Recently growing number of studies analyzed the relationship between electoral systems and ideological congruence (Blais and Bodet 2006, Powell 2009). Although prior works point to higher congruence in PR systems, lately it is argued there are no significant differences between PR and SMD systems (Golder and Stramski 2010). This is now known as the ideological congruence controversy. Yet, these studies fail to draw a complete picture of political representation in general and ideological congruence in particular, due to simplistic assumptions. In my paper I will challenge these in a number of ways: First of all, as dependent variable I will compare distribution of preferences of voters with distribution of parties rather than median positions. As Braumoeller (2006) points variance altering causal mechanisms had been significantly under appreciated in the political science literature. However, analysis of changes in variance both statistically and substantively bolsters our understanding as it provides a more detailed depiction of political phenomena. In our case, showing how electoral rules affect a system's ability to represent the whole distribution of ideological preferences, rather than central tendency of voters will be a significant achievement. Also, different than previous studies I will compare ideological positions in a multidimensional ideological space rather than a unidimensional one (Benoit and Laver 2006); and instead of dumping various complicated electoral systems into PR-SMD dichotomy I will use electoral system permissiveness or proportionality (Cox 1997, Taagapera 2002). I believe these innovations would help us to see a broader picture regarding how different electoral rules of all configurations (not only pure PR or SMD) affect how preferences of the whole spectrum of voters (not only the median voter) are to be represented in the parliament. Comparative data from CSES and national election surveys of old and new Western parliamentary democracies will be used in the study and a series of regression models will be constructed in order to analyze this data.
