Previous comparative electoral studies using aggregate data indicate the importance of party-system variables, such as polarization and the number of parties, with regard to the level of volatility between two elections. Research using individual level data has shown elements, such as political knowledge, political disaffection and party identification, that explain why voters remained faithful to their party or not. Until now, no study has investigated these variables simultaneously on individual level data using a large set of elections. This study fills that important gap in the literature using data from 29,591 voters in 33 elections. We find polarization influences party-switching at the individual level, rather than the sheer number of parties, as aggregate-level analyses suggest.