This study examines gender differences on the American Mathematics Competition AMC 8 contest between 2003 and 2007 by comparing the performances of male and female United States eighth grade students after controlling for ability. During these years 183,857 males and 178,857 females participated in the contest. Research on gender differences frequently measures impact which is a difference in performance between two groups that can often be explained by different ability distributions. In contrast, differential item functioning (DIF) is a difference in performance after controlling for ability. Three types of analyses were performed to compare the performances. First, statistical analyses identified items with impact, DIF, and uniform or nonuniform DIF. Differences in proportion correct were used to identify impact and type of DIF while the Mantel-Haenszel procedure was used to identify items with gender DIF. Second, substantive analyses placed the items into multiple categories based on NCTM, Gierl, and Harnisch's classifications of mathematics problems. Third, subtest analyses used these categories to look for gender differences in terms of impact and DIF on subsets of the contest. While a majority of the items favored males in terms of impact, after controlling for ability, few items demonstrated gender DIF. None of the hypotheses of differing abilities in males and females suggested by earlier studies were supported by the subtest analyses.
