Methods: Structural Equation Modeling (SEM) was used to examine the meditational role of emotional and behavioral engagement between CRI and math end-of-grade (EOG) test scores. In order to demonstrate mediation in this case five conditions must be established: 1) CRI must predict EOGs, 2) CRI must predict engagement, 3) engagement must impact EOG, 4) given CRI predicting engagement and engagement predicting EOG, the effect of CRI on math achievement must be at or near zero to indicate full mediation, and finally and often neglected, 5) there must be evidence of a causal relationship between CRI and math scores. In the current SEM model, CRI, emotional engagement, and behavioral engagement were latent variables estimated by multiple student-report survey items.
Results: Initial models established the first three mediation conditions. Condition 5—causality between CRI and math EOG—is supported by analyses detailed in another paper within this symposium. That leaves condition four. The full mediated model was specified in which CRI in sixth, seventh, and eighth grades predicted eighth grade behavioral and emotional engagement, which in turn predicted math EOGs. This model had fair fit (RMSEA = .056, CFI=.89, IFI=.89), with all but one of the pathways significant. To test for mediation the direct pathways were added from CRI to EOGs and the change in chi-square in relation to degrees of freedom was examined. A significant but small chi-square change (9.2 @ 3 df) suggested a partially mediated model. Examining the coefficients for the three added pathways revealed a significant pathway, from CRI in eight grade to EOG, therefore the final model included that pathway. Fit for the final model was improved (RMSEA = .045, CFI=.92, IFI=.92) and indicated that engagement partially mediated the impact of CRI on math achievement, and that CRI strongly predicted emotional and behavioral engagement, with both types of engagement predicting math achievement. Squared multiple correlations indicated the model explained 32% of behavioral and 17% of emotional engagement, and 8% of math EOG scores.
Conclusions and Implications: The current findings support the multidimensional conceptualization of student engagement and the role of engagement as a partial mediator of the impact of CRI on academic achievement. This suggests that we may first need to change the way students feel and behave toward school in order to effect change in academic outcomes. And finally, that when students perceive that what they are learning in school is relevant to their futures, we see such positive changes.
