Describing the determinants of electoral behaviour is one of the fundamental research problems in studies of electoral democracies and serves to test many hypotheses about political behaviour in modern democracies.
The standard analytical approach for determining factors responsible for party choice is the discrete choice model. In this approach dependent variable indicates respondents' choices, i.e. parties or candidates to whom respondents voted. The discrete choice model approach includes such analytical techniques as multinomial logit, conditional logit or multinomial probit. An alternative approach to the analysis of determinants of electoral behaviour is the concept of the propensity to vote (PTV) and y-hats (van der Eijk et al. 2006; van der Brug, van der Eijk, and Franklin 2007). In this approach, dependent variables are utilities of voting for particular parties. These utilities are revealed by voters and are related to each party significant in a given party system. The analytical techniques applied in PTV approach is OLS regression.
An application of both approaches is compared in this paper. Discrete choice model and PTV were applied simultaneously to the analyses of the economic voting model in the case of the Polish parliamentary elections 1997-2011. Data under study comes from the Polish National Election Study. Due to the high level of the volatility of the party system as well as electoral behaviour, the analysed data allow to track the effects of both approaches in cases that differ significantly in the configuration of support for particular parties.
The analyses carried out with both approaches show coherent results. Moreover, both methods complement one another - some dependencies that are not visible in a discrete choice model approach are revealed when the analysis is performed using PTV, and vice versa. The results of this comparison lead to the conclusion that a full understanding of the mechanisms determining electoral choices requires applying both approaches at the same time.
