PDF pending restoration
ERIC Number: ED181077
Record Type: RIE
Publication Date: 1979-Sep
Sex Bias in Educational Aptitude and Achievement Tests: What Do We Know? Draft.
Lockheed, Marlaine E.
Empirical studies are reviewed of sex bias and psychometric bias in aptitude and achievement tests. It is concluded that: (1) of four studies of sex bias, all found that the language used in the tests favored males, but, at the elementary-secondary level, sex bias has been decreasing; (2) of 15 analyses of psychometric bias, all found an item-by-sex interaction effect on performance; (3) of two studies which examined both sex and psychometric bias, one found a relationship between sex bias and sex differences in performance; (4) no evidence that changes in sex bias relate to changes in psychometric bias was noted; and (5) most recent studies of sex bias divert attention from the relative performance of males and females to the item difficulty within sex group; but do not point out that males outperform females overall on mathematics and science tests and that there are basically no items (by content or item type,) on which females outperform males. However, the author cautions interpreting this to imply innate sex differences in mathematical or scientific abilities because it has been determined that, given equal training, such differences disappear. (Author/MH)
Descriptors: Achievement Tests, Aptitude Tests, College Entrance Examinations, Difficulty Level, Educational Testing, Elementary Secondary Education, Females, Higher Education, Item Analysis, Measurement Techniques, Psychometrics, Research Reviews (Publications), Sex Differences, Sex Discrimination, Test Bias, Test Construction, Test Interpretation, Test Items, Test Validity
Publication Type: Speeches/Meeting Papers; Information Analyses
Education Level: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A
Note: Paper presented at the Research Conference on Educational Environments and the Undergraduate Women (Wellesley, MA, September 13-15, 1979) ; Appendix may be marginally legible