Public schools have reported a notable increase in the numbers of students with autism served. To help schools provide more efficacious services for students with autism, information is needed on predictors of educational outcomes. Although information is available on pre-treatment child predictors (i.e., intelligence, language, social abilities, and autism severity), little information is available on caregiver and teacher predictors of school-based educational outcomes.
Objectives:
To identify the child, caregiver and teacher pre-treatment characteristics predictive of educational outcomes of children with autism.
Methods:
Thirty-five special education teachers participated in a randomized controlled study of a consultation intervention. Teachers were responsible for the individual educational programs (IEPs) of students with autism ages 3 to 8 years old. Following teacher recruitment, a student with autism was randomly selected for participation and the caregiver was recruited. Prior to group assignment, each teacher, caregiver, and child triad completed a comprehensive baseline evaluation at the start of the school year (Time 1). The experimental group consisted of 18 teachers. Both groups received a Time 2 evaluation at the end of the school year. Observational rating of child goal attainment of IEP objectives using curriculum based assessment at Time 2 was used to assess outcomes. The difference in the goal attainment score from Time 1 and 2 was used as the dependent variable. A correlation analysis was conducted to identify potential predictor variables from four child (age; IQ; language; autism severity), six teacher (years teaching autism; number children taught; stress related to child; child problem behavior; knowledge of autism; instructional engagement), and six caregiver (race; income; maternal education level; stress related to child; child problem behavior; teacher alliance) characteristics possibly associated with the dependent variable. Separate analyses for teacher, caregiver, and child variables were conducted using standard multiple regression.
Results:
Analysis of the correlation matrix indicated that of the 16 variables, no child, two teacher (stress; report of child problem behavior), and two caregiver (stress; report of child problem behavior) variables were significantly associated with educational outcome at p <.05. Two separate analyses were conducted for teacher and caregiver predictors separately (with a criterion of p < .05 to enter variables). Assumptions (multicollinearity, outliers, normality, linearity, homoscedasticity, and independence of residuals) were met for analysis. For teacher predictors, teacher report of child problem behavior was strongly negatively related to GAS improvement (Beta = -.551 for , p=.001); however, teacher stress did not remain significant in the regression (p = .085). For parent predictors, parent stress was strongly negatively related to GAS improvement (Beta = -.674, p = .007). Caregiver report of problem behavior was no longer significant once the impact of parent stress was regressed on the outcome (p = .808).
Conclusions:
Preliminary analysis of child, teacher, and caregiver pre-treatment predictors of educational outcomes suggest that teacher report of child problem behavior and parent report of stress related to the child were significant pre-treatment predictors of child educational outcome; surprisingly, pretreatment child variables found significant in other outcome research were not predictive in this outcome study.