Using Parental Ratings to Study the Etiology of 3-year-old Twins' Problem Behaviors: Different Views or Rater Bias?

Abstract

Child Behavior Checklist questionnaires (Achenbach, 1992), filled in separately by mothers and fathers, were collected for an effective sample of 3501 Dutch 3-year-old twin pairs. To disentangle the child's phenotype from that of the rater, two contrasting models were fitted to the data. One model, called a Rater Bias model, is based on the assumption that both parents assess exactly the same behaviors in the child. A weaker alternative of this model, called a Psychometric model, assumes that apart from these common behavioral views, each parent also assesses a unique aspect of the child's behavior. A Psychometric model fitted the data of both Internalizing and Externalizing scales significantly better than a Rater Bias model. This implied that each parent provided unique information from his or her own perspective, apart from the common behavioral view. Using this best fitting model, the etiology of both the Internalizing and Externalizing scales was studied. Common factors (influencing behaviors similarly assessed by both parents) were more important than unique factors (influencing behaviors uniquely assessed by one parent). Common genetic factors explained about 50% of the variance of both scales, indicating a possible inborn vulnerability to childhood psychopathology. Common environmental factors not shared between twins (free of unreliability and error) explained around 14% of both scales, suggesting the importance of pure idiosyncratic experiences even for children as young as 3 years. Common environmental factors shared between twins (unconfounded by rater bias) were only found for the Externalizing scale, explaining 18% of the variance. Rater bias and unreliability, if present in the data, were included in the estimates of the unique factors. Unique genetic, shared, and nonshared environmental factors each explained around 8% of the variance for both scales. These small effects could be detected because of the large sample of twin pairs used.