The Specter of Violence: Perceptions of Violence and Political Behavior in Mexico

Violence stemming primarily from organized crime plagues Mexico. The sharp increase in criminality and insecurity, particularly since the start of the war on drugs in 2006, is one of the most significant developments for the country since the turn of the century. Officially recorded crimes, which are estimated to represent only one tenth of all crimes committed, have grown by 70% in the last two decades. The homicide rate in particular has exploded: between 2000 and 2018 the rate nearly tripled, climbing from eleven murders per 100,000 inhabitants to twenty-nine. Two hundred and fifty thousand deaths since 2006 are considered directly attributable to the drug war. This violence has led to the internal displacement of 345,000 Mexicans as of 2019, an increase of 431% from a decade prior, and more than 35,000 disappearances.A surge of scholarly research on the causes of bloodshed in the country have highlighted the growing intractability of public security issues and have raised questions about the efficacy of various policy initiatives aimed to curtail the violence. Yet, little attention has been paid to how violence is understood by the populace and to the political consequences of those perceptions. This dissertation addresses that deficit by using large-scale survey data with embedded experiments to systematically address two driving questions: 1) how do Mexicans perceive violence? and 2) how do perceptions of violence affect engagement in electoral politics?Using data from an original survey of 17,451 Mexican voters contacted six weeks prior to the 2018 presidential election, I examine the ways in which Mexicans perceive violence, specifically asking a random subsection of respondents to estimate levels of homicide and kidnapping in their state. I find that misperceptions of violence are nearly universal, with underestimation of homicides and overestimation of kidnappings being dominant. Yet, misperception within each type of violence is not uniform: A fifth of respondents overestimated homicides and nearly a quarter underestimated kidnappings. Elite discourse and media coverage of violence interact with different cognitive biases and adaptations to shape the way that information is assimilated. The salience of violence, more than actual, objective levels of violence, drive these misperceptions. Moreover, perceptions of violence in one’s state often reflect violence levels of years prior, indicating that these perceptions are relatively stable, slow to update, and even dramatic recent changes in violence levels have little effect on people’s beliefs. In exploring the heterogeneity in perceptions of violence, I find that a number of state-level and individual-level factors influence how respondents interpret the violence around them. Respondents along both the northern and southern border were more likely to overestimate homicide. Those living along the southern border were substantially more accurate in their assessment of kidnapping than those along the northern border or living in the interior; they still tended to overestimate kidnapping, but not to the same extent as their peers elsewhere. Those living in more wealthy and/or in more economically unequal states were also more substantial overestimators of violence than their peers in either poorer or more equal states. At an individual level, one’s level of education and the attention paid to political processes drive overestimation of violence, supporting the hypothesis that exposure to media and elite messaging, as well as retention of that information, are key drivers in overestimation of violence. Additionally, support for leftist candidate and current president Andrés Manuel López Obrador was highly correlated with overestimation of violence.These misperceptions have serious political consequences. I use both experimental and observational data to explore how perceptions of violence, and its two key driving components, actual violence and the salience of violence, alter political attitudes and behaviors. In doing so, a clear picture emerges: the citizens of Mexico who experience violence the most strongly are hiding in fear, expressing uncertainty over their electoral decisions, and withdrawing from political life. Insecurity has dramatically decreased citizens’ the sense of personal safety. It has led them to retreat from electoral politics in clear ways: they are less likely to participate in elections, are more indecisive about who to vote for when they do chose to engage, and feel alienated from political parties. They are also more likely to reject reforms and tolerate societal ills, including corruption and criminal organizations themselves, than those whose overall experience of violence is less heightened. Fear, apathy, and uncertainty drive this withdrawal.While the majority of scholarly research on the effects of violence on political behavior has focused on the relationship between actual violence levels and participation and vote choice, researchers have largely missed that perceptions of violence, not violence itself, are the key driver of these processes. Now, a decade and a half after the start of the militarized war on drugs in Mexico, the death toll has reached new highs for three consecutive years, with no signs of tapering off. The specter of violence hangs over Mexico and is pushing citizens backward in retreat