Predicting Election Outcomes with Demographic Data

Abstract

Abstract Predicting Election Outcomes with Demographic Data Vlad Catalin Balint College of Social Sciences/Department of Economics The Graduate School Seoul National University Models of voting choice in multiparty elections are influenced by underlying assumptions about the voting-decision process of individuals. Usually the statistical methods used to estimate these models impose very high restrictions about the choices of voters that can give rise to some suspect inferences. Therefore in my analysis I will use two less restrictive models, first the Linear Model and then a Simple Logit Model for more robust results, where besides the IIA assumption I also impose the assumption that the political competition takes place along a single ideological axis between Incumbent Government and Opposition. Using demographic variables I try to predict which Coalition will form a government after parliamentary elections. Theres a multi-party system in Norway where usually no party has the chance to win a majority in the Parliament and hence to govern by itself. This makes parties to collaborate with each other and try to form governmental coalitions or even minority cabinets. The complexity raises in part due to the fact that a voter can only cast his/her vote for an individual party that may be or not part of the coalition that will be formed afterwards. And even if it will be part of it, there is no guarantee that its views will coincide with those of the coalition as a whole or its ex-ante conditions will be satisfied. Keywords: multiparty elections, coalition government, elections prediction, demographics. Student Number: 2013-23780