Analysis of Regional Aspects of Voting Behaviour: The Case of Polish Presidential Election

Transition of votes is observed when a voter who voted for a given candidate in the first round of election, votes for another candidate in the runoff round of election. There are two types of transitions. Transition is obvious when preferred candidate is a loser in the first round. In other case transition also may happen. Exact information about the scale of transitions is usually unavailable. There are many opinion pools, even exit pools, but exact transition’s data is not collected during election. Ecological regression techniques give an opportunity to obtain quantitative description of electoral behaviour from aggregated data. Aggregated data set is published after election. Ecological regression approach gives reliable results under homogeneity assumption. Homogeneity in this case is considered in term of electoral behaviour. Homogeneity assumption if applied for the whole country is usually not held. Regional decomposition of estimation process, used for the description of voters’ behaviour, extends application the ecological regression to large regions or even for the whole country. In this analysis, maximum likelihood approach and regional decomposition of voting results is used. The last presidential election 2010 in Poland is used as an empirical example. Presidential election in Poland consists of two rounds if none wins more than 50% of the votes, there is runoff round. The interval between rounds of election in Poland is two weeks.