The total survey error paradigm and pre-election polls: the case of the 2006 Italian general elections

Abstract

Pre-election polls sometimes fail to reach the purpose for which they are carried out: to provide accurate predictions of electoral out-comes. By looking at the 2006 Italian General Elections, this paper aims to assess the role that different factors play in determining the accuracy of the pre-election polls. We find strong evidence that the quality of the sampling frame and non-respondents may contribute to biasing the polls results. This paper also aims to show how to over-come some of the limitations of the survey data by using statistical matching techniques and weighing procedures.