III: Preliminary analyses and analytic plan

In this chapter, we describe our preliminary analyses and analytic plan. The preliminary analyses were focused on several aspects of construct validity. We first distinguish between construct invariance and measurement invariance as two different approaches to establishing that our scales mean the same thing to adolescents categorized as African American or European American and female or male. We then consider the extent to which (a) correlations between the same construct assessed at different points in time provide evidence of convergent validity and (b) correlations between a given construct and theoretically similar and dissimilar construct provide evidence of convergent and discriminant validity, respectively. Given that our assessment of construct invariance (described below) revealed some differences in the magnitude and direction of correlations for the R/E discrimination measures across the four R/E and gender groups, we next describe our analysis of measurement invariance for these measures. Finally, we provide an overview of the HLM analyses that were used to examine the growth curve trajectories in Chapters IV through IX and describe how the results are presented in tables and figures.