Passage Dependence of Multiple-Choice Items in Reading Comprehension Component of English Academic Test of Guangdong Province

Abstract

With the deepening of the reform in examination and enrollment system in China, the English Academic Test plays an increasingly important role in selecting the secondary school graduates for institutions of higher vocational education. Therefore, it is of great significance to ensure the quality of the test. A substantial body of research has shown that test takers are able to correctly respond to reading comprehension (RC) multiple-choice (MC) items without reading or comprehending the accompanying passage (Hannon & Daneman, 2001; Katz, Blackburn, &Lautenschlager, 1991; Lifson, Scruggs, & Bennion, 1984), confirming that the passage dependence issue does exist in the testing of RC. The term passage dependence refers to the extent to which students can correctly answer the RC questions depending on whether they have read the accompanying passage. This study seeks to determine the extent to which the RC MC items in the English Academic Test is passage-dependent from the perspective of passage types and item types. A quantitative analysis (descriptive statistics, Independent-Samples T Test, One-Sample T Test, Chi-Square Test) of performance in RC component of the English Academic Test 2017 and 2018 were conducted on 343Grade 11 students in two senior secondary schools in Guangzhou, China. The results of the present study are as follows： (1) Participants’ scores of the items for both narrative and expository passages under the with-passage condition were significantly higher than the scores obtained under the without- passage condition. Furthermore, the items associated with narrative passages were more passage-dependent than the items associated with expository passages. (2) Factual item type was more dependent on the passage than the inferential item type. (3) There was a significant relationship between item types and Passage Dependence Index. In addition, the main idea item type and vocabulary-in-context item type were the least passage-dependent item types. The result of passage types indicates that the RC component of the English Academic Test was with relatively high validity regarding passage dependence. Students and teachers are advised to be aware of the fact that it is impossible to gain success in the RC test simply by applying test-wise strategies. Emphasis should be made on the development of students’ RC ability. Meanwhile, the result of item types provides empirical evidence for the test designers for improving the quality of the test.