本研究探討國小高年級數學科成就測驗中的性別DIF現象。試題以國立編譯館主編國小六下數學課本及習作為選題範圍，以選擇題呈現，施測對象為臺灣北部一所國小全體六年級學生677位，以ConQuest及EZDIF兩程式進行DIF檢定。研究結果為：(1)女生群體的數學平均得分顯著高於男生群體。(2)數列規律及位值關係兩試題不利於女生群體。(3)以去除DIF 試題的純化總分為分組變項，DIF容易發生在得分接近平均數的男、女次群體之中。(4)以去除DIF試題的純化總分為分組變項，DIF的產生與平均數似乎無關。最後最出教學與研究上的議議：(1)在教學上，教師宜注意程度居中學生的學習過程，尤其是解題較不利的女生。(2)在研究上，則可從教學情境、試題難度、試題脈絡等三方面探討與DIF試題的關聯性。

The purpose of this study is to examine the gender DIF in a sixth grade mathematics achievement test. Test items, presented in multiple-choice format, were prepared based on an edition of mathematics textbooks published by the National Institute for Compilation and Translation. Six hundred and seventy-seven students in an elementary school located in Taipei County participated in the study. ConQuest and EZDIF computer programs were applied in data analysis to conduct DIF detection. The results are as follows: (1)girls’ average score was significantly higher than of that of boys’; (2)items about the concepts of “sequence” and “place value relation” were more difficult for girls; (3) DIF ore likely to occur in the gender subgroups for those scores are close to total mean of raw score; (4) DIF’s occurrence seemed to have nothing to do with purified total score. Based on the above findings, the researchers provide suggestions for math teachers and future researchers.