英文學科能力測驗選擇題之性別差異與差異試題功能分析

學科能力測驗（簡稱學測）在臺灣為一重要而普及的大學入學測驗，因此該測驗是否具有信度、效度與公平性等議題，值得進一步探討。其中，影響測驗效度與公平性的可能因素之一即為考生的性別；不同性別的考生測驗表現是否存在顯著差異？尤其是性別不同，但能力相同的個人，是否在答對某個試題上的機率有所不同，因而呈現差異試題功能（DIF）的現象？換句話說，是否學測的考題可能會對某種性別的考生特別有利？然而目前文獻較少有針對大學學測英文考科是否存有性別差異與DIF試題的相關研究。因此，本研究主要目的即在比較男女考生在九十八學年度大學學測英文考科選擇題的測驗表現；並進一步採用M－H分析法（Mantel-Haenszel analysis）與logistic迴歸分析（logistic regression），進行差異試題功能分析檢測，逐一檢視各題是否因性別而存在DIF。研究結果顯示女性考生的英文學測成績略高於男性考生，並發現少數幾題具有性別DIF的現象，但尚不構成明顯的試題偏誤。

關鍵字

英文科學科能力測驗；性別差異；差異試題功能分析；測驗公平性；結構效度

並列摘要

The General Scholastic Ability Test (GSAT) is a widely-used college entrance exam in Taiwan. It is thus more critical than ever that the test scores be reliable and that the inferences drawn from the results be valid and fair. One of the factors influencing test validity and fairness is gender. The major concern is to examine if there are significant gender differences in test scores and if there are items which favor a certain group. In other words, if female test takers with the same trait level as male counterparts have different expected scores on the same item, differential item functioning (DIF) may be present in the item. To date, these issues are yet to receive a lot of attention in the English GSAT context. The aim of the current study is to investigate gender differences in the 98 English GSAT multiple-choice scores and detect gender DIF items using the Mantel-Haenszel method and logistic regression analysis. The results showed that female test-takers scored slightly higher than males in the English GSAT. Few potentially gender DIF items were flagged, but no systematic bias was detected.