Surveys have shown that the evaluation of different public policies have low correlations with each other, correlations were particularly low among illiterates and poorly educated respondents and they were poor predictors of voting intention and party preferences. This was particularly true of research carried during the military regime, but was not limited to it. Data from a large pre-election poll, with a stratified random household sample in the Distrito Federal contradict these findings. Evaluations of public po-licies cluster, yet are not "block and blind" answers, the magnitude of their intercorrelations does not follow educational lines, and they are excellent predictors of voting preference for governor.The disparities among these findings may result from real changes in the population during the last decade and a half, to specificities of the Distrito Federal (highest per capita income and one of the highest educational levels in the nation), to an election in which the incumbent governor ran against a former governor, to the campaign emphasis of the incumbent governor, who was running on his record, to characteristics of the military regime or to combinations thereof. Regardless of the explanation that one may favored, the 1998 elections showed the existence of a rationality that linked voting intention to the evaluation of public policies.
Elections; Public policies; Rational choice; Social classes; Authoritarian regime; Democracy
