Use of the generalized logit model in the analysis of crime data in Santarém-PA

Abstract

Crime is a social phenomenon spread in different modern civilizations and studied by researchers in different fields of knowledge. Understanding the relation between different variables related to crime can be a tool of great importance in planning and implementation of specific public policies. One of the tools available for this understanding can be accomplished by through the use of the analysis of categorical data and its branches. The work presented here objective using the generalized logit model to analyze the relationship between income and the use of drugs on the type of crime (against life, against property, against customs and trafficking narcotics) committed by inmates from agricultural Silvio Hall de Moura prision, in Santar´em-Par´a-Brazil; 344 detainee interviewed in 2004 and 2005, studying the agent asset of the crime and not the victim. Considering the type of crime as the dependent variable, the logit generalized model  was the best fit to the data, from which probabilities of occurrences of each intersection of the categories were predicted, and finally odds and odds ratios were obtained. The use of the logit generalized model proved efficient in the analysis of categorical data process identifying, among other factors, which inmates with lower incomes are more likely to commit crimes (equity, customs and narcotics trafficking) when compared with crimes against life, and this is the most frequent type of crime in the sentencing process those detainees.