著者
杉野 直樹 斎藤 栄二 高橋 貞雄 清水 裕子 根岸 雅史 野澤 健 石塚 智一 内田 照久 前川 眞一
出版者
全国英語教育学会
雑誌
ARELE : annual review of English language education in Japan (ISSN:13448560)
巻号頁・発行日
vol.14, pp.221-230, 2003-03

The present study aims at explicating the influence of test taking strategies on the test item reliability in English language proficiency tests. Widely known test taking strategies include starting with a certain section that might require more time than others to answer so that test takers can allocate more of their time allowance on it, or, especially in multiple-choice format, marking an alternative based on wild guessing. Also widely speculated is that, in the Daigaku Nyushi Center English test (DNC test, henceforth), those test takers who find themselves running out of time are forced to rely on wild guessing in answering test items. Some English language proficiency tests, such as TOEFL or TOEIC, strictly instruct test takers to tackle a specific section so that they cannot use the first strategy, however, many of the entrance examinations administered in Japan do not have such restrictions. In order to examine the influence of these two test taking strategies, viz. the 'answering order' strategy and the 'wild guessing' strategy, we conducted a large-scale survey using two parallel tests with different question/answering orders. Our analysis of the data with simulated wild guessing shows that wild guessing would deteriorate the test item reliability. Furthermore, it shows that those test takers who had more time to answer the same section seem to be using the wild guessing strategy anyway, which has significant implications on the test format itself.