Objectives: The goals were to (1) determine if the CPI was a reliable measure and could discriminate between the comprehensive treatment models of TEACCH or LEAP as well as discriminate those models from an eclectic, non-model specific treatment approach (“Business as Usual or BAU”) based on the classroom practices endorsed by those teachers, (2) provide descriptive information on the types of practices teachers reported they used across model types, and (3) explore variables that predict teachers’ selection of specific classroom practices.
Methods: 24 teachers across 3 states (9=TEACCH, 6=LEAP, 9=BAU) completed the CPI at the beginning and end of the school year as part of a larger study. In addition, data related to number of years teaching and highest level of education attained were collected. The reliability of the measure was determined by examining test-retest reliability and internal consistency. Discriminant analysis was conducted to identify the classroom practices on the CPI that best discriminated between classroom types. Regression analyses were conducted to determine which demographic variables predicted teachers’ use of specific classroom practices.
Test-retest reliability for pre/post-test time points was M = .80 (.29-.95 across 10 subscales)
Internal consistency was examined using Cronbach’s alpha and was .77.
Discriminant analyses revealed two canonical variates. Canonical variate 1 significantly discriminated the three models from each other based on the practices endorsed by teachers. The practices that discriminated the models included Floortime and social stories (endorsed most by BAU teachers) as well as discrete trial training and peer-mediated instruction (endorsed most by LEAP teachers). Canonical variate 2 significantly separated LEAP from BAU and LEAP from TEACCH.
Regression models indicated that number of years teaching children with autism and education level were not significant predictors of teachers’ selection of classroom practices. Only two practices, structured teaching and social stories were negatively associated with years teaching (structured teaching, -.145, p <.05) and level of education (social stories, -1.61, p <.05).
Conclusions: We found that the CPI is a psychometrically robust measure and discriminates between intervention models. Both the development of this measure and the analysis of its psychometric properties are valuable contributions to classroom-based intervention research for children with ASD. Adequately describing and measuring additional practices in use during treatment evaluation is critical to explain intervention effects.