The primary data analyzed in this book are collected from a series of panel studies. Since there are some statistical issues with this kind of analysis, I include a brief discussion of those problems and issues in this appendix, but this section is not essential to understand the analysis presented in the empirical chapters. Specifically, there are various concerns with how the dependent variables are measured. In panel studies, there are a number of approaches to analyzing effects over time. Two of the most common groups of approaches are “change score” models and “regression adjustment” or “static score” models (Judd and Kenney 1984; Finkel 1995). Each of these models improves upon standard cross-sectional approaches by incorporating the dynamic components, but in so doing they can suffer from a different set of statistical problems.

The “change score” method uses simple differences between the level of the dependent variable at the various time points. In a two-wave study, the change score is simply the difference between the dependent variable at the first and second measurements. This method regresses the change in Y on the change in X. Variations in this general model include using either X from a single time point rather than the change in X. This model is also known as the “unconditional change score” or the method of first differences (Finkel 1995, pg. 5).