Race, gender and omissions on standard achievement tests

Abstract

The purpose of this study was to examine the effects of race and gender on omissions in multiple choice tests.

Modest but significant (p <= .05) correlations were observed between gender and omissions and between race and omissions on the mathematics subtest of the Tests of Achievement and Proficiency, a standardized achievement test administered to all eleventh graders in the Commonwealth of Virginia. Item characteristics (difficulty, discrimination and an index of differential item functioning) were used as independent variables in regression equations for the male/female case and the black/white case. In both cases item difficulty was the only significant (p <= .05) predictor of omissions.

Principal components analysis was used to create composite variables characterizing school divisions. These composites together with race (proportion of black students) were used as independent variables in a regression equation with omissions as the dependent variable. Race was the only variable which was a significant predictor of omissions.