ERIC Number: ED202902
Record Type: RIE
Publication Date: 1981-Mar
Reference Count: 0
An Analysis of the Individual Effects of Sex Bias.
Smith, Richard M.
Most attempts to correct for the presence of biased test items in a measurement instrument have been either to remove the items or to adjust the scores to correct for the bias. Using the Rasch Dichotomous Response Model and the independent ability estimates derived from three sets of items, those which favor females, those which favor males, and neutral items, it is possible to examine the individual variability of the ability estimates and the consequences of group correction efforts. In this study, in which the Logical Relationships test contained in the New Jersey College Basics Skills Placement Test was administered to approximately 47,000 freshman, it was found that only 43% of the women and 46% of the men had their highest ability estimates on the sets of items that supposedly favored their sex, while 34% of the women and 25% of the men had their highest ability estimates on the sets of items that were supposedly biased against them. Any correction for bias based solely on group membership would harm as many people as it helped. Correcting for bias must include some measure of the effect of bias on that individual. (Author/RL)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: New Jersey College Basic Skills Placement Test
Note: Paper presented at the Annual Meeting of the National Council on Measurement in Education (Los Angeles, CA, April 13-17, 1981).