Teachers' Rating of Classroom Competence and Correlations with Objective Measures of Peformance

Teachers have primary responsibility for rating children's classroom and academic performance and these teacher ratings influence many important aspects of a child's academic trajectory. Prior research suggests that teacher assessments correlate moderately well with objective measures of performance such as standardized test scores. However, there is concern that teacher ratings may systematically vary across groups of students defined by ethnicity, gender, or classroom behavior. This analysis used a latent profile model to derive discrete latent classes of students based on teacher ratings of classroom behavior and academic competence. The variables included in the model were cooperation, assertion, self-control, externalizing problems, internalizing problems, hyper-activity, and academic competence. A three class model provided the best fit; the three latent classes were high achievers, moderate performers, and disruptive/low performers. The mean value for academic competence was lowest in the disruptive/low performers.  After determining the latent class structure, special education services and standardized test scores for reading and math were included as distal outcomes to the model to determine how these three measures varied as a function of class membership.  Wald tests were used to test for between-group differences for each of the three variables.  The probability of receiving special education services for students in the high achieving class was significantly lower than either the moderate performers or the disruptive/low performers class. However, the proportion of students receiving special education services in moderate performers and the disruptive/low performers class was equivalent.  For the standardized reading test scores, the proportion of students ranked as basic, proficient, and advanced was equivalent across all three latent classes of students. For the standardized math test scores, the proportion of students with basic test scores was lower in the high achieving class but was comparable in the moderate performers and the disruptive/low performers classes.  Likewise, the proportion of students with proficient math scores was also comparable between the moderate performers and the disruptive/low performers classes. However, the proportion of students with advanced standardized math scores was higher in the disruptive/low performers classes relative to the moderate performers (but was significantly lower than the high achieving class). The findings are consistent with past research which suggests that teachers are more accurate at rating student performance for high achieving students, but the accuracy of teacher ratings decline as students express higher levels of problem behavior.