Multilevel Student Engagement and Confidence in Learning: Assessing Construct Validity Across Students and School

The purpose of this study was to examine the psychometric properties of six student outcome measures. In particular, it examined the unidimensionality of student engagement (ENG-M, ENG-S, and ENG-R) measures, and confidence in learning (CON-M, CON-S, and CON-R) measures across mathematics, science, and reading. To empirically investigate these measures' psychometric properties, the study hypothesized that a) these measures are multilevel constructs, b) based on the way these measures administered, method effects were a source of these measures misfit, and c) indicators' translation errors were another source of these measures' misfit. The data was obtained from the Trend in International Mathematics and Science Study (TIMSS-2011) and Progress in Reading Literacy Study (PIRLS-2011) of Saudi 4th grade students. Utilizing Multi Confirmatory Factor Analysis (ML-CFA), the three hypotheses of the study were supported. Interestingly, controlling for the hypothesized sources of misfit was not as effective in student confidence measures as in the engagement measures. Several implications for policy makers and practitioners as well as future research were discussed.