Discrepancies between observers are common in studies of child behavior problems that rely on behavior ratings. Although modest concordance between informants is well-documented, little is known about characteristics that predict discrepancies. In 477 children aged 5 to 12 years, maternal age and indicators of socioeconomic status (SES; maternal education, family income) were evaluated in relation to score discrepancies between the Child Behavior Checklist and Teacher Report Form for Total, Externalizing and Internalizing Problems. Family income < $35,000 was independently associated with discrepancies in which mothers rated more clinically significant child behavior problems than teachers for Total, Externalizing and Internalizing Problems [odds ratio (OR) = 3.26, 95 % confidence interval (CI), 1.19–8.96, OR = 2.76, 95 % CI 1.03–7.34 and OR = 3.07, 95 % CI 1.30–7.26, respectively]. Maternal education was not associated with discrepancies, but younger mothers were less likely to rate child’s behavior in the clinical range for Externalizing Problems than teachers (OR per year of age = 0.88, 95 % CI 0.81–0.96). These results suggest that studies that utilize only maternal or teacher report of child behavior may have misclassification of outcomes that is dependent on SES and could produce biased results.
