Gender differences in performance on multiple-choice and constructed response mathematics items

The purpose of this study was to examine gender differences in performance on multiple-choice and constructed response items in mathematics, administe red within the context of a state high school graduation test that was desi gned to match a specified curriculum. A secondary purpose was to demonstrat e a method based on the many-faceted Rasch (1980) measurement model for com paring performance on different item types and exploring differential item functioning (DIF). A random sample of 3,952 eleventh graders who took the 1 994 Georgia High School Graduation Test was used for the analysis. In both mean scores and DIF indexes, women showed a statistically significant and c onsistent advantage over men on multiple-choice items involving algebra, wh ereas men showed a less consistent advantage on items involving geometry an d measurement, number and computation, data analysis, and proportional reas oning. Mean scores were significantly higher for men than for women on 2 ou t of 8 constructed response items. However, when men and women were statist ically matched according to ability, the only significant difference in per formance on constructed response items was in favor of women. It was conclu ded that gender differences in mathematics may well be linked to content an d item format.