obrador activated populist attitudes among morenistas and independent voters. Second, this study also contributes to the populism literature by including a variable that is rarely analyzed but that moderates the relationship between populist attitudes and voting behavior, namely partisanship. Specifically, we find that not all voters are mobilized equally in response to populist framing, even if they register a similar level of populist attitudes. Voters’ partisanship constitutes an information filter that makes it more likely that they will accept populist framing and mobilize, as long as this framing is consistent with their political predispositions.

There are several aspects that this study has not investigated that could be relevant to understanding the conditions conducive to populist activation. For example, future studies should analyze in depth the moderating role of partisanship. While our theory proposes that partisanship is a political predisposition that conditions voters’ attitudes and electoral behavior —consistent with ^{Lupu (2015)}— some voters may self-identify with Morena (a personalist party-movement) because they have populist attitudes and not because partisan self-identification is a political predisposition that precedes such populist attitudes. However, our results suggest that this is not the case. First, the correlation between populist attitudes and self- identification with Morena is not significantly different from correlation with other parties (PAN, PRI, and independents). Likewise, the fact that the relationship between populism, self-identification with Morena, and the three conditions for populist mobilization (context, framing, and anger) decreased after Election Day also suggests that self-identification with Morena and populist attitudes are empirically and conceptually distinct phenomena. These results are not entirely surprising given that although Morena is a new party, it inherited the political brand of Andrés Manuel López Obrador, who has led the political opposition in Mexico since 2000 and whose 2018 presidential campaign was his third bid for office. Now, if this research was replicated in countries with less institutionalized party systems (^{Mainwaring, 2018}) in which new political parties emerge in each electoral period or outsider candidates pursue populist mobilization of the electorate, it is possible that both the populist attitudes and political self-identification with those parties are conceptually and empirically intertwined.

Another topic that future studies could analyze further is context. Firstly, it is possible that a negative perception of the country’s situation may increase populist attitudes, rather than populist attitudes exacerbating a negative perception of context, as our article argues, based on the populism literature. If such a possibility is indeed the case, future studies may explore whether these two variables —context and populist attitudes— mutually reinforce each other or, alternatively, isolate the effect each variable has on electoral mobilization. Likewise, although our work shows that there was fertile ground for populist mobilization given the negative evaluations of the economy, security, and corruption in the 2018 presidential election in Mexico, future studies may ask which particular issue constitutes the most important predictor for mobilization of López Obrador’s voters. As Altamirano and Ley find in this special volume, it seems that the economy and security, rather than corruption, are the most important dimensions to understand López Obrador’s victory in terms of voting intention. This result is especially interesting given the centrality of López Obrador’s discourse denouncing the corruption of the PAN and PRI governments throughout his electoral campaign. Finally, something similar should be investigated in future works regarding voters’ anger. It is possible that there are different reasons why voters are angry about the situation in the country. If this is the case, it may be that identifying the nature of voters’ anger—which could be motivated by various issues such as representation deficits or perceived systemic corruption— may help understand the conditions under which voters will be more or less supportive of or likely to be activated by a populist candidate.