According to the proximity theory of voting, if parties and voters can all be represented on the same policy issue dimension, voters will vote for the party that stands closest to their policy issue position. Contrary to the spatial tradition, the Michigan school of voting professes that voters do not take rational decisions on the grounds of policy issue proximity but vote for the party that they feel psychologically close to, meaning the party of identification.

Relying on Adams, Merrill and Grofman (2005) work that reconciles the above-mentioned traditions of voting, the thesis at hand shows the effect that a psychological characteristic such as party identification imparts on parties’ rational strategies to locate their optimal positions on the policy issue dimension. The analysis is divided into two parts. The first part assumes that voters vote with certainty or deterministically for the party of choice on the grounds of proximity and party identification. The second part assumes that voters’ decisions are probabilistic in the sense that their utilities are perturbed by unmeasured components that render their decisions uncertain from the parties’ perspectives.

Under deterministic voting it is shown by means of a simulation analysis, that parties have incentives to present different optimal positions when party identification affects the voting and when it does not. Under probabilistic voting it is shown that party identification has a curvilinear effect on Finnish parties’ Nash equilibrium positions; low levels of party identification lead to strong centripetal competition; medium levels of party identification lead to less strong centripetal competition; high levels of party identification bring the competition back to strong centripetal outcomes. The afore-described curvilinear effect of party identification also holds sway when data on Swedish elections is used and is in line with the results reported by Adams, Merrill and Grofman (2005) when applying the same algorithm of voting to French politics.

Applying spatial model techniques to survey data on real elections this work also shows the extent to which voters’ decisions in Finland are based on long-term psychological characteristics such as party identification as opposed to left and right policy orientation; whether Nash equilibrium positions exist in the case of Finland and Sweden and how the characteristics of such equilibrium configurations compare with previous empirical studies; the extent to which party identification and left and right policy orientation affects parties’ expected vote share; why Finnish parties with large partisan constituencies have electoral incentives to move centripetally on the policy issue dimension; whether there is a similarity between Finnish parties’ policy issue positions in 2007 with their Nash equilibrium configurations; and whether Finnish voters’ perceptions of parties’ left and right positions in 2007 are shaped by the psychologically based concept of projection.