As described in the parent grant, the primary questions will be tested using a 3-level mixed-effects regression models. In fact, the same mixed-effects model can be used to test all of the hypotheses, where the inclusion of higher order interactions are the only significant changes in model specification for the different hypotheses. The generalized mixed-effects regression model will contain two levels of nesting (repeated observations within children, and children within classrooms), with school entered as a fixed effect covariate (due to the relatively small number of schools). For the hypotheses noted below, we will examine main effects and interactions of teacher practices (TP) and psychological climate (PC) on child outcomes, and of teacher practices and teacher-student relationships (TS) on child outcomes. A significant interaction would indicate a moderation effect of teacher practices with classroom characteristics. The general model can be augmented to include any relevant time varying covariate at level 1 (repeated measurements), or time invariant covariates at levels 2 (child), or 3 (classroom). Hypothesis 1: Teacher practices will predict child outcomes. This model will contain main effects of TP, time, and the TP*time interaction. Hypothesis 2: Classroom characteristics (teacher psychological climate and teacher-student relationships - TSR) will impact on teacher practices. This is a 2-level model with teachers nested within schools and TP used as the outcome variable. The model will include main effects of PC, TSR, and time, and the PC*time and TSR*time interactions to predict TP. Hypothesis 3: Classroom characteristics (teacher psychological climate and teacher-student relationships) will impact on child outcomes. This model will add main effects of PC, and RE, and the PC*time and RE*time interactions to the model in H1 and remove the TP main effect and TP*time interaction. Hypothesis 4: Teacher practices will mediate the association between classroom characteristics and child outcomes. This model will reintroduce the TP main effect and TP*time interaction to the model in H3. If the model in H3 identifies significant PC and RE related effects on child outcomes, but they are no longer significant when the TP effects are added in H4, then we can conclude that the TP effects mediate the relationship between PC, TS, and child outcomes. Hypothesis 5: Classroom characteristics will moderate the association between teacher practices and child outcomes. This model will add the PC*TP*time, and TS*TP*time interactions to the model in H4. This model will test to see if PC and TS moderate the relationship between TP and child outcomes (i.e., does the TP*time interaction depend on classroom characteristics).
Showing the most recent 10 out of 13 publications
