This study uses the method of two-stage least squares of dynamic panel data models to examine the variables closely related to electoral stability and change in Taiwan, especially focusing on the impact of electoral reform of the Legislative Yuan in 2008 on voting behavior. To account for causal effects, the analysis includes 1,820 observations consisted of the 364 township and village-level units of five legislative elections from 1998 to 2012. The data sources come from the archive of election outcomes officially released by the Central Election Commission, and the statistical yearbooks published by county and city governments. The findings reveal that the variables of electoral reforms, types of electoral system, gender ratio, the level of educational attainment, and proportion of civil servants present statistically significant associations with the index of vote volatility. In addition, the factors of macroeconomic conditions, voting patterns, and geographical contexts emerge as statistically significant and in the anticipated directions. The empirical results demonstrate the Duverger’s law and also the self-fulfilling prophecy; i.e., the reforms of electoral system exert a profound effect of the stability and change of voting behavior. There are at least three academic implications that can be drawn from the outcome of this study. First, this study verifies the relationship between electoral system, party system, and electoral stability and change. Second, it empirically measures the vote volatility index which might be valuable for campaign strategies. Last, different from the qualitative and limited dependent variable models in the previous studies of political participation, this study takes advantage of a dynamic panel data model to assess the impact of electoral reforms on stability and change in voting behavior. Developing an electoral model that is both concise and accurate awaits future research.