Because of the concrete operational cognitive stage of elementary students, it is difficult to create high quality child-report measures. Reading and comprehension skills of students in elementary school, including students eligible for special education services, vary greatly. School practitioners need to have confidence in the quality of data used to inform their practice decisions. This study's aim was to test the factor structure and quality of child-reported perceptions of adults collected with the online Elementary School Success Profile. The ESSP is a comprehensive social environmental assessment for 3rd through 5th graders. It collects data from children, parents, and teachers.
Methods: Data were collected from 3rd through 5th graders in eight schools in two North Carolina districts were combined for the study. Students in one school district were part of a longitudinal intervention study targeting all students in four schools in the 2007 third grade cohort. In the second district, a random sample of low performing students in all three grades was targeted for a one-year intervention. Data were collected on 451 students; 423 cases had child-report data.
Exploratory factor analyses and internal consistency reliability tests were conducted with SPSS 16.0. Confirmatory factor analyses were conducted with Mplus 4.2, using appropriate procedures for missing values, ordinal data, and the clustering of students in schools. Multiple group analyses and chi square difference tests were used to compare factor loadings for participants and non-participants in special education programs at the schools. Three ordinal subscales of the ESSP assessing child perceptions of caring adult neighbors, teachers, and parents were examined.
Results: The best fitting model was a first-order factor model with three factors—one for each social environmental domain. Multiple group analyses indicated that the factor loadings were not statistically significantly different for special education participants and non-participants. The CFI and TLI of the final model were both .988, the RMSEA was .047, and the WRMR was .896, indicating excellent fit. Standardized factor loadings ranged from .534 to .730 on the six items measuring perceptions of caring adults, .504 to .801 on the six items measuring perceptions of caring teachers, and .668 to .873 on 8 items measuring perceptions of caring adults at home. Alpha coefficients were .757, .775, and .877 respectively.
Implications: Psychometric tests using procedures appropriate for categorical, nested data indicated a strong factor structure for ESSP child-report data about caring adults in their lives. Importantly, the factor structure and loadings were invariant for special education participants and non-participants. School staff can confidently use the tool to assess the presence of caring adults in student lives.