Testing sampling bias in estimates of adolescent social competence and behavioral control

Abstract

In 5 of the 6 large Dutch developmental cohorts investigated here, lower SES adolescents are underrepresented and higher SES adolescents overrepresented. With former studies clearly revealing differences between SES strata in adolescent social competence and behavioral control, this misrepresentation may contribute to an overestimation of normative adolescent competence. Using a raking procedure, we used national census statistics to weigh the cohorts to be more representative of the Dutch population. Contrary to our expectations, in all cohorts, little to no differences between SES strata were found in the two outcomes. Accordingly, no differences between weighted and unweighted mean scores were observed across all cohorts. Furthermore, no clear change in correlations between social competence and behavioral control was found. These findings are most probably explained by the fact that measures of SES in the samples were quite limited, and the low SES participants in the cohorts could not be considered as representative of the low SES groups in the general population. Developmental outcomes associated with SES may be affected by a raking procedure in other cohorts that have a sufficient number and sufficient variation of low SES adolescents.