A correction for recruitment bias in norms derived from meta-analysis.

Abstract

Normative comparisons are an integral component of neuropsychological test interpretation and provide the basis for an inference of abnormal function and impairment. In order to remedy a deficit of normative standards for a large number of neuropsychology tests, Mitrushina, Boone, Razani, and D'Elia (2005) used the meta-analysis of studies that incorporated normal volunteers to create a type of normative standard for many tests in neuropsychology that were not adequately normed in the past. The present study examined this method by contrasting meta-analysis norms of the Wechsler Adult Intelligence Scale-Revised (WAIS-R; Wechsler, 1981) with its published quota norms. The study examined 12 experimental studies of the WAIS-R that included normal, control volunteers (N = 2,147). These were combined by meta-analysis, and the summary scores were compared. The meta-analysis revealed a significant sampling bias with studies recruiting more White and higher educated respondents than indicated by the U.S. Census. This bias was successfully corrected using Monte Carlo simulation and adjustments for quota sampling. The corrections could be applied to all meta-analysis norms currently in use and bring them in line with the U.S. Census demographics.