Search

This paper has as its goal the measure of the accuracy of the pre-election surveys published in the Spanish media for the 2014 european Parliament, the regionals in the first half of 2015 and the locals in seven of the biggest Spanish cities. Starting with 120 analysed surveys, the aggregated distance model consists of the differences between predictions and actual results by every party and it develops an error coefficient that allows the measure and comparison of those predictions. Results are presented by election, communication group where predictions were published, research company producing them, political party and time distance to election day. Also it is analysed the existing relation between these and some other variables, such as sample size, and the error coefficient as dependent variable with the objective of knowing which are the most decisive factors affecting polls accuracy. [ABSTRACT FROM AUTHOR]