Whether specification of focal group membership based on ethnicity as well as educational opportunity affects differential item functioning (DIF) of test items was studied. Subjects were 4,342 freshmen admitted to a large midwestern university who completed the 40-item Advanced Mathematics Placement Examination in the spring of 1991. There were 3,896 White, 338 African American, and 108 Hispanic American examinees. All of the Hispanic Americans and 166 of the African Americans were from high schools judged to be educationally and economically disadvantaged. The Mantel-Haenszel (MH) procedure was used to examine DIF. Although there were few verbal items, the mean MH DIF delta difference did change substantially in relation to focal group specification. There were also changes in item classification depending on the specification of focal group membership. The question of whether sample sizes for focal group measurement were adequate in this investigation is discussed. There are five tables of study data and one illustrative plot. A 23-item list of references is included. (SLD)