Selective attrition does not affect cross-sectional estimates of associations with emotional and behavioral problems in a longitudinal study with onset in adolescence

Abstract

The present examination raises the question whether attrition in a longitudinal study leads to biased findings. The Zurich Adolescent Psychology and Psychopathology Study (ZAPPS) originated in 1994 by following a sample of 1239 adolescents with follow-ups after three, six, and 22 years. The study collected data on life-events, self-esteem, and coping behavior as independent and mental problems as dependent variables. The baseline sample was partitioned according to follow-up status for the three subsequent waves of assessment. Baseline measures of associations between independent and dependent variables were compared for those retained in the study (group A) and those lost to follow-up (group B) at each phase of cross-sectional data collection by use of multiple linear regression analyses. There were significant differences for some baseline independent variables between the two groups with small effect sizes. Males and migrants dropped out more frequently. The main findings indicated that the strength of the associations between independent and dependent variables at baseline in the total sample and in each of the two groups as defined by sample sizes at the three follow-ups was close to equal. Thus, one may conclude from the current study that attrition in longitudinal studies rarely affects cross-sectional estimates of associations.