An examination of bias in volunteer subject selection: findings from an in-depth child abuse study.

Abstract

Remarkably few reported studies tested the assumption that a research sample can be constructed which is representative of the population of interest. In order to investigate potential volunteer bias in abuse research, we utilized a database assembled for an NIMH funded study investigating the relationship among adolescent physical abuse, suicidal behavior, and psychopathology. Extensive information was available concerning the nonparticipant pool from which this sample was assembled, allowing for a comprehensive assessment of possible sample bias. The volunteer sample of 99 abused families who agreed to participate in our study was compared on a large number of variables with a random sample of 99 abused families who declined to participate. Comparisons of the two groups did not support the hypothesis that the non-participating families represented a more dysfunctional population. The two groups were far more similar to, than disparate from, each other.