Some problems encountered with the telephone survey as used for electoral predictions

Abstract

Telephone surveys were first used in the context of electoral predictions (to be specific, in the United States presidential elections of 1936). From then on, their use in opinion and market research has increased steadily. This paper¿s aim is to analyse the impact that the lack of coverage has on electoral polls based on telephone surveys, revealing the dangers of electoral predictions which only consider homes with listed telephone numbers. In order to carry out our inquiry, we have used information from the barometers carried out in 2004 by the Centro de Investigaciones Sociológicas.