School Engagement: Testing the Factorial Validity, Measurement, Structural and Latent Means Invariance between African American and White Students
Echeverria, Roy Arnon
MetadataShow full item record
This study was designed to accomplish three main objectives. The first objective was to test the hypothesis that school engagement is a multidimensional construct with three factors: behavioral, emotional, and cognitive or academic engagement. The second objective was to test for invariance of the measurement and factorial structures of school engagement across white and African-American students. And the third objective of the study was to test for invariance of the latent mean structures of school engagement across white and African-American students. In order to accomplish the objectives of the study a step by step approached, using structural equation modeling, was followed. First, the best fitting model of school engagement for both white and African American students were identified. Second, invariance of the number of underlying factors of school engagement across white and African-American students was tested. Third, invariance of factor loadings across the two racial/ethnic groups was tested. Fourth, invariance of the factor variances and covariances was tested. Fifth, latent mean structures of school engagement between white and African-American were compared. Finally, the results of the calibrating sample were cross-validated with the second half of the sample. Results from this study produced consistent support for a three-factor model of school engagement and without cross-loadings to other dimensions of school engagement. However, some parameters including factor loadings, factor variances and latent means were found non-invariant across white and African American students. African American students rated themselves statistically significantly higher on emotional engagement than white students. In addition, weaknesses in the measurement model especially the reliability coefficients of observed indicators and variance accounted for by the latent factors were identified. Cognitive engagement proved to be the most difficult to measure among all three dimensions of school engagement. Finally, analysis of the cross-validating sample produced some important differences which included one additional non-invariant factor loading, one factor covariance, and one additional latent mean difference between white and African American students.
- Doctoral Dissertations