Measurement Invariance of a Classroom Engagement Measure among Academically At-Risk Students

Abstract

Read online

The current study investigated the measurement invariance of a classroom engagement measure across time points, genders, and ethnicities using a sample of 523 academically at-risk students across grades 7 through 9; this measure was based on Skinner et al.'s (1990) original engagement measure. The engagement measure was comprised of 16 items, yielding three factors: Behavioral Engagement, Behavioral Disaffection, and Emotional Engagement. Configural, metric, and scalar invariance held across the three time points, as did invariance of factor covariances and means, indicating that scores have a similar meaning across all 3 years. The engagement measure also featured adequate configural, metric, and scalar invariance, and invariance of factor covariances and means across genders and ethnicities. These findings suggest the measure is appropriate for investigating substantive hypotheses regarding classroom engagement across different grade levels, genders, and ethnicities. In summary, the current results indicate this measure of classroom engagement is suitable for testing hypotheses regarding group differences in engagement across grade levels, genders, and ethnicities. Researchers may also use this measure to examine relationships between the engagement factors and other important academic outcomes. Limitations of the current study, such as certain caveats regarding convergent validity and internal consistency, are also discussed.

Keywords

• engagement
• longitudinal data analysis
• measurement invariance
• multigroup comparison
• confirmatory factor analysis