Aim 2a

Goal of Aim 2a: To test the relative effectiveness of the program for various sub-populations of students

Hypothesis: Positive Action is equally effective for students from different racial, ethnic, and other demographic backgrounds

To test these hypotheses, the model depicted in Figure 6 will be used (i.e., time-invariant moderator) to investigate if any of these moderators explain different PA effects on student outcomes. At times group differences in a measure, in addition to their demographic backgrounds, may confound the identification of actual effect variations (e.g., SECD differences between males and females). In these situations, these potential confounding variables will be added to the model to control for group variations.

Figure 6:

Considering other potential moderators,  it is hypothesized that students with different onset/developmental pathways will respond differently to PA. For example, the hypothesis is that a moderator representing the level of engagement in serious forms of violence will help to explain PA effect differences. In this example, the moderator is time-variant in that the level of violence can change for each student over time. Other time-variant examples include substance use, family engagement, and school dynamics. Therefore, to test the hypotheses involving time-variant moderators, the model represented in Figure 7 will be followed.

Figure 7:

For all the moderation models, a traditional moderator analysis will first be conducted using the baseline levels for a moderator, independent variable (e.g., PA component), and a student outcome. Next, the moderator effect on the growth/change of a student outcome effected by a PA component will be analyzed. In these analyses, the hypothesis is that PA effects will be greater for different student characteristics (i.e., time-invariant) and student behavior/beliefs/environment (i.e., time-variant). The comparison of baseline-only moderator findings to growth/change findings will greatly increase the understanding of moderator effects over time and identify potential best-practices in moderation analyses when researching SECD and academic student outcomes.