Schools are increasingly measuring school climate and social emotional learning, yet few measures conduct invariance testing by race, gender, or language despite known differences in perceptions of these constructs based on these identities.
This study reports on the validation process of a school climate and social emotional learning (SEL) student survey in the context of a research practice partnership (RPP), including ways the research process was shaped by the partnership. Guided by the principles of practical measurement and the goal of creating a measure used for continuous improvement, we examined item quality, construct validity, internal consistency, and measurement invariance.
Data are from a district-wide survey of 29,415 students enrolled in 3rd through 12th grade. Confirmatory factor analysis (CFA) was used to establish a measurement model. Multiple group CFA was used to assess configural, metric, and scalar invariance by race, gender, home language and grade band.
The measurement model had adequate fit after minor post hoc modification, albeit with high correlations among factors, showing evidence of construct validity. Scales were sufficiently reliable. Evidence was found for configural and metric invariance but not scalar invariance for race or grade band.
Findings suggest that districts measuring SEL and school climate should take additional care to attend to differences in how students perceive these constructs. Ongoing consultation with the RPP was crucial to designing a measure that was practical and useful for continuous improvement.