Investigação da relação entre a cor do veículo e o risco de furto e roubo:uma análise para o estado de são paulo de 2016 a 2021

Abstract

The present study aimed to investigate whether there is a relationship between the color of the vehicle and the risk of a theft or robbery event. The data used were obtained from the Department of Public Security of the State of São Paulo, containing the reports of thefts or thefts of vehicles between 2016 and 2021. As there is no individualized data of vehicles that were not stolen/stolen, the induced exposure method was applied. For the application of the method in this work, the luminosity condition at the time of theft or theft of the vehicle was the factor chosen to define the two comparison groups: a group called "prone to colors", formed by the occurrences in which the natural luminosity was sufficient and, thus, the color of the vehicle could have contributed to the occurrence of a theft or robbery; and another group, called "induced exposure", formed by the records of thefts and robberies that occurred in conditions of low natural light, less conducive to the influence of the color of the automobile due to the lower visibility. The Binomial Negative and Poisson distributions were used to model the number of robberies and thefts using generalized linear models. The adhesion tests applied showed that only the negative binomial model adjusted to the data, which presented an overdispersion behavior. The rates of theft/theft (division of the number of occurrences by the size of the fleet) for the vehicles of the colors black, white, silver and red were higher than those of the colors blue, green, brown and yellow. However, the results of the negative binomial model showed that, statistically, there is no car color more or less conducive to suffering a robbery or theft when compared to white vehicles, a color used as a comparative baseline of the study.