ERIC Number: ED286914
Record Type: RIE
Publication Date: 1987-Apr
Gender and Race Effects on Standardized Tests Predictive Validity: A Meta-analytical Study.
Sanber, Shukri R.; Millman, Jason
This paper summarizes and analyzes data from studies of differential validity and differential prediction for blacks versus whites and females versus males. Forty studies on racial differences and 53 on gender differences were selected from searches of the ERIC, PsychInfo, and Dissertation Abstracts International databases; Buros' Mental Measurements Yearbooks; and major test reviews. All data were derived from standardized achievement tests. Two questions were investigated: (1) is there empirical evidence of differential predictive validity on race and/or gender?; and (2) is there an under- or over-prediction of black or female students' performance when combined regression equations are used? Results of the meta-analysis showed: (1) averaged validity coefficients were higher for whites than blacks, and for females than males, but the differences were too small to be meaningful; (2) averaged regression coefficients were higher for whites than blacks and for males than females; and (3) slopes were equal in 78% of the black/white comparisons and 81% of the male/female comparisons. In 41% of these black/white comparisons, intercepts were higher for blacks, and in 53% of the male/female comparisons intercepts were higher for females. It was concluded that, despite small differences in regression equations between the racial and gender groups, tests as predictors of future academic performance areas valid for blacks and females as they are for whites and males. (A list of studies included in the synthesis and a six-page bibliography are included.) (JGL)
Publication Type: Speeches/Meeting Papers; Reports - Research; Information Analyses
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Annual Meeting of the American Educational Research Association (Washington, DC, April 20-24, 1987).