Electoral change can happen in just three ways: population replacement, turnout switching or party switching. However, we know relatively little about how common each of these factors are, how they vary across electoral contexts, and how they much each contributes to electoral change. To answer these questions I compile a dataset of 104 inter-election panel surveys covering 18 countries. Across the 104 election pairs, party switching contributes three times as much to aggregate volatility as turnout switching on average, and party switching is the most important factor in 97% of election pairs. These results are robust to using validated measures of turnout. I show that previous research emphasizing the importance of turnout switching focused on the highly atypical case of recent US elections, which does not generalize to most electoral contexts.