Sociophysics in the mexican case (Sociof\'isica en el caso mexicano)

TL;DR

This paper analyzes electoral data from Mexico using statistical physics methods, revealing regularities like power laws and Daisy models, and shows that entropy-based approaches cannot detect vote-buying or electoral fraud.

Contribution

It applies statistical mechanics models to Mexican electoral data, identifying regularities and testing entropy methods for fraud detection, which proved ineffective.

Findings

Power law regularities found in Mexican election data

Daisy models linked to corporate voting patterns

Entropy methods do not detect vote-buying in Mexican elections

Abstract

The availability of electoral data in electronic format allowed to physicist and mathematicians their own analysis. Beyond theoretical models for electoral processes, new regularities had been found in the data analysis of many countries and many elections. Here, we discuss some of the statistical regularities found for M\'exico, from power laws Daisy models linked in corporate vote. The methods used are those from standard statistics and statistical mechanics. As a particular application we discuss an entropic approach to the distribution of votes for the Partido Revolucionario Institucional and we demonstrate that this kind of method DOES NOT allow to discover a massive buy of vote. The argument is that such a distribution has been amazingly consistent in the last two decades elections. Other methods have been unsuccessful to discover signal of fraud. RESUMEN: Con la disponibilidad de datos electorales en formato electr\'onico los f\'isicos y matem\'aticos iniciaron sus propios an\'alisis de los datos. M\'as all\'a de los modelos te\'oricos desarrollados para los procesos electorales, nuevas regularidades empezaron a surgir de los datos en muchas elecciones y en muchos pa\'ises. Aqu\'i discutimos algunas de las regularidades estad\'isticas encontradas para M\'exico. En particular, la aparici\'on de leyes de potencia y modelos Margarita (Daisy models) que se pueden asociar al voto corporativo. Los m\'etodos utilizados son de estad\'istica y f\'isica estad\'istica est\'andar. Como una aplicaci\'on particular, discutimos una aproximaci\'on entr\'opica a la distribuci\'on de votos del Partido Revolucionario Institucional y demostramos que \'este m\'etodo NO permite diferenciar una compra masiva de votos pues su comportamiento ha sido sorprendentemente consistente. De igual forma otros m\'etodos han resultado infructuosos en determinar signos de fraude.