Türkiye'deki Seçim Anketlerinin Toplam Anket Hatası Perspektifinden Bir İncelemesi

Abstract

Election polls are not only essential to measure electorates' political attitudes, preferences, and behavior but may also influence their formation. In this study, we evaluate election polls in Turkey, which have been increasingly popular and criticized by political actors in the last years, from a survey methodology perspective and investigate whether they follow internationally recognized standards. To this end, we collected a comprehensive dataset covering all election polls on eight elections between 2011 and 2019 published at traditional or social media outlets. Our assessment of a total of 295 election polls from the perspective of the "total survey error" paradigm and CNN's transparency guidelines shows that, most election polls in Turkey do not follow scientific standards, especially in terms of representation, and are thus unsuccessful at predicting election outcomes. In the conclusion, we discuss how survey data from public opinion and election polls, and cross-national surveys can be employed in political science research.