ARE ETHNIC-DIFFERENCES IN DIAGNOSIS OF CHILDHOOD PSYCHOPATHOLOGY AN ARTIFACT OF PSYCHOMETRIC METHODS - AN EXPERIMENTAL EVALUATION OF HARRINGTONS HYPOTHESIS USING PARENT-REPORTED SYMPTOMATOLOGY

Based on a series of animal studies in the 1960s and 1970s, Harrington (1968, 1975, 1984, 1988), a developmental psychobiologist, hypothesiz ed and subsequently provided what he believed was strong empirical evi dence that the mean differences in scores evident on various measures of aptitude among different ethnic groups were the result of flawed ps ychometric methods used in test item selection. Since test-items tradi tionally are selected on the basis of total sample statistics, Harring ton argued that the selection of items is affected primarily by the Wh ites (the numerical majority of a population proportionate sample) due to problems with classical test theory that cause items to favor the numerical majority of any test development sample. The present study c onstructed a human experimental model to test empirically Harrington's hypothesis of race differences on a test assessing behavior and used data from the Parent Rating Scale (Adolescent form; PRS-A) from the Be havior Assessment System for Children (BASC; Reynolds & Kamphaus, 1992 ). Three separate forms of the PRS-A were developed in which each ethn ic group completely and separately was the test development population . According to Harrington, by developing homogeneous ethnic tests, eac h ethnic group should score in the direction of less pathology on thei r dedicated test. Results indicate that Harrington's proposed effect i s not evident for this set of behavioral data. (C) 1998 Society for th e Study of School Psychology. Published by Elsevier Science Ltd.