The Economy, Security, and Corruption in the 2018 Presidential Election Campaign Issues and Electoral Preferences in Mexico

Altamirano, Melina; Ley, Sandra; Altamirano, Melina; Ley, Sandra

Servicios Personalizados

Revista

Articulo

Indicadores

Citado por SciELO
Accesos

Links relacionados

Similares en SciELO

Otros
Otros

Permalink

Política y gobierno

versión impresa ISSN 1665-2037

Polít. gob vol.27 no.2 Ciudad de México jul./dic. 2020 Epub 17-Feb-2021

Articles

The Economy, Security, and Corruption in the 2018 Presidential Election Campaign Issues and Electoral Preferences in Mexico

Melina Altamirano¹

Sandra Ley²

^¹Profesora-investigadora del Centro de Estudios Internacionales en El Colegio de México. Camino al Ajusco 20, Pedregal de Santa Teresa, 10740, Ciudad de México. Tel: (52) 55 5449 3000, ext. 4033. Correo-e: maltamirano@colmex.mx.

^²Profesora-investigadora en la División de Estudios Políticos del Centro de Investigación y Docencia Económicas (CIDE). Carretera México-Toluca 3655, Lomas de Santa Fe, 01210, Ciudad de México. Tel: 52 (55) 5727 9800. Correo-e: sandra.ley@cide.edu.

Abstract

Throughout Enrique Peña Nieto’s administration and during the 2018 presidential campaign, three problems stood out among the Mexican electorate: the limited economic growth, a sustained increase in violence, and multiple corruption scandals. Therefore, based on the CIDE-CSES 2018 National Electoral Study, we analyze the simultaneous effects of individual evaluations of the economy, violence and corruption on electoral preferences. Our findings indicate the prevalence of a retrospective economic vote that coexists with a security vote, but without clear support from crime victims. Although corruption was widely discussed during the electoral campaign, this issue did not play a major role in voter preferences.

Keywords: issue voting; elections; economic voting; security; corruption

Resumen

Durante la administración de Enrique Peña Nieto y la campaña presidencial de 2018 resaltaron tres problemas entre el electorado mexicano: un limitado crecimiento económico, un aumento sostenido de la violencia y múltiples escándalos de corrupción. Con base en el Estudio Nacional Electoral de México (CIDE-CSES) 2018, analizamos los efectos simultáneos de las evaluaciones individuales respecto a la economía, la violencia y la corrupción sobre las preferencias electorales. Nuestros hallazgos indican la prevalencia de un voto económico retrospectivo que coexiste con un voto de seguridad. Las evaluaciones negativas de la seguridad favorecieron al candidato puntero, aunque sin un claro respaldo de las víctimas. A pesar de su centralidad en la campaña electoral, el tema de la corrupción no afectó sustantivamente las preferencias del votante.

Palabras clave: voto temático; elecciones; voto económico; seguridad; corrupción

Enrique Peña Nieto’s administration was permeated by multiple controversies regarding the economy, security and corruption. The implementation of labor, energy, economic competition, and telecommunications reforms, among others, placed president Peña Nieto on the cover of Time magazine in early 2014. In the fall of that year, however, his administration was facing its worst crisis, due to the disappearance of the 43 students from the Ayotzinapa Teachers’ College and the conflict of interest investigation also known as the Casa Blanca scandal. As a result, the country’s economy and security, in addition to corruption, were three issues that did not go unnoticed by the media and were constantly discussed throughout the 2018 presidential campaign. The economic reforms implemented during the six-year term were subject to constant public discussion. In contrast to the 2012 election, security became a major topic of presidential debates, with diverging proposals, ranging from an iron fist approach, to social prevention policies. Also, the candidates from the three major parties were involved in corruption scandals during the election campaign.

Given this multiplicity of debates and concerns, it is important to understand how Mexican voters took the country’s economic, security and corruption situation into account to define their electoral preferences. What factors mediated the consideration of economic, political and social issues in voting decisions? Which issues prevailed in the evaluation of the different candidates for voting intentions, particularly for the winning candidate? And, for which group of voters was one issue more important than another when casting their vote? These are some of the questions we address in this article, with a specific focus on the 2018 elections in Mexico. The analysis of the determinants of electoral preferences is crucial to understand the results of this electoral process, which points to a reconfiguration of Mexico’s party system. However, at a theoretical level, our motivation and research questions revolve around the simultaneous role that economic and non-economic issues play in voting decisions, and the possibility of identifying the different issue publics that are mobilized in response to three important topics: the economy, security and corruption.

The analysis of these features of the electoral behavior among voters during Mexico’s past presidential election is relevant for both theoretical and practical reasons. Multiple studies have revealed the limitations faced by voters when demanding accountability from their government authorities due to a lack of information (^{Holbrook and Garand, 1996}; ^{Aidt, 2000}) and the subsequent media interpretation of the available information (^{Hetherington, 1996}), along with the complex responsibility attribution processes voters face when deciding who to reward or punish at the polling station (^{Gélineau and Remmer, 2006}; ^{Arceneaux, 2006}; ^{Hobolt et al., 2013}). Additionally, although the economy is usually a major factor in electoral decisions, recent works show that, in times of political crisis, other priorities define voting decisions (^{Singer, 2011}), though these may differ from voter to voter (^{Krosnick, 1990}). Together, these findings suggest that electoral behavior may have important consequences related to accountability and to the policies that will prevail among subsequent governments. In the longer term, electoral preferences and outcomes have fundamental implications for the citizens’ future well-being.

One limitation of this literature, however, has an empirical nature. Due to the restrictions in the design and availability of surveys, it is difficult to simultaneously evaluate the role that different social problems —beyond the economy— play in electoral behavior. However, these issues rarely arise in isolation. For example, criminal violence in Latin America is often accompanied by corruption and can also have profound economic effects. In terms of accountability, it is then necessary to identify which dimension weighs most heavily on the electorate. Without a rigorous and comparative analysis, we could over or underestimate the effect of each issue on electoral results. In this article we analyze the post-electoral survey of the CIDE-CSES 2018 National Electoral Study,^¹ an instrument that allows us to examine, within the same study, the effects of voters’ evaluations of the economy, security, and corruption on electoral preferences and to weigh their impact on individual voting decisions.

In order to understand the logic of the Mexican vote in the last presidential election, we organized the article as follows. First, we present a brief review of the literature on economic and non-economic voting and the main findings that guide our work. Next, we present our argument and hypotheses. Later, we review the main issues that marked the 2018 presidential campaign and their evolution. We then describe our research design and present our results. Finally, we discuss our conclusions and the implications of our findings.

Our results allow us to understand the Mexican voter at a historic political juncture. The evidence presented here points to the prevalence of a retrospective economic vote among the Mexican electorate, a finding in line with previous analyses of past elections (^{Buendía, 1996}, ²⁰⁰⁰; ^{Poiré,
1999}; ^{Beltrán, 2003}, ²⁰¹⁵; ^{Singer,
2009}). However, given the increase of violence in the country, we also find that such an economic vote coexists with a security vote: negative evaluations due to the situation of insecurity favored the leading candidate, Andrés Manuel López Obrador (AMLO), for Morena, although crime victims seem to have had reservations about his proposals and did not widely support him. Furthermore, although the issue of corruption was present in the campaign, particularly in the winning candidate’s discourse, this was not the main issue on which Mexican voters based their electoral decision, nor was it a determining factor in Andrés Manuel López Obrador’s victory. Therefore, this article contributes to a strict evaluation of the logic of the economic and issue voting and offers the possibility of identifying the different issue publics that prevailed or lost relevance in the 2018 election.

Economic voting VS. Non-economic voting

Previous works on economic voting predict that voters, characterized as rational individuals, will decide to reward the ruling party during economic good times and punish it during an economic crisis or decline (^{Fiorina, 1978}). This prediction is based on two assumptions: 1) that economic conditions provide voters with information about political actors and 2) that economic conditions indicate government capacity (^{Dorussen and Palmer, 2002}). In addition, economic voting models assume that voters can easily evaluate economic performance, because the consequences are tangible and reflected in their daily lives. Despite the rationality of economic voting and the vast empirical evidence related to it, there are important limitations.

To analyze the impact of economic conditions on electoral outcomes, one must take into account the political, institutional and social contextual features in which elections take place (^{Powell and Whitten,
1993}; ^{Pacek and Radcliff,
1995}; ^{Anderson, 2000}, ²⁰⁰⁷). For economic models to work, the management of the economy must be a major concern. However, this depends on the individual psycho-socio-logical context (^{Krosnick, 1990}; ^{Dorussen and Palmer,
2002}).

During economic good times or periods of great political or institutional turmoil, voters are more likely to turn their attention to other non-economic issues (^{Singer, 2011}). Comparative policy research has shown that in elections that occur in the midst of government crises involving corruption, human rights violations or terrorist attacks, voters do not pay as much attention to the economy (^{Bali, 2007}; ^{Kibris,
2011}; ^{Singer, 2011}).

Even in the midst of a major governance crisis, voters assign varying degrees of importance to an issue, depending on their own personal concerns and experiences (^{Krosnick, 1990}). The importance of issues is critical to accountability. If a non- economic issue is not relevant to voters, they will not take it into account when deciding who to vote for or when evaluating the authorities in office. According to ^{Krosnick (1990)}, this means that there are multiple “issue publics” within the electorate —each of them is composed of citizens who are especially concerned with a single issue, either because it affects a relevant interest or a personal value. For example, a teacher might be more focused on the candidates’ proposals for education and decide his or her vote based on that dimension.

There are two non-economic issues that have attracted the attention of experts in electoral behavior in recent years: crime and corruption.^² These are visible issues with profound economic, social, and political consequences for voters.^³ Both issues are also widely covered by the media and are often mentioned by politicians in their speeches, so that voters are often exposed to what is happening in terms of both insecurity and corruption, making them more aware of these issues (^{Chiricos et al., 2000}; ^{Altheide, 2002}; ^{Ferraz and Finan, 2008}; ^{Chang et al., 2010}).

Despite the relevance of these issues, the existing evidence suggests that it is only under very particular conditions that voters punish parties for criminal violence or corruption and that, consequently, the chances of making the government electorally accountable for these issues are relatively limited. On the one hand, voters punish their rulers when insecurity is associated with organized crime and there is a partisan alignment that facilitates their accountability attribution process (^{Ley,
2017}). Additionally, previous works reveal that victims of crime do not show statistically significant effects on electoral support (^{Ley, 2017}) or presidential approval (^{Romero et al., 2016}), perhaps because victims tend to disengage from the electoral process and stop participating in elections (^{Ley, 2018}). On the other hand, when corruption is widespread —as it is in Latin America— it is difficult for voters to identify politicians who are not corrupt; because of this, they disregard this issue (^{Pavão,
2018}), and prioritize other social needs in their electoral decision (^{Boas et al., 2018}). Thus, the issue of corruption can influence the vote only when the source reporting on corruption scandals is credible (^{Botero et al., 2015}) and voters are politically sophisticated enough to process such information (^{Weitz and Winters, 2017}).

Considering the limitations that voters have in considering the different issues that directly affect them when casting their votes, this article seeks to examine the extent to which perceptions of economic performance, security, and corruption determined the voting decision in the 2018 Mexican presidential election.

Argument

This article seeks to contribute to the literature on economic and issue voting based on the Mexican case and the most recent presidential election (2018).

Based on the theories of economic voting, we assume that economic evaluations were among the most important determinants of the electoral preferences in Mexico’s 2018 electoral process, particularly those regarding the national economy, given our focus on the federal election. In fact, evidence on the Mexican case points to the widespread prevalence of an economic vote among the electorate since the 1990s (^{Buendía,
1996}, ²⁰⁰⁰; ^{Poiré, 1999}; ^{Beltrán,
2003}, ²⁰¹⁵; ^{Singer, 2009}). However, considering that, in recent years, economic fluctuations coexist with non-economic phenomena such as criminal violence and corruption scandals —following the theory of issue publics (^{Krosnick, 1990})—, we consider it likely that Mexican voters also directed their attention to non-economic issues when defining their voting intentions, particularly toward those issues affecting them directly and that were relevant during the campaign. Based on these theoretical expectations, we propose to explore the following hypotheses. We first focus on electoral preferences regarding the candidate of the incumbent party:

H1a. The better the evaluation of economic performance, the greater the likelihood of supporting the ruling government’s candidate.
H1b. The better the evaluation of public safety performance, the greater the probability of supporting the ruling government’s candidate.
H1c. Direct experiences with crime and insecurity (victimization) are associated with diminished support for the ruling government’s candidate.
H1d. The better the evaluation of performance in corruption, the more likely it is to support the ruling government’s candidate.

In a complementary manner, and focusing on the factors that could have influenced the victory of the winning opposition candidate in the 2018 election, we propose the following:

H2a. The better the evaluation of economic performance, the less likely it is to support the winning candidate of the opposing party.
H2b. The better the evaluation of public security performance, the less likely it is to support for the winning candidate of the opposing party.
H2c. The better the evaluation of performance in corruption, the less likely it is to support for the winning candidate of the opposing party.

Although the hypotheses presented here follow the logic of consolidated studies in the literature on electoral behavior, we reiterate the value of analyzing the impact of three topics of major theoretical and practical importance on electoral preferences within the same study, and thus to be able to evaluate their effect in a comparative fashion. As we explain in detail below, the data on which this study is based offers this possibility. Beyond this empirical contribution, we propose the identification of issue publics (^{Krosnick, 1990}) for each of these concerns, whose characteristics —as we argue— could enhance the relevance of each topic and its effects on electoral preferences.

First, with respect to the economy, we consider that, given the structure of the labor market in Mexico, the informal sector is the most vulnerable group due to the lack of access to health care, retirement savings plans, housing loans and childcare services, among others ( ^{Alba Vega and
Kruijt, 1995}, ^{Altamirano,
2019}). Therefore, we argue that informality conditions the effect of economic evaluations.

H.3a. Individual economic vulnerability conditions the effect of economic evaluations.

With regard to security, we propose that it is especially the direct victims of criminal violence who give greater weight to their security assessments when expressing their electoral preferences.

H3b. Direct experiences with crime condition the effect of security evaluations.

Finally, while it is difficult to identify a group that is particularly affected by corruption due to its wide dissemination within the Mexican political system, the existing evidence clearly points to political sophistication as a relevant individual characteristic that conditions the effect of corruption issues on electoral behavior (^{Weitz and Winters,
2017}). Politically sophisticated voters have the capacity to process information about acts of corruption and to incorporate it into the definition of their vote. Therefore, although there is not a specific issue public with respect to corruption, we propose the existence of a possible public that is particularly sensitive, attentive, and mobilized regarding the issue of corruption.

H3c. Individual information levels condition the effect of corruption evaluations.

Overall, we argue that performance evaluations of the economy, security, and corruption influenced the electoral preferences of the Mexican electorate in 2018, but with particular impact among voters in the informal sector, victims of crime and politically informed individuals, respectively. Based on these theoretical expectations, we provide a brief account of the 2018 presidential campaign and present our empirical analysis in the following sections.

The issues of the 2018 presidential election

The balance of Enrique Peña Nieto’s administration

Between 2012 and 2014, the Congress approved 11 structural reforms: fiscal, financial, energy, education, telecommunications, economic competition, transparency, labor, criminal justice, political-electoral, and a new legal protection (amparo) law. In general terms, the purpose of these initiatives was to accelerate the country’s economic growth and development. Thus, this set of reforms won former president Enrique Peña Nieto the cover of Time magazine, emphasizing his work toward “saving Mexico.” These initiatives, however, faced many problems in terms of implementation (^{Flores-Macías,
2016}; ^{Arroyo et
al., 2018}). The fiscal reform fell short of its tax collection objectives; the energy reform was affected by the decline in oil prices; the education reform faced enormous protests and consequent problems in the implementation of the proposed teachers’ evaluation; and the telecommunications reform lent itself to clientelistic practices, among other problms (^{Flores Macías, 2016}). More importantly, overall, poverty, inequality, and lack of social mobility continued to mark the Mexican economy (^{Arroyo et al., 2018}). Additionally, by 2017, gasoline was no longer subsidized, resulting in Peña Nieto’s lowest level of presidential approval during his administration.

In terms of security, president Peña Nieto’s administration had multiple challenges. In his first year of government, he faced the flourishing of self-defense groups in 13 of the country’s 32 states (^{Phillips, 2017}), thus generating non-governmental armed organizations that added to the complexity of violence in the country. Although Joaquín “El Chapo” Guzmán was captured in February 2014 —in parallel to the series of reforms that strengthened the president’s image—, the drug trafficker escaped a year later, in July 2015. His escape followed the biggest disaster of the Peña Nieto’s presidency: the disappearance of the 43 students from Ayotzinapa, with clear involvement of authorities at different levels of government, in collusion with organized crime. A clumsy and unfortunate handling of the events by Attorney General Jesús Murillo Karam further complicated this scenario when he infamously declared he had “had enough” and was “fed up” (ya me cansé) in front of the media and thousands of victims who had tirelessly searched for their relatives for years, along with the parents of the 43. In fact, during the Peñista administration, homicides rose to more than 100 000 and more than 21 000 people disappeared (^{Cacelin,
2018}), making it the most violent six-year period in recent history.

Following the failure of structural reforms and the escalation of violence, the president faced two major corruption scandals involving two of his closest associates: the first lady, Angélica Rivera, and the Secretary of Finance, Luis Videgaray. Several news reports revealed that both individuals acquired expensive real estate through Grupo Higa, a company that had benefited extensively from contracts with the federal government (^{Aristegui Noticias,
2014}; ^{Montes, 2014}). The president ordered an investigation into the matter, but also chose the person in charge, Virgilio Andrade Martínez. Andrade’s final report did not identify any conflict of interest. Thus, any commitment made by the president to the fight against corruption and impunity was publicly perceived as completely empty.

As a result of this series of scandals and problems, Enrique Peña Nieto’s presidential approval rating quickly plummeted from 56 per cent at the start of his administration in February 2013 to 26 per cent just before the presidential election in May 2018 (^{Buendía&Laredo, 2018}). Thus, not only did the PRI arrive to the election with a huge credibility deficit in the eyes of the electorate, but, given the performance of the federal government, economic problems, violence, and corruption were, to some extent, present in the minds of voters.

According to data from CIDE-CSES 2018 National Electoral Study, one month before the presidential election, 44 per cent of Mexicans considered that insecurity and other associated problems such as crime, homicides, and drug trafficking were the country’s main problem; while for 36 per cent, economic issues such as unemployment, inflation, and poverty were the main concern. Despite wide coverage and discussion of corruption, only 9 per cent of the population considered it to be the country’s main problem, although it is important to note that 82 per cent considered corruption to be widespread in the country. From ^{Krosnick’s (1990)} point of view, this would suggest that there was a public widely concerned with the issues of insecurity and economy, probably due to direct effects on personal and family welfare, while corruption, having more diffuse effects, might not have generated a public specifically mobilized by the issue.

Regardless of the priorities within the electorate, given the diverse agenda and scandals of the Peña Nieto administration, it is crucial to know how these different issues were addressed during the presidential campaign. It should be noted that, although in this study we do not intend to evaluate the effect of the campaigns,^⁴ knowing their contents is fundamental in order to identify the possible differentiation between candidates for each of the three issues we examine here and to better understand their impact on electoral preferences. In the following section, we briefly examine the candidates’ attention and proposals on the issues of economy, security, and corruption.

The 2018 presidential campaign

The 2018 electoral process was characterized, among other things, by three elements that, to some extent, distinguish it from previous processes: 1) the first PAN-PRD alliance for a presidential candidacy; 2) the participation of independent candidates; but, above all, 3) a new party (Morena) that not only competed in the presidential election, but also, despite its short history, led the polls from the beginning of the campaign. With respect to this last element, it is important to note that the electoral success of new parties in both recent and established democracies has been associated, to a large extent, with their ability to take advantage of problems of representation in the face of the emergence of new cleavages among the electorate (^{Harmel and Robertson, 1985}; ^{Kitschelt, 1988}), as well as of the poor performance of ruling authorities (^{Tavits, 2007}). Therefore —and given the low approval of Enrique Peña Nieto—, it becomes even more relevant to consider the proposals made by the different candidates, particularly Morena, in the 2018 campaign.

On the economic front, the leading candidate, Andrés Manuel López Obrador (Morena) offered to return to the stabilizing development model of the 1960s, refocusing priorities in current spending and favoring investment in scholarships for young people who neither study nor work, therefore placing “the poor first”. The Morena candidate also emphasized the need to revise the structural reforms implemented by the Peñista administration, particularly those related to energy and education. In contrast, José Antonio Meade (PRI) did not propose changes but rather a continuity of the Peñista administration’s economic model and reforms. Ricardo Anaya’s (PAN) economic proposal focused on establishing a universal basic income and raising the minimum wage, while independent candidate Jaime Rodríguez Calderón proposed eliminating the minimum wage.

Although security is an issue where candidates and voters converge on the importance and necessity of the fight against crime (a valence issue), this was also an issue of wide contrasts in terms of the candidates’ proposals. While Ricardo Anaya and José Antonio Meade insisted on maintaining the participation of the armed forces to fight organized crime, Andrés Manuel López Obrador’s emphasis was on addressing the root causes of insecurity: corruption and poverty. López Obrador’s proposal also included an amnesty law for those who participated in illegal activities out of necessity or forced by organized crime —e.g. peasants who grew poppies or teenagers who collaborated with drug trafficking gangs. López Obrador’s offer contrasted sharply with that of independent candidate Jaime Rodríguez Calderón, who, among other measures, proposed militarizing high schools to discipline teenagers, and cutting off criminals’ hands.

It is possible that the Morenista candidate’s differentiation in terms of his security proposals, focused on addressing the roots of violence and an amnesty law —in contrast to the PRI and PAN candidates’ bet on the continuity of the military strategy or the heavy-handed option of “El Bronco”— placed him in the position of being evaluated among voters on this particular issue. The expectation in this regard, however, is ambiguous. Despite offering a different proposal in terms of security, a large sector of victims rejected the idea of an amnesty (^{Barragán, 2018}), which became the subject of multiple debates and political ad spots (^{Reporte
Índigo, 2018}).

As far as corruption is concerned, it was the candidate for Morena in particular who focused much of his discourse on this issue, emphasizing the importance of removing privileges for high-level officials, the possibility of prosecuting the president and revoking his mandate through a referendum. In this regard, López Obrador tried to characterize himself as an outsider of the political system in order to make his commitment to the fight against corruption credible, and to position the issue in his campaign platform. At the same time, the PRI candidate focused on the presentation of seven different types of tax returns and official statements for public servants in order to inform of their personal and family wealth. The PAN candidate’s proposal highlighted his insistence on the elimination of the use of cash in government operations. Finally, “El Bronco” captured the media attention with his proposal to cut off the hands of officials involved in embezzlement of public funds.

It should be noted that, despite their proposals to fight corruption, the four candidates were confronted personally for being involved in scandals of greater or lesser scope throughout the campaign or during their periods of public service. Ricardo Anaya was accused of money laundering. José Antonio Meade was questioned for diversion of public resources in the Ministry of Social Development (Sedesol) during his tenure as head of the agency. López Obrador formed alliances with people who were controversial because of their history of fraud accusations, such as Napoleón Gómez Urrutia. Jaime Rodríguez Calderón was accused by the electoral authority of forging signatures to obtain his candidacy.

Thus, the perception of high and widespread corruption and allegations of corrupt practices among the four candidates may have diminished the weight of this problem (^{Pavão, 2018}), in addition to making it difficult to differentiate between candidates, which is crucial for a specific issue to become important in an election (^{Krosnick, 1990}). However, the leading candidate’s emphasis on the issue of corruption may have made it relevant to the voters’ decision or at least to the evaluation of Morena as a viable electoral alternative.

The balance

In the 2006 election, the campaign focused mainly on the characterization of López Obrador as a danger to Mexico. In 2012, contrary to expectations, the issue of security was not a prominent theme of the campaign, despite the escalation of violence throughout the previous administration. However, in 2018, both issues —the perceived “threat” of López Obrador and insecurity— were present, along with broader debates about the economy and corruption. Moreover, as surveys show, the economy, security and corruption were somehow present in the minds of the voters. And whether these public concerns shaped the candidates’ references to them or, on the contrary, the candidates encouraged voters to think about these issues, the fact is that all three issues were present from the beginning to the end of the election campaign, with significant contrasts among the four candidates. It is therefore crucial to understand the extent to which perceptions of the economy, security and corruption played a role in the electoral preferences of the electorate.

It is also worth noting that it was particularly the candidate for Morena —as part of a new party in the political arena at the federal level— who offered the greatest contrast of proposals on the three issues reviewed, seeking to address specific groups of the electorate who might have felt relegated or underrepresented under the government in office. This behavior is in line with the expectations of the literature on new parties, which tend to address new cleavages among the electorate and have a greater chance of electoral success in a context of poor or deficient government performance (^{Tavits, 2007}). Thus, our empirical analysis also seeks to examine the extent to which the different dimensions of disapproval or dissatisfaction of Mexican voters, given the performance of the Peña Nieto administration, contributed to the electoral victory of Andrés Manuel López Obrador.

Research design

To study the effect of campaign issues on electoral preferences, we used individual data from the CIDE-CSES 2018 National Electoral Study . This project consisted of panel surveys conducted at three points in the 2018 federal election process: two pre-election surveys and one post-election survey.^⁵ The survey we used corresponds to the post-electoral period and is representative of the Mexican population at a national level.^⁶ In addition to information on the electoral decision of the voters, this wave includes questions that refer to the evaluation of the situation of the economy, security, and corruption, the three thematic dimensions that we address in our argument. Therefore, as we have previously mentioned, this survey has unique characteristics that allow us to evaluate the impact of three crucial issues —from a theoretical and public policy standpoint— on electoral preferences in the most objective and comparative way.

We conducted our analysis in two stages. First, we used a multinomial model to estimate the effect of issue evaluations on electoral preferences with respect to the ruling party candidate. This analysis allows us to compare the differentiated weight of each of the issues in voters’ decisions to change or keep the ruling party in the government and the relative gains of the opposition parties in each of the thematic agendas. The second set of models deepens the analysis of the determinants of the vote for the candidate representing Morena to explore the specific effect of perceptions on the economy in comparison with the issues highlighted in his campaign: corruption and security.

Dependent variable: Electoral preferences

Consistent with previous studies, we measure citizens’ electoral preference with a question that asks respondents to indicate the party or candidate for which they voted in the last presidential election. This question is posed with a paper ballot similar to the one used on Election Day. Interviewees mark their choice and place the ballot in a box, so that the interviewer cannot immediately see which option they chose. Compared to other possible question formats, this option increases the likelihood that respondents will genuinely indicate how they voted. On the basis of this question, we first generate a variable that takes different values according to the interviewee’s electoral preference: 1 if they voted for PAN/Ricardo Anaya; 2 if they voted for PRI/José Antonio Meade; 3 if they voted for Morena/Andrés Manuel López Obrador, and 4 if they voted for the independent candidate Jaime Ramírez “El Bronco”. This categorical variable is the dependent variable in our multinomial model. For the logistic models of electoral preference for the winning candidate, we generate a new dependent variable from the previous one, which takes a value of 1 if the interviewee voted for Morena/Andrés Manuel López Obrador and 0 if he voted for any other party or candidate.

Independent variables: Evaluation of the economy, security, and corruption

To operationalize citizens’ perceptions on key issues in the electoral process, we use evaluations of the national situation and, if available, of the personal situation or the voter’s direct experience. Higher values of these variables indicate more negative evaluations. In the case of questions related to the economic situation, we used two standard questions for retrospective evaluation. First, the sociotropic evaluation (“Would you say that during the last 12 months the economic situation of Mexico...?”), with a response scale ranging from 1 (improved) to 4 (worsened); and, second, the egotropic evaluation (“Would you say that during the last 12 months your personal economic situation...?”), with a response scale ranging from 1 (improved considerably) to 5 (worsened considerably).^⁷ To measure the effect of economic vulnerability associated with the informal sector, we constructed a variable that takes a value of 1 if the interviewee is a beneficiary of either of the two major public health care systems (IMSS or the ISSSTE) and 0 if he or she does not have access to the services provided by these institutions. Given the structure of the labor market in Mexico, eligibility for either social security institution indicates membership in the formal sector, which is associated with a series of benefits such as health care, retirement savings plans, housing loans, and childcare services, among others (^{Ghai, 2003}; ^{Benería and Floro, 2006}). Therefore, the economic vulnerability is greater among the population that lacks access to social security services.

For the issue of security, we include the national retrospective evaluation (“Would you say that, during the last twelve months, public security in the country...?”) with a response scale ranging from 1 (improved considerably) to 5 (worsened considerably). To assess the effect of personal experiences with crime, we use the following question: “Please tell me if you or a family member or friend have been a direct victim of any of these crimes in the last 12 months” Affirmative responses take a value of 1 and a value of 0 is assigned if no victimization experience is reported in the last year.

To capture perceptions regarding the issue of corruption, we used a question that asks about the situation of this problem in the country: “With respect to the previous six years, do you think that corruption in Mexico has decreased or increased?” Possible responses follow a scale of 1 (decreased) to 5 (increased). In the set of models that deepen the analysis of the electoral preference for the candidate representing Morena, we include an item asking about the credibility of specific accusations in his case: Tell me, how true or false do you think it is that Andrés Manuel López Obrador got together with corrupt politicians like Elba Esther Gordillo and Napoleón Gómez Urrutia in exchange for support for his campaign? This variable takes the value of 1 if the interviewees believe that the accusation is true/ very true, and 0 if they consider it to be false.^⁸

Finally, we constructed a variable that takes the value of 1 if the interviewee considers the accusations of corruption involving Ricardo Anaya and José Antonio Meade to be credible (true/very true), and 0 if the accusations do not seem credible in either case.^⁹ With this variable we seek to capture the effect of a widespread perception of corruption among the candidates who competed against Andrés Manuel López Obrador. Specifically, we are interested in finding out if this perception conditioned a possible negative electoral impact of the accusations against the leading candidate.

Controls

Previous studies show the relevance of several individual political characteristics in order to understand the logic of the voters’ electoral decisions. One of the main determinants of voting is party identification. In the multinomial models we include dichotomous variables to indicate the identification with each of the main parties: PAN, PRD, PRI and Morena (with a value of 1 if the interviewee identifies with the party and 0 if not). In addition to party identification, the ideological orientation of the voters can have an independent effect on their electoral decision. To measure this dimension, we include the position of the interviewee on the liberal-conservative scale (with a value of 0 if the interviewee identifies with the left and 10 if he or she identifies with the right).

We also control for a set of sociodemographic variables. We include the gender of the interviewee and his/her age. The variable of years of formal education can approximate the effect of the level of information and political sophistication of the interviewees.^¹⁰Finally, we include an index that captures the level of political awareness of the interviewee. This is an additive index based on three questions regarding knowledge of the Mexican political system.^¹¹ Table A1 in the Appendix shows the descriptive statistics of the variables included in the models.

Results

Table 1 presents the results of the multilevel model of electoral preferences. The reference category is the vote for the ruling party, PRI/José Antonio Meade. The estimate shows that a negative evaluation of the economy increased the preference for Ricardo Anaya and Andrés Manuel López Obrador, compared to the ruling party’s candidate. The negative perception of the personal economic situation, however, only favored López Obrador. In this model, the negative evaluation of corruption had a positive effect on electoral support for Ricardo Anaya and López Obrador, compared to the candidate of the PRI.

Table 1 Multinomial model: Campaign and voting issues regarding the governing party in 2018

	PAN/Anaya Vote	Morena/AMLO Vote	Independent/Bronco Vote
National economy evaluation	0.405** (3.18)	0.655*** (5.55)	0.372 (1.26)
Personal economy evaluation	-0.0214 (-0.21)	0.246** (2.61)	-0.291 (-1.28)
National security evaluation	0.0783 (0.74)	0.251** (2.59)	0.254 (1.00)
Victim	0.193 (0.78)	-0.142 (-0.61)	1.415** -2.89
National corruption evaluation	0.263* (2.46)	0.216* (2.24)	-0.0813 (-0.32)
PAN identification	0.503 (1.73)	-0.291 (-1.01)	-0.889 (-1.28)
PRD identification	0.627 (1.19)	0.085 (0.17)	-14.46 (-0.01)
Morena identification	-0.283 (-1.02)	0.0819 (0.34)	-1.002 (-1.64)
PRI identification	-1.068*** (-3.56)	-1.345*** (-4.93)	-15.77 (-0.02)
Liberal-Conservative	0.023 (0.54)	-0.0763* (-1.98)	-0.136 (-1.54)
Female	-0.33 (-1.58)	-0.458* (-2.39)	-0.477 (-1.01)
Age	-0.0151* (-2.02)	-0.0135* (-2.01)	-0.0423* (-2.30)
Education	0.146 (1.24)	-0.00681 (-0.06)	-0.177 (-0.62)
Constant	-1.672 (-1.96)	-2.112** (-2.69)	0.702 (-0.36)
Number of observations Pseudo R²		966 0.12

Source: Authors’ elaboration based on CIDE-CSES 2018 National Electoral Study (^{Beltrán, Ley and Castro Cornejo, 2020}). Note: Multinomial model with PRI/Meade vote as reference category. The variables of party identification and personal economic situation were included in wave 1. *p < 0.05, **p < 0.01, ***p < 0.001, t-statistics in brackets.

The security evaluation is not significant in the comparison between José Antonio Meade and Ricardo Anaya, but it is significant in the decision to vote for López Obrador instead of supporting the ruling party.

Direct experiences with crime did not have a significant effect in the case of Ricardo Anaya and Andrés Manuel López Obrador, but they did have an effect in the comparison with independent candidate Jaime Rodríguez. Finally, as expected, voters who identified with the PRI were less likely to vote for the candidates of the PAN and Morena-led coalitions.

The results of this first multinomial model point to a differentiated effect of the issues that were at the center of the 2018 federal election campaign. The first finding is that perceptions about the state of the economy were a central dimension in voters’ decisions.

Dissatisfaction with the state of the country’s economy favored candidates from opposition parties (H1a). The worsening of the personal situation translated into specific support for the candidate representing Morena, taking the ruling party as a reference. The deterioration of the security situation also had an uneven effect on the relative loss of support for the PRI (H1b).

The negative evaluations of national security favored the candidate for Morena, but not that of the PAN. When compared to the ruling party’s candidate, those with direct experience of crime tended to favor the independent candidate. It is possible that this last finding can be explained by the iron fist proposal of “El Bronco”, which would resonate with recent work such as that of ^{Visconti (2019)}, who finds that victims are more likely to support this type of security policies. Finally, voters who perceived a worsening of corruption did favor candidates Anaya and López Obrador over candidate Meade (H1d).

The second set of models deepens the analysis of the factors underlying the specific vote for the candidate representing Morena. Table 2 presents the logistical estimates of the electoral preference for Andrés Manuel López Obrador. Model 1 considers the effects of the independent variables included in the multinomial specification. Model 2 includes the additional variable of credibility of the accusations regarding possible links between the candidate for Morena and characters identified in corruption scandals. Model 3 introduces the perception of corruption allegations against candidates opposing López Obrador and an interactive term for the two variables (credibility of AMLO’S corruption allegations and credibility of Anaya/Meade’s corruption allegations).

Table 2 Logit model: Campaign issues and voting for the candidate representing Morena

	Modelo 1	Modelo 2	Modelo 3
National economy	0.442***	0.472***	0.380**
evaluation	(4.49)	(4.25)	(3.25)
Personal economy	0.322***	0.370***	0.326**
evaluation	(3.80)	(4.01)	(3.26)
National security	0.210**	0.249**	0.281**
evaluation	(3.12)	(3.03)	(2.64)
Victim	-0.346*	-0.320*	-0.250
	(-2.56)	(-2.35)	(-1.53)
National corruption	0.0627	0.106	-0.0261
evaluation	(0.74)	(1.19)	(-0.25)
Corruption AMLO		-1.309***	-0.552**
		(-6.12)	(-2.68)
Corruption opposition			1.843***
(Anaya and Meade)			(6.59)
Corruption AMLO x			-1.546***
Corruption opposition			(-3.31)
Morena identification	0.550**	0.508*	0.420
Liberal-Conservative	(2.58) -0.0743*	(2.23) -0.0802*	(1.70) -0.0509
	(-2.30)	(-2.53)	(-1.46)
Female	-0.229	-0.323	-0.270
	(-1.62)	(-1.86)	(-1.50)
Age	-0.00283	-0.00464	-0.00378
	(-0.72)	(-0.96)	(-0.74)
Education	-0.0625	-0.0788	-0.000623
	(-0.74)	(-0.84)	(-0.01)
Constant	-2.801*** (-4.38)	-2.670*** (-3.81)	-2.987*** (-4.09)
Number of observations	966	865	804
Pseudo R2	0.09	0.17	0.23

The results confirm the centrality of the economic dimension in the electoral support of the candidate for Morena. Individuals with negative evaluations of the national economic situation were more likely to express that they voted for López Obrador, as are those who perceived a deterioration in their personal economic situation —according to our expectations under H2a.^¹² The effect of both evaluations is significant and robust to alternative specifications (see Tables A4 and A5 in the Appendix). Table A6 reports similar results controlling for the (dis)agreement with President Peña Nieto’s administration.

Figure 1 reports the estimated probabilities of voting for the candidate representing Morena for different levels of evaluation of the national economic situation (Panel a) and the personal economic situation (Panel b), keeping the rest of the variables at their average values. The panel on the left (a) shows that the main effect occurs between the “equally good ” category and the most negative evaluation: “worsened ”. The intervals corresponding to the most positive evaluation (“improved ”) are larger and overlap with the intervals estimated for the other evaluations. It should be noted that the most positive category has a small number of cases, as the observations on this variable are concentrated at the most negative end.

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

Figure 1 Effect of the sociotropic and egotropic economic evaluations on the voting probability for the candidate representing Morena

The panel on the right (b) shows the probabilities for the evaluations of personal economic status. The perception of deterioration in the personal situation is clearly associated with increased electoral support for the candidate for Morena. The effects of the more negative evaluations on the probability of voting for López Obrador are significantly higher than those corresponding to the more positive evaluations.

Regarding the security dimension, the findings are mixed. The negative evaluation of the national security situation had a positive effect on electoral support for the candidate representing Morena —consistent with H.2b. The effect of direct experiences with crime is less evident. In the first two specifications the victimization variable was negatively associated with voting for AMLO-Morena (p < 0.05). However, this result is sensitive to alternative specifications, as Model 3 shows.

Figure 2 shows the estimated probabilities of voting for the candidate representing Morena according to the perception of the security issue and the victimization experience. Panel (a) presents the probabilities corresponding to the different assessments of the security situation at the national level. The figure indicates that there are no major differences in the electoral support of citizens who positively or negatively assessed the country’s security situation at the extremes of the scale. However, the probabilities clearly differ between those who considered that it “improved somewhat” and “worsened considerably”. Panel (b) shows that while respondents who have been victims of crime are less likely to support the candidate for Morena than those who have not had direct experience with crime, this effect is less robust than the contextual assessment. The figure shows an overlap in the confidence intervals of the estimated probabilities (AMLO/Morena vote) for victims and non-victims.

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

Figure 2 Effect of national security evaluations and personal experience with crime on the voting probability for the candidate representing Morena

The results on the issue of insecurity suggest, first, that sociotropic evaluations of security and personal experiences with crime can have differentiated effects on electoral preferences. This finding is consistent with previous works that analyzed their effects on the 2012 election (^{Romero et al., 2016}; ^{Ley, 2017}). However, in the specific case of López Obrador’s campaign, it is possible that the campaign discussion about a possible amnesty had a counterproductive effect among voters who had already been personally affected by violence. This result resonates with the rejection of a potential amnesty previously expressed by victims’ organizations during the presidential campaign (^{Barragán,
2018}).

The evaluation on national corruption is not significant in the level of support for the candidate representing Morena —contrary to the expectation in H2c. Although dissatisfaction with the worsening of corruption was a component of the rejection of the ruling party, it does not appear to have been the central issue that mobilized a specific electorate in favor of López Obrador. Since it had a diffuse effect, the problem of corruption did not generate a clearly defined issue public (^{Krosnic, 1990}). Despite being a central theme in the winning candidate’s discourse, the dissatisfaction and indignation regarding this issue does not seem to have been decisive in the individual considerations of the voters who elected him.

However, Model 2 shows that the perception of the existence of possible links between López Obrador and people accused of corruption did affect the electoral support for the candidate representing Morena. This means that the credibility of specific accusations seems to have come at a cost to the leading candidate. In order to identify the prevalence of such an effect in the face of the multiple accusations of corruption against the other candidates, Model 3 introduces the interactive term between the perception of AMLO’S corruption and the credibility of the corruption accusations against the PRI and PAN candidates.

Figure 3 shows the interactive effect of both variables, which is significant in the model. Those who consider the accusations against the candidate for Morena to be false (value of 0 on the horizontal axis) are much more likely to vote for him when the accusations against his opponents seem credible to them (dotted line).^¹³ However, the probability of voting for AMLO decreases significantly when the accusations against him gain credibility (value 1), especially for those who consider that their opponents are also corrupt —that is, when there is a perception of widespread corruption. Finally, most of the political and sociodemographic variables are not significant in these models. As expected, those interviewed who self-identified as morenistas voted for their party.

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

Figure 3 Effects of the credibility of the accusations against PRI and PAN candidates, according to the credibility of the accusations against the candidate for Morena

In order to explore in detail the conformation of electoral preferences and to identify a possible conditional effect of the sociotropic and egotropic evaluations according to the direct experiences with each topic and/or the individual characteristics of the interviewees (H3a-H3c), we present a series of interactive models.

These examine the joint effect of evaluations of the economy, security, and corruption according to the individuals’ insertion in the informal market, their experience of victimization, and their level of information, respectively. The logic of these statistical tests is the identification of issue sectors or publics that have been mobilized to a greater extent due to characteristics that make them more sensitive to each topic, either because of direct effects on the specific topics of economy (economic vulnerability) or security (victimization) (^{Krosnick, 1990}), or because of their greater knowledge and information regarding the country’s situation (^{Gomez and Wilson, 2001}) —for which we use the political information index described above. Table 3 shows the results of this set of models.

Table 3 Logit model: Campaign issues and voting for the candidate representing Morena. Interactive models

Model 1		Model 2	Model 3	Model 4
National economy evaluation	0.431*** (4.36)	0.357* (2.17)	0.448*** (4.62)	0.442*** (4.51)
Personal economy evaluation	0.591*** (4.67)	0.327*** (3.86)	0.320*** (3.80)	0.319*** (3.68)
National security evaluation	0.219** (3.25)	0.212** (3.09)	0.180* (2.32)	0.211** (3.14)
Victim	-0.364** (-2.64)	-0.344* (-2.56)	-0.828 (-1.17)	-0.339* (-2.42)
National corruption evaluation	0.0727 (0.83)	0.0715 (0.82)	0.0599 (0.70)	0.0512 (0.29)
Index of political information (IPI)				-0.0537 (-0.19)
Informal sector	1.833** (3.04)	-0.295 (-0.49)
Informal * Personal economy evaluation	-0.431** (-2.95)
Informal * National economy evaluation		0.130 (0.76)
Victim * National security evaluation			0.131 (0.74)
IPI National * corruption evaluation				0.00415 (0.06)
Morena identification	0.557* (2.57)	0.556** (2.58)	0.555** (2.62)	0.548** (2.59)
Female	-0.227 (-1.61)	-0.224 (-1.59)	-0.218 (-1.51)	-0.228 (-1.61)
Age	-0.00292 (-0.72)	-0.00269 (-0.67)	-0.00276 (-0.70)	-0.00267 (-0.66)
Liberal-Conservative	-0.0748* (-2.31)	-0.0w748* (-2.32)	-0.0743* (-2.30)	-0.0731* (-2.27)
Education	-0.0635 (-0.78)	-0.0591 (-0.71)	-0.0630 (-0.75)	-0.0572 (-0.66)
Constant	-3.977*** (-5.36)	-2.678*** (-3.56)	-2.706*** (-4.15)	-2.662* (-2.52)
Number of observations Pseudo R²	966 0.10	966 0.10	966 0.10	966 0.10

Models 1 and 2 study the effect of evaluations of national and personal economic conditions in terms of their interaction with the economic vulnerability associated with informality. Figure 4 shows the estimated voting probabilities for López Obrador for both interactive terms. Economic vulnerability does not seem to condition the effect of the evaluations of the national economic situation (Panel a). However, consistent with our expectation under H3a, the interactive term between individual economic vulnerability and personal economic evaluation is significant.

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

Figure 4 Effects of economic vulnerability at different levels of economic evaluation

Panel (b) in Figure 4 shows that, in the range of positive or neutral assessments of personal economic status, individuals in the formal sector are less likely to express their electoral support for the candidate representing Morena as compared to those in the informal sector. However, as the perception of personal economic well- being deteriorates, the intervals overlap between individuals in both sectors and support for López Obrador increases significantly for those in the formal sector as well. This finding provides additional evidence on the economic micro-foundations of support for the candidate representing Morena. For voters, the perception of a decline in their personal well-being was decisive in their support for the leading candidate. The results suggest that the perceived decline in the personal economic situation implied a convergence between the preferences of voters in the formal sector and those of the most economically vulnerable one (the informal sector). Previous studies have explored the differentiated weight of economic evaluations on the electoral preferences of voters in the formal/informal sector in Argentina (^{Singer, 2016}). Consistently, the results for the federal election in Mexico in 2018 suggest that economic vulnerability conditioned the effect of pocketbook evaluations of the economy. However, contrary to the Argentine case, Mexican voters in the informal sector who made positive assessments of their personal situation did not favor the ruling party.^¹⁴ Support for the candidate representing Morena tended to be greater among the most vulnerable voters across the scale and increased slightly as the egotropic evaluation worsened. In contrast, the preferences of formal voters were particularly sensitive to the deterioration in their personal economic situation, which benefited candidate López Obrador.

Model 3 explores the combined effect of victimization and the assessment of the national security situation. The interactive term is not significant, indicating that direct experiences with crime did not necessarily condition the effect of evaluations of insecurity on the probability of voting for the candidate representing Morena —contrary to our hypothesis H.3b.

To find out whether the assessment of corruption had a differentiated effect on voters according to their level of information, Model 4 tests the effect of an interactive term between political information and the evaluation of the situation of corruption. The model includes the set of control variables incorporated in the previous models. Contrary to our expectations under H.3c, the results do not show a significant interactive effect of the information and the socio-political evaluation of this dimension. Hence, there is no evidence that the most informed voters were particularly sensitive to the issue of corruption when they cast their vote.

Discussion and conclusions

The 2018 presidential election in Mexico marked the arrival of a new party in office. This historic party alternation necessarily requires a clear and timely explanation of the elements that allowed the triumph of a candidate who ran for president for the third time. In this article we have focused on an explanation based on the public policy elements that were present in the previous administration and that could have impacted López Obrador’s electoral success on this occasion.

Based on the extensive literature on issue voting, we explore three themes that permeated the political agenda prior to the presidential election: the economy, security, and corruption. We found that the economy was a particularly important factor in the definition of electoral preferences in favor of the candidate for Morena. Certainly, economic reforms were among the most controversial issues of Enrique Peña Nieto’s administration, and our findings suggest that, given the high expectations for such changes along with the poor economic performance, AMLO benefited the most from these negative economic evaluations.

As previous work has documented, it is particularly difficult for the electorate to exercise consistent accountability on the dimension of security. The results presented here indicate that López Obrador was also favored by those who had negative evaluations of the national security, but victims —precisely the group less favored by the country’s security situation— did not show an equally consistent support for the winning candidate. As we noted, one possibility is that López Obrador’s amnesty proposals may have backfired on his campaign, at least among victims.

In this study we also found that general evaluations of the state of corruption at the national level did not have a direct effect on electoral preference for López Obrador. Consistent with studies on the electoral effect of corruption, our results confirm the difficulty of voters in using this dimension in the final definition of their electoral preferences, which can have negative effects on electoral accountability. In fact, our extended analysis showed that the possibility of punishing candidates for an allegation of corruption depends both on the perceived credibility of such observations, and on their comparison with the rest of the candidates —even after controlling for party identification. This finding resonates with the work of ^{Botero
et al. (2015)}, reiterating the importance of understanding the logic of such notions of credibility, particularly in a context where corruption scandals prevail across political fronts, regardless of the ideological spectrum.

Thus, in general terms, it is possible to say that the electoral victory of president Andrés Manuel López Obrador is largely the result of the dissatisfaction of an electorate in matters regarding economy and security, although it is likely that some of his proposals in the second area have alienated or divided the sector most affected by insecurity: the victims. And, although the candidate for Morena tried to capitalize on the issue of corruption and make it a banner of his campaign, the reality is that this was not an issue that permeated the final definition of electoral preferences. The reasons behind this may be many: the complexity of the issue, the wide prevalence of corruption throughout the political system, and the difficulty of evaluating the information and accusations regarding corruption, among others.

Beyond the ability to explain an important dimension of a historical electoral result in Mexico, this study contributes to existing analyses of thematic voting, incorporating three issues that have been widely discussed in the literature on electoral behavior, but that can rarely be studied together and in a comparative manner. Moreover, our work expands the traditional approach of studies on issue voting, identifying the different publics particularly mobilized around each topic. Our findings also point to the importance of deepening certain aspects of the non-economic vote on which political science still needs greater understanding, such as the political behavior of victims; the conditions under which victims demand accountability from government authorities on the specific issue of security; and the difficulty with which problems which are both generalized, as well as complicated in terms of political attribution —such as corruption— can become a determining factor in an election.

REFERENCES

Aidt, Toke S. (2000), “Economic Voting and Information”, Electoral Studies, 19(2), pp. 349-362. [ Links ]

Alba Vega, Carlos y Dirk Kruijt (1995), “El significado del sector informal y la microempresa en América Latina y en México”, en Thomas Calvo y Bernardo Méndez (eds.), Micro y pequeña empresa en México: Frente a los retos de la globalización, Ciudad de México: Centro de Estudios Mexicanos y Centroamericanos. [ Links ]

Altamirano, Melina (2019), “Economic Vulnerability and Partisanship in Latin America”, Latin American Politics and Society, 61(3), pp. 80-103. [ Links ]

Altheide, David L. (2002), Creating Fear: News and the Construction of Crisis, Nueva York: Aldine de Gruyter. [ Links ]

Anderson, Christopher (2000), “Economic Voting and Political Context: A Comparative Perspective”, Electoral Studies, 19(2-3), pp. 151-170. [ Links ]

Anderson, Christopher (2007), “The End of Economic Voting? Contingency Dilemmas and Limits of Democratic Accountability”, Annual Review of Political Science, 10, pp. 271-296. [ Links ]

Arceneaux, Kevin (2006), “The Federal Face of Voting: Are Elected Officials Held Accountable for the Functions Relevant to Their Office?”, Political Psychology, 27(5), pp. 731-754. [ Links ]

Arceneaux, Kevin y Robert M. Stein (2006), “Who is Held Responsible when Disaster Strikes? The Attribution of Responsibility for a Natural Disaster in an Urban Election”, Journal of Urban Affairs, 28(1), pp. 43-53. [ Links ]

Aristegui Noticias (2014), “La casa blanca de Enrique Peña Nieto (investigación especial)”, Aristegui Noticias, 9 de noviembre, disponible en: https://aristeguinoticias.com/0911/mexico/la-casa-blanca-de-enrique-pena-nieto/ [fecha de consulta: 13 de marzo de 2019]. [ Links ]

Arroyo Ortiz, Juan Pablo, Isabel Islas Arredondo, Ariadna Díaz Castillo y María Ángeles Cortés Basurto (eds.) (2018), Balance de las reformas estructurales, t. I, Ciudad de México: Instituto Belisario Domínguez. [ Links ]

Ashby, Nathan J. y Miguel A. Ramos (2013), “Foreign Direct Investment and Industry Response to Organized Crime: The Mexican Case”, European Journal of Political Economy 30, pp. 80-91. [ Links ]

Bali, Valentina A. (2007), “Terror and Elections: Lessons from Spain”, Electoral Studies, 26(3), pp. 669-687. [ Links ]

Barragán, Sebastián (2018), “‘Perdón después de la justicia’, reiteran víctimas a López Obrador”, Aristegui Noticias, 14 de septiembre, disponible en: https://aristeguinoticias.com/1409/lomasdestacado/perdon-despues-de-la-justicia-reiteran-victimas-a-lopez- obrador/ [fecha de consulta: 25 de marzo de 2019]. [ Links ]

Becerra, Lorena (2017), “Da ventaja a Morena desaprobación de EPN”, Reforma, 18 de enero, disponible en: https://gruporeforma.reforma.com/interactivo/encuestas/enc_13a_epn/#continuar [fecha de consulta: 11 de noviembre de 2018]. [ Links ]

Beltrán, Ulises (2003), “Venciendo la incertidumbre: El voto retrospectivo en la elección presidencial de julio de 2000”, Política y Gobierno, X(2), pp. 325-358. [ Links ]

Beltrán, Ulises (2015), “Percepciones económicas retrospectivas y voto por el partido en el poder, 1994-2012”, Política y Gobierno, XXII(1), pp. 45-85. [ Links ]

Beltrán, Ulises, Sandra Ley y Rodrigo Castro Cornejo (2020), Estudio Nacional Electoral de México (CIDE-CSES) 2018, México: Centro de Investigación y Docencia Económicas. [ Links ]

Benería, Lourdes y Maria S. Floro (2006), “Labour Market Informalization, Gender and Social Protection: Reflections on Poor Urban Households in Bolivia and Ecuador”, en Shahra Razavi y Shireen Hassim (eds.), Gender and Social Policy in a Global Context: Unco- vering the Gendered Structure of the Social, Londres: Palgrave Macmillan, pp. 193-216. [ Links ]

Berrebi, Claude y Esteban F. Klor (2008), “Are Voters Sensitive to Terrorism? Direct Evidence from the Israeli Electorate”, American Political Science Review, 102(3), pp. 279-301. [ Links ]

Boas, Taylor C., F. Daniel Hidalgo y Marcus André Melo (2018), “Norms versus Action: Why Voters Fail to Sanction Malfeasance in Brazil”, American Journal of Political Science, 63(2), pp. 385-400, doi: https://doi.org/10.1111/ajps.12413. [ Links ]

Botero, Sandra, Rodrigo Castro Cornejo , Laura Gamboa, Nara Pavão y David W. Nickerson (2015), “Says Who? An Experiment on Allegations of Corruption and Credibility of Sources”, Political Research Quarterly, 68(3), pp. 493-504. [ Links ]

Brown, Ryan (2018), “The Mexican Drug War and Early-life Health: The Impact of Vio- lent Crime on Birth Outcomes”, Demography, 55(1), pp. 319-340. [ Links ]

Brown, Ryan y Andrea Velásquez (2017), “The Effect of Violent Crime on the Human Capital Accumulation of Young Adults”, Journal of Development Economics, 127, pp. 1-12. [ Links ]

Buendía, Jorge (1996), “Economic Reform, Public Opinion, and Presidential Approval in Mexico, 1988-1993”, Comparative Political Studies, 29(5), 566-591. [ Links ]

Buendía, Jorge (2000), “El elector mexicano en los noventa: ¿Un nuevo tipo de votante?” Política y Gobierno, VII(2), pp. 317-352. [ Links ]

Buendía&Laredo (2018), “Aprobación presidencial”, disponible en: http://www.buendia-ylaredo.com/publicaciones/444/APROBACION.pdf [fecha de consulta: 11 de noviembre de 2018]. [ Links ]

Cacelin, Janet (2018), “Cifras récord de asesinatos, desapariciones e impunidad: El balance que deja Peña Nieto al terminar mandato”, Univisión, 29 de noviembre, disponible en: https://www.univision.com/noticias/america-latina/cifras-record-de-asesinatos-desapariciones-e-impunidad-el-balance-que-deja-pena-nieto-al-terminar-mandato [fecha de consulta: 18 de noviembre de 2018]. [ Links ]

Caudillo, Mónica L. y Florencia Torche (2014), “Exposure to Local Homicides and Early Educational Achievement in Mexico”, Sociology of Education, 87(2), pp. 89-105. [ Links ]

Carreras, Miguel (2013), “The Impact of Criminal Violence on Regime Legitimacy in La- tin America”, Latin American Research Review, 48(3), pp. 85-107. [ Links ]

Chang, Eric C.C., Miriam A. Golden ySeth J. Hill (2010), “Legislative Malfeasance and Political Accountability”, World Politics, 62(2), pp. 177-220. [ Links ]

Chiricos, Ted, Kathy Padgett yMarc Gertz (2000), “Fear, TV News, and the Reality of Crime”, Criminology, 38(3), pp. 755-786. [ Links ]

Chong, Alberto, Ana de la O, Dean Karlan y Leonard Wantchekon (2015), “Does Corruption Information Inspire the Fight or Quash the Hope? A Field Experiment in Mexico on Voter Turnout, Choice, and Party Identification”, Journal of Politics, 77(1), pp. 55-71. [ Links ]

Dorussen, Han y Harvey D. Palmer (2002), “The Context of Economic Voting: An Introduction”, en Han Dorussen y Michael Taylor (eds.), Economic Voting, Nueva York: Routledge, pp. 1-14. [ Links ]

Ferraz, Claudio y Frederico Finan (2008), “Exposing Corrupt Politicians: The Effects of Brazil’s Publicly Released Audits on Electoral Outcomes”, Quarterly Journal of Economics, 123(2), pp. 703-745. [ Links ]

Flores-Macías, Gustavo (2016), “Latin America’s New Turbulence: Mexico’s Stalled Reforms”, Journal of Democracy, 27(2), pp. 66-78. [ Links ]

Fiorina, Morris P. (1978), “Economic Retrospective Voting in American National Elections: A Microanalysis”, American Journal of Political Science, 22(2), pp. 426-443. [ Links ]

Gasper, John T. y Andrew Reeves (2011), “Make it Rain? Retrospection and the Attentive Electorate in the Context of Natural Disasters”, American Journal of Political Science, 55(2), pp. 340-355. [ Links ]

Gélineau, Francois y Karen L. Remmer (2006), “Political Decentralization and Electoral Accountability: The Argentine Experience, 1983-2001”, British Journal of Political Science, 36(1), pp. 133-157. [ Links ]

Gelpi, Christopher, Peter D. Feaver y Jason Reifler (2005), “Success Matters: Casualty Sensitivity and the War in Iraq”, International Security, 30(3), pp. 7-46. [ Links ]

Ghai, D. (2003), “Trabajo decente: Concepto e indicadores”, Revista Internacional del Trabajo, 122(2), pp. 125-160. [ Links ]

Gomez, Brad T. y J. Matthew Wilson (2001), “Political Sophistication and Economic Voting in the American electorate: A Theory of Heterogeneous Attribution”, American Journal of Political Science, 45(4), pp. 899-914. [ Links ]

Harmel, Robert y John D. Robertson (1985), “Formation and Success of New Parties: A Cross-national Analysis”, International Political Science Review, 6(4), pp. 501-523. [ Links ]

Hetherington, Marc J. (1996), “The Media’s Role in Forming Voters’ National Economic Evaluations in 1992”, American Journal of Political Science, 40(2), pp. 372-395. [ Links ]

Hobolt, Sara, James Tilley y Susan Banducci (2013), “Clarity of Responsibility: How Go- vernment Cohesion Conditions Performance Voting”, European Journal of Political Research, 52(2), pp. 164-187. [ Links ]

Holbrook, Thomas y James C. Garand (1996), “Homo Economus? Economic Information and Economic Voting”, Political Research Quarterly, 49(2), pp. 351-375. [ Links ]

Karol, David y Edward Miguel (2007), “The Electoral Cost of War: Iraq Casualties and the 2004 US Presidential Election”, The Journal of Politics, 69(3), pp. 633-648. [ Links ]

Kibris, Arzu (2011), “Funerals and Elections: The Effects of Terrorism on Voting Behavior in Turkey”, Journal of Conflict Resolution, 55(2), pp. 220-247. [ Links ]

Kitschelt, Herbert (1988), “Left-Libertarian Parties: Explaining Innovation in Competiti- ve Party Systems”, World Politics, 40(2), pp. 194-234, doi: 10.2307/2010362. [ Links ]

Krosnick, Jon A. (1990), “Government Policy and Citizen Passion: A Study of Issue Publics in Contemporary America”, Political Behavior, 12(1), 59-92. [ Links ]

Ley, Sandra (2017), “Security and Crime Issue Voting: Electoral Accountability in the Midst of Violence”, Latin American Politics and Society, 59(1), pp. 3-27. [ Links ]

Ley, Sandra (2018), “To Vote or Not to Vote: How Criminal Violence Shapes Electoral Participation”, Journal of Conflict Resolution, 62(9), pp. 1963-1990. [ Links ]

Manzetti, Luigi y Guillermo Rosas (2015), “Corruption and the Latin American Voter”, en Ryan E. Carlin, Matthew M. Singer y Elizabeth J. Zechmeister (eds.), The Latin American Voter: Pursuing Representation and Accountability in Challenging Contexts, Ann Arbor: University of Michigan Press, pp. 313-356. [ Links ]

Mauro, Paolo (1995), “Corruption and Growth”, Quarterly Journal of Economics, 110(3), pp. 681-712. [ Links ]

Mauro, Paolo (1998), “Corruption and the Composition of Government Expenditure”, Journal of Public Economics, 69(2), pp. 263-279. [ Links ]

Mishler, William y Richard Rose (2001), “What are the Origins of Political Trust? Testing Institutional and Cultural Theories in Post-Communist Societies”, Comparative Political Studies, 34(1), pp. 30-62. [ Links ]

Montes, Juan (2014), “Mexico Finance Minister Bought House from Government Contractor”, Wall Street Journal, 11 de diciembre. [ Links ]

Pacek, Alexander C. yBenjamin Radcliff (1995), “Economic Voting and the Welfare State: A Cross-National Analysis”, Journal of Politics, 57(1), pp. 44-61. [ Links ]

Pavão, Nara (2018), “Corruption as the Only Option: The Limits to Electoral Accountability”, Journal of Politics, 80(3), pp. 996-1010. [ Links ]

Pérez, Orlando (2015), “The Impact of Crime on Voter Choice in Latin America”, en Ryan E. Carlin, Matthew M. Singer y Elizabeth J. Zechmeister (eds.), The Latin American Voter: Pursuing Representation and Accountability in Challenging Contexts, Ann Arbor: University of Michigan Press, pp. 337-358. [ Links ]

Phillips, Brian J. (2017), “Inequality and the Emergence of Vigilante Organizations: The Case of Mexican Autodefensas”, Comparative Political Studies, 50(10), pp. 1358-1389. [ Links ]

Poiré, Alejandro (1999), “Retrospective Voting, Partisanship, and Loyalty in Presidential Elections: 1994”, en Jorge Domínguez y Alejandro Poiré (eds.), Toward Mexico’s Democratization: Parties, Campaigns, Elections and Public Opinion, Nueva York: Routledge. [ Links ]

Powell, Bingham y Guy Whitten (1993), “A Cross-National Analysis of Economic Voting: Taking Account of the Political Context”, American Journal of Political Science, 37(2), pp. 391-414. [ Links ]

Reporte Índigo (2018), “Meade retrata ‘miedo’ de la sociedad a AMLO en nuevos spots”, Reporte Índigo, 9 de abril, disponible en: https://www.reporteindigo.com/reporte/meade-retrata-miedo-la-sociedad-a-amlo-en-nuevos-spots-videos/ [fecha de consulta: 30 de marzo de 2019]. [ Links ]

Robles, Gustavo, Gabriela Calderóny Beatriz Magaloni (2013), “The Economic Conse- quences of Drug-Trafficking Violence in Mexico”, documento de trabajo 42, Banco Interamericano de Desarrollo. [ Links ]

Romero, Vidal, Beatriz Magaloni y Alberto Díaz-Cayeros (2016), “Presidential Approval and Public Security in Mexico’s War on Crime”, Latin American Politics and Society, 58(2), pp. 100-123. [ Links ]

Rose-Ackerman, Susan (1999), Corruption and Government: Causes, Consequences and Reform, Nueva York: Cambridge University Press. [ Links ]

Seligson, Mitchell (2002), “The Impact of Corruption on Regime Legitimacy: A Comparative Study of Four Latin American Countries”, Journal of Politics, 64(2), pp. 408-433. [ Links ]

Singer, Matthew M. (2009), “‘Defendamos lo que hemos logrado’: El voto económico en México durante la elección presidencial de 2006”, Política y Gobierno, volumen temático, pp. 199-235. [ Links ]

Singer, Matthew M. (2011), “Who Says It’s the Economy? Cross-National and Cross-Indi- vidual Variation in the Salience of Economic Performance”, Comparative Political Studies, 44(3), pp. 284-312. [ Links ]

Singer, Matthew M. (2016), “Informal Sector Work and Evaluations of the Incumbent: The Electoral Effect of Vulnerability on Economic Voting”, Latin American Politics and Society, 58(2), pp. 49-73. [ Links ]

Tavits, Margit (2007), “Principle vs. Pragmatism: Policy Shifts and Political Competition”, American Journal of Political Science, 51(1), pp. 151-165. [ Links ]

Trejo, Guillermo y Sandra Ley (2019), “High-Profile Criminal Violence: Why Drug Cartels Murder Government Officials and Party Candidates in Mexico”, British Journal of Political Science, pp. 1-27, doi: 10.1017/S0007123418000637. [ Links ]

Visconti, Giancarlo (2019), “Policy Preferences after Crime Victimization: Panel and Survey Evidence from Latin America”, British Journal of Political Science, pp. 1-15, doi:https://doi.org/10.1017/S0007123418000297. [ Links ]

Wei, Shang-Jin (1999), “Corruption in Economic Development: Beneficial Grease, Minor Annoyance, or Major Obstacle?” documento de trabajo 2444, Washington D.C.: Banco Mundial. [ Links ]

Weitz-Shapiro, Rebecca y Matthew S. Winters (2017), “Can Citizens Discern? Information Credibility, Political Sophistication, and the Punishment of Corruption in Brazil”, Journal of Politics, 79(1), pp. 60-74. [ Links ]

¹Given the panel design of this survey, we took advantage of some of the data collected in different waves to complement information on some variables.

²Other non-economic issues that have generated research on their impact on electoral preferences include natural disasters: ^{Arceneaux and Stein (2006)}, ^{Gasper and Reeves (2011)}; terrorism:^{Bali (2007)}, ^{Berrebi and Klor (2008)}, ^{Kibris (2011)}; and war casualties: ^{Gelpi et a. (2005)}; ^{Karol and Miguel (2007)}, among others.

³On the socio-economic and political consequences of criminal activity, see ^{Ashby and Ramos (2013)}, ^{Robles et al. (2013)}, ^{Carreras (2013)}, ^{Caudillo and Torche (2014)}, ^{Brown and Velásquez (2017)}, ^{Brown (2018)}, ^{Ley (2018)}, and ^{Trejo
and Ley (2019)}. On the economic and political effects of corruption, see ^{Mauro (1995}, ¹⁹⁹⁸⁾, ^{Rose-Ackerman (1999)}, ^{Wei
(1999)}, ^{Mishler and Rose
(2001)}, and ^{Seligson
(2002)}.

⁴The analysis of the exposure and impact of the campaign messages is beyond the scope of our study. However, other articles in this special issue focus specifically on those effects.

⁵The two pre-election surveys were conducted in March and June 2018, respectively, and the post-election survey was conducted one week after the July 2018 election.

⁶The ENEM, 2018 post-electoral survey consisted on the application of 1 239 effective surveys to respondents over 18 years old and was conducted face-to-face in households across the national territory. The survey was carried out on the basis of a probability sample of electoral sections.

⁷The evaluation of the personal economic situation was not included in the third wave of the ENEM 2018 panel, so we used the question that was included in the first wave.

⁸The survey did not include questions on respondents’ direct experiences with corruption.

⁹The specific questions for both candidates are: 1) Tell me, how true or false do you think it is that José Antonio Meade covered up government corruption scandals? and 2) Tell me, how true or false do you think it is that Ricardo Anaya used his political career to get rich?

¹⁰While the education variable captures part of the socioeconomic status in Mexico, we ran a series of additional models with two measures that approximate the level of wealth of individuals. First, we added a self-reported income variable (which, by its nature, has a high level of non-response). In an other set of models, we added an index of socioeconomic level calculated from a battery of questions about goods and services available in the respondent’s home. The results of these specifications are reported in Tables A2-A5 in the Appendix.

¹¹The questions are 1) Can you please tell me the name of the current Governor of your state? 2) In general, how many years does a congressperson’s term last? and 3) Given what you know, which are the chambers of the Mexican Congress?

¹²To test the robustness of these findings, we added several controls in additional estimations. We included a variable to consider the level of urbanization of the location of the individuals. In other mod- els we added variables corresponding to the ethnic-racial identity of the respondents (indigenous, mes- tizo and white, the latter as a reference category). Finally, we tested the effect of the frequency of news consumption by different media (TV, radio, newspapers, Facebook, Twitter, and WhatsApp) separately and also using two additive indices: one for consumption in traditional media and another for consumption in social networks. Our main findings hold with these alternative specifications. These results are reported in Table A7.

¹³The probabilities shown in Figure 3 come from the estimates of Model 3, which includes the Morena party identification as a control variable.

¹⁴An important difference with the Argentine election analyzed by ^{Singer (2016)} is the ideological orientation of the party in power. His results indicate that voters with positive economic evaluations in the informal sector favored Néstor Kirchner in 2005. While informality does not appear to have had a direct effect on electoral support for president Kirchner, his party may have had a relative advantage in capitalizing on economic improvement in the pockets of vulnerable voters.

Appendix

Table A1 Descriptive statistics

Variable	Obs.	Mean	Std. Dev.	Min.	Max.
Morena/AMLO Vote	966	0.52	0.50	0	1
Vote for candidates	966	2.32	0.87	1	4
National economy evaluation	966	3.07	0.90	1	4
Personal economy evaluation	966	3.95	0.99	1	5
National security evaluation	966	3.49	1.14	1	5
National corruption evaluation	966	3.85	0.98	1	5
Victim	966	0.23	0.42	0	1
Liberal-Conservative	966	5.58	2.51	0	10
Female	966	0.56	0.50	0	1
Age	966	40.55	14.89	18	86
Education	966	2.26	0.93	0	4
Informal	966	0.63	0.48	0	1
Morena identification	966	0.25	0.43	0	1
PAN identification	966	0.16	0.37	0	1
PRD identification	966	0.04	0.20	0	1
PRI identification	966	0.14	0.34	0	1
Index of political information (IPI)	966	3.01	1.05	1	4
Corruption AMLO	865	0.41	0.49	0	1
Income	636	2.08	0.76	1	7
NSE Index	920	4.17	1.55	1	7
Corruption of opponents (Anaya and Meade)	857	0.37	0.48	0	1

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

Table A2 Multinomial model: Campaign a nd voting issues with respect to the party in government in 2018 (+ income)

	PAN/Anaya Vote	Morena/AMLO Vote	Independent/ Bronco Vote
National economy evaluation	0.231	0.508***	0.426
	(1.45)	(3.37)	(0.99)
Personal economy evaluation	0.164	0.362**	-0.127
	(1.34)	(3.13)	(-0.42)
National security evaluation	0.0791	0.282*	0.152
	(0.60)	(2.28)	(0.44)
Victim	0.303	-0.240	1.493*
	(0.96)	(-0.79)	(2.29)
National corruption evaluation	0.365**	0.355**	0.321
	(2.74)	(2.91)	(0.90)
PAN identification	0.764*	-0.0733	-2.035
	(2.04)	(-0.20)	(-1.80)
PRD identification	0.981	-0.0252	-14.08
	(1.55)	(-0.04)	(-0.02)
Morena identification	-0.278	0.104	-1.911*
	(-0.79)	(0.34)	(-2.26)
PRI identification	-0.934*	-1.460***	-15.66
	(-2.57)	(-4.30)	(-0.02)
Liberal-Conservative	-0.00746	-0.0777	-0.198
	(-0.14)	(-1.58)	(-1.73)
Female	-0.292	-0.624**	-0.0158
	(-1.12)	(-2.58)	(-0.03)
Age	-0.0152	-0.0173*	-0.0582*
	(-1.58)	(-1.97)	(-2.25)
Education	0.265	0.0244	-0.106
	(1.65)	(0.16)	(-0.29)
Income	0.172	0.119	0.419
	(0.91)	(0.67)	(1.14)
Constant	-2.688*	-2.679*	-1.269
	(-2.36)	(-2.53)	(-0.46)
Number of observations		636
Pseudo R²		0.14

Source: Authors’ elaboration based on CIDE-CSES 2018 National Electoral Study (^{Beltrán, Ley and Castro
Cornejo, 2020}). Note: Multinomial model with PRI/Meade vote as reference category. The variables of party identification and personal economic situation were included in wave 1. *p < 0.05, **p < 0.01, ***p < 0.001, t-statistics in brackets.

Table A3 Multinominal model: Campaign and voting issues with respect to the party in government in 2018 (+ NSE index)

	PAN/Anaya Vote	Morena/AMLO Vote	Independent/ Bronco Vote
National economy evaluation	0.372**	0.629***	0.274
	(2.83)	(5.19)	(0.90)
Personal economy evaluation	-0.0205	0.224*	-0.33
	(-0.19)	(2.26)	(-1.40)
National security evaluation	0.105	0.284**	0.286
	(0.96)	(2.84)	(1.12)
Victim	0.357	-0.0626	1.449**
	(1.39)	(-0.25)	(2.87)
National corruption evaluation	0.277*	0.234*	-0.0852
	(-2.53)	(-2.37)	(-0.34)
PAN identification	0.679*	-0.208	-0.762
	(2.24)	(-0.70)	(-1.09)
PRD identification	0.646	0.0572	-14.52
	(1.21)	(0.11)	(-0.01)
Morena identification	-0.213	0.117	-0.913
	(-0.74)	(-0.47)	(-1.49)
PRI identification	-0.945**	-1.288***	-15.72
	(-3.08)	(-4.62)	(-0.02)
Liberal-Conservative	0.0431	-0.0572	-0.119
	(0.98)	(-1.44)	(-1.31)
Female	-0.37	-0.474*	-0.347
	(-1.72)	(-2.41)	(-0.72)
Age	-0.0149	-0.0136	-0.0469*
	(-1.92)	(-1.95)	(-2.45)
Education	0.265	0.0868	-0.0193
	(-1.96)	(-0.7)	(-0.06)
NSE Index	-0.137	-0.125	-0.198
	(-1.76)	(-1.78)	(-1.14)
Constant	-1.617	-1.937*	1.53
	(-1.77)	(-2.30)	-0.74
Number of observations		920
Pseudo R²		0.12

Source: Authors’ elaboration based on CIDE-CSES 2018 National Electoral Study (^{Beltrán, Ley and Castro
Cornejo, 2020}). Note: Multinomial model with PRI/Meade vote as reference category. The variables of party identification and personal economic situation were included in wave 1. *p < 0.05, **p < 0.01, ***p < 0.001, t-statistics in brackets.

Table A4 Logit model: Campaign issues and voting for the candidate representing Morena (+ income)

	Model 1	Model 2	Model 3
National economy evaluation	0.384**	0.411**	0.332*
	(2.98)	(2.97)	(2.10)
Personal economy evaluation	0.322**	0.368**	0.341**
	(3.16)	(3.08)	(2.86)
National security evaluation	0.230**	0.280*	0.326*
	(2.61)	(2.43)	(2.20)
Victim	-0.544**	-0.576**	-0.501*
	(-3.07)	(-3.09)	(-2.42)
National corruption evaluation	0.127	0.149	-0.0000805
	(1.15)	(1.28)	(-0.00)
Corruption AMLO		-1.265***	-0.595**
		(-5.21)	(-2.62)
Corruption opposition (Anaya and Meade)			1.649*** (5.58)
Corruption AMLO * Corruption			-1.089*
opposition			(-2.32)
Morena identification	0.691*	0.592*	0.503
	(2.56)	(1.99)	(1.60)
Liberal-Conservative	-0.0524	-0.0469	-0.0251
	(-1.47)	(-1.37)	(-0.61)
Female	-0.432* (-2.50)	-0.555** (-2.63)	-0.537* (-2.32)
Age	-0.00460	-0.00686	-0.00519
	(-0.84)	(-0.99)	(-0.74)
Education	-0.0773	-0.0605	0.0143
	(-0.67)	(-0.42)	(0.09)
Income	-0.0438	-0.0516	-0.0670
	(-0.43)	(-0.41)	(-0.45)
Constant	-2.735*** (-3.57)	-2.587** (-2.79)	-2.896** (-3.04)
Number of observations	636	559	521
Pseudo R²	0.10	0.18	0.22

Table A5 Logit model: Campaign issues and voting for the candidate representing Morena (+ NSE index)

	Model 1	Model 2	Model 3
National economy evaluation	0.439***	0.457***	0.350**
	(4.42)	(3.99)	(2.83)
Personal economy evaluation	0.304***	0.348***	0.308**
	(3.34)	(3.44)	(2.80)
National security evaluation	0.228**	0.260**	0.306**
	(3.18)	(2.98)	(2.70)
Victim	-0.370*	-0.321*	-0.247
	(-2.48)	(-2.21)	(-1.49)
National corruption evaluation	0.0731	0.115	-0.0236
	(0.85)	(1.27)	(-0.22)
Corruption AMLO		-1.269***	-0.499*
		(-5.75)	(-2.42)
Corruption opposition (Anaya and			1.845***
Meade)			(6.32)
Corruption AMLO * Corruption opposition			-1.494** (-3.20)
Morena identification	0.553*	0.479*	0.386
	(2.46)	(2.03)	(1.47)
Liberal-Conservative	-0.0669*	-0.0703*	-0.0400
	(-2.01)	(-2.11)	(-1.12)
Female	-0.239	-0.330	-0.268
	(-1.64)	(-1.85)	(-1.43)
Age	-0.00275	-0.00487	-0.00449
	(-0.69)	(-1.00)	(-0.89)
Education	-0.0342	-0.0617	0.0405
	(-0.37)	(-0.59)	(0.38)
NSE Index	-0.0479	-0.0344	-0.0708
	(-0.72)	(-0.50)	(-0.83)
Constant	-2.718*** (-3.86)	-2.560** (-3.23)	-2.778** (-3.22)
Number of observations	920	824	766
Pseudo R²	0.09	0.17	0.22

Table A6 Logit model: Campaign issues and voting for the candidate representing Morena (+ presidential approval)

	Modelo 1	Modelo 2
National economy evaluation	0.396***	0.431***
	(4.08)	(3.94
Personal economy evaluation	0.301***	0.344***
	(3.68)	(3.81)
National security evaluation	0.193**	0.220**
	(2.89)	(2.73)
Victim	-0.331*	-0.314*
	(-2.36)	(-2.22)
National corruption evaluation	0.0355 (0.4)	0.0796 (0.82)
Corruption AMLO		-1.243***
		(-5.75)
Morena identification	0.585**	0.536*
	(2.8)	(2.39)
Liberal-Conservative	-0.0729* (-2.24)	-0.0860** (-2.63)
Female	-0.202	-0.293
	(-1.42)	(-1.67)
Age	-0.00309	-0.00475
	(-0.84)	(-1.04)
Education	-0.059	-0.0728
	(-0.72)	(-0.78)
Disagreement with EPN ’s presidential performance	0.383***	0.374*
	-3.56	-3.06
Constant	-3.800***	-3.589***
	(-5.73)	(-4.51)
Number of observations Pseudo R²	947	850
	0.107	0.178

Source: Authors’ elaboration based on CIDE-CSES 2018 National Electoral Study (^{Beltrán, Ley and Castro
Cornejo, 2020}). Note: Logit model. The variables of party identification and personal economic situation were included in wave 1. The presidential agreement variable was included in wave 2. Higher values for the variable of disagreement with EPN’S presidential performance indicate more negative evaluations. *p < 0.05, **p < 0.01, ***p < 0.001, t-statistics in parentheses.

Table A7 Logit model: Campaign issues and voting for the candidate representing Morena (+ additional controls)

	Model 1	Model 2	Model 3	Model 4	Model 5
National economy	0.442***	0.392***	0.427***	0.437***	0.424***
evaluation	(4.48)	(3.72)	(4.36)	(4.52)	(4.37)
Personal economy	0.335***	0.354***	0.336***	0.332***	0.342***
evaluation	(3.90)	(4.12)	(3.97)	(3.91)	(4.00)
National security	0.213**	0.233***	0.229***	0.223***	0.216***
evaluation	(3.22)	(3.40)	(3.49)	(3.40)	(3.29)
Victim	-0.343*	-0.290*	-0.363*	-0.387**	-0.311*
	(-2.47)	(-1.97)	(-2.38)	(-2.77)	(-2.17)
National corruption	0.0766	0.111	0.125	0.0967	0.0957
evaluation	(0.91)	(1.23)	(1.44)	(1.14)	(1.16)
Morena identification	0.533*	0.481*	0.578**	0.575**	0.538*
	(2.49)	(2.26)	(2.67)	(2.71)	(2.52)
Liberal-Conservative	-0.0754* (-2.30)	-0.0931* (-2.57)	-0.0832** (-2.65)	-0.0774* (-2.44)	-0.0769* (-2.37)
Female	-0.232	-0.229	-0.182	-0.212	-0.237
	(-1.63)	(-1.57)	(-1.19)	(-1.45)	(-1.69)
Age	-0.00346	-0.00276	-0.00376	-0.00290	-0.00370
	(-0.88)	(-0.73)	(-0.88)	(-0.73)	(-0.92)
Education	-0.0884	-0.0896	-0.0796	-0.0743	-0.0899
	(-1.06)	(-1.07)	(-0.99)	(-0.90)	(-1.09)
Urban	0.248	0.212	0.310	0.262	0.275
	(1.32)	(1.15)	(1.67)	(1.37)	(1.48)
Indigenous		0.0677
		(0.24)
Mestizo		-0.343
		(-1.60)
Frequency of radio			-0.0819
news consumption			(-1.92)
Frequency of TV news			0.0396
consumption			(0.86)
Frequency of print			0.0616
news consumption			(1.44)
Frequency of news			-0.0450
consumption on			(-1.07)
Facebook
Frequency of news			0.0815
consumption on			(1.38)
Twitter
Frequency of			-0.00287
WhatsApp News			(-0.06)
Usage
Index of news consumption in traditional media				0.00141 (0.07)
Index of news consumption on social networks					-0.00434 (-0.25)
Constant	-2.982*** (-4.60)	-2.851*** (-4.06)	-3.293*** (-4.77)	-3.154*** (-4.72)	-3.018*** (-4.47)
Number of observations	966	923	945	955	952
Pseudo R²	0.097	0.100	0.106	0.100	0.097

Source: Authors’ elaboration based on CIDE-CSES 2018 National Electoral Study (^{Beltrán, Ley and Castro
Cornejo, 2020}). Note: The variables of party identification and personal economic situation were included in wave 1. The variables of media consumption were included in wave 2. Notes: Logit model. The index of news consumption in traditional media includes the frequency of news consumption on radio, TV and newspaper. The Social Media News Usage Index includes the frequency of news usage on Facebook, Twitter, and WhatsApp. *p < 0.05, **p < 0.01, ***p < 0.001, t-statistics in brackets.

Received: June 12, 2019; Accepted: April 03, 2020

Este es un artículo publicado en acceso abierto bajo una licencia Creative Commons

Servicios Personalizados

Revista

Articulo

Indicadores

Links relacionados

Compartir

Política y gobierno

versión impresa ISSN 1665-2037

Polít. gob vol.27 no.2 Ciudad de México jul./dic. 2020 Epub 17-Feb-2021