Uma análise dos efeitos da violência municipal sobre os resultados do referendo do Estatuto do Desarmamento de 2005

Abstract

This paper aimed to verify a possible relationship between the proportion of the votes in the option NÃO in the Referendo do Estatuto do Desarmamento of 2005 and the level of violence at municipalities. The Regreesions Discountinuity Designg method was used for a sample of 5591 brazilian municipalities, whose variables were defined for the years of 2000, 2004 and 2005. The sample was taken from Repositório de Dados Eleitorais, of the Tribunal Superior Eleitoral (TSE), DATASUS/Ministério da Saúde, IBGE (Instituto Brasileiro de Geografia e Estatística)) and Federação das Indústrias do Estado do Rio de Janeiro (FIRJAN). The treatment group was defined by municipalities with was reported less than 10 homicides per 100,000 inhabitants, and de control group was composed by those defined as violent for Organização Mundial da Saúde (OMS), in which index presented value above 10. Two models were estimated, with voting margin between the two possible alternatives in the election - YES and NO - less than 3% and 5%. Most of the estimates of the local mean effect of the treatment over the treaties were not statistically significant at traditional levels. Only the estimated coefficient for the model with 3% voting margin showed a statistically non-zero value indicating a 0,43 increase in the proportion of the votes in NÃO in the most violent municipalities, in case of was not inserted covariates. Thus, there is no robust evidence to conclude a causal relationship between municipal violence and the 2005 Referendum results.