Methods: All students (N=29,593) from the 94 SPS schools in grades 3-12 take the annual school climate survey (44.4% male, 45.8% female, 6.5% prefer not to state; 59.2% English at home). Students self-reported their race: 15.1% Asian, 10.7% African American, 6.7% Latino or Hispanic, 14.9% Multiracial, 1.5% Native American, 1.7% Pacific Islander, and 44.9% White. Confirmatory factor analysis (CFA) was used to make initial refinements of the measures. Multiple group CFA models were then used to assess configural, metric, and scalar invariance across gender, language, and racial groups. Results are being used to redevelop the survey to better align with district goals.
Results: The CFA resulted in a five-factor model for climate and a two-factor model for social emotional skills fit the data well (RMSEA=0 .04; CFI=0.92; TLI=0.91), with sufficient factor loadings (0.50-.79) and internal reliability (Cronbach’s Alpha 0.72-0.85). The correlation between high-order global factors for climate and social emotional skills was r=0.61 (p<.001). Multiple group invariance for differences in measurement across racial/ethnic groups supported configural and metric invariance. Fit was substantially worse, however, for models testing scalar invariance model that included equality constraints on intercepts of indicators (RMSEA=0.04, CFI=0.91; TLI=0.91; Cheung & Rensvold, 2002), suggesting that some items differed across racial/ethnic groups in terms of their “severity” with respect to underlying constructs. These results are guiding the reformulation of the constructs of interest to the school district, to align to the local context and practice goals of schools.
Conclusions: Partnering with SPS resulted in a psychometrically stronger, practice relevant survey, making it more useful to schools as a tool for monitoring progress. Results suggested important areas not captured by the previous survey, such as the importance of identity safety and adult-student relationships especially for historically underserved students of color. Measuring constructs related to equity, and ensuring the survey is relevant to all student groups are first steps towards eliminating the opportunity gap.