This paper examines the tension between two prominent policy change models—incrementalism and punctuated equilibrium—and their statistical underpinnings. It argues that heavy-tailed distributions of policy changes, often used to distinguish these models, rely on an assumption that policy inputs are normally distributed if variance remains constant over time.
However, this study shows that the within-time variance in such input measures—like public opinion survey items—actually varies across periods. This empirical finding challenges conventional assumptions about data distribution, suggesting normality tests may be flawed when applied to real-world political processes with changing variances.
The paper concludes by urging scholars testing punctuated equilibrium theory to reconsider their methodological foundations and explore more flexible statistical approaches.
