Abstract:

This dissertation aims to propose a methodological approach to apply modelling to public opinion polls in order to favour the process which leads from description to explanation and further to prediction. The inclusion of modelling would also increase the precision of the vote intention estimates in the middle term as well as improve the diagnostic- power of the positioning studies carried out for political figures or entities. Two eases are presented to demonstrate the viability of the proposal: one having to do with voting tendency which was conducted among a sample of 800 cases; and another conducted amongst 500 respondents and dealing with the positioning and repositioning carried out in the province of Buenos Aires Argentina between March and September 1997.