A caixa de pandora dos gastos de campanha: eleições para deputados federais no Brasil (2010-2018)

Abstract

What is the effect of the types of campaign spending on votes? Our overall goal is to answer this research question. Technically, we will use descriptive statistics and a linear regression model to test the hypothesis that changes in the types of campaign spending have an impact on winning votes. Our unit of analysis are candidates for the position of federal deputy in Brazil for the 2010, 2014 and 2018 elections. Our dependent variable (VD) is votes obtained and our main independent variable (VI) is the types of campaign spending (modern and traditional). Our data was systematized from the Superior Electoral Court (TSE) repository. The results indicate that (1) traditional expenditures present statistically significant differences between challengers and incumbents; (2) the 2010 elections showed the highest correlation between campaign spending and votes (r = 0.670; p-value < 0.000); and (3) the linear regression model that best explains votes is the one that tests traditional spending (2010, 2014 and 2018), with an adjusted r2 of 0.613, where each increase of one log unit in traditional spending returns 0.517 votes.