An Empirical Study On The Validity Of The Three Reading Item Types In The Effectiveness Of Testing Reading Subskills

English reading comprehension has mostly adopted the traditional and singular reading item type, the MCQ for a long time. However, with the development of language testing theories, more and more researchers have disputes about the validity of the current way of testing reading. Opinions on whether different reading item types have influence on test-takers'reading performance differ greatly. Some researchers reach the conclusion that different reading item types are effective to test different reading abilities or the components of a kind of reading ability. With reference to previous studies, it is known that there is hardly any research about English major university students in China. No sufficient studies were carried out to support their viewpoints. Therefore, there is a strong requirement to conduct this kind of study in China.Thus this study selects 106 sophomores of English major from Hunan University as subjects to investigate the validity of different reading item types in the effectiveness of testing reading subskills. A formal experimental reading test and a questionnaire about reading test are carried out to answer these four questions: (1) Which reading item type (MCQ, T/F or SC) has more validity? (2) Is test-takers'performance of reading test influenced by the item types? (3) What influence does each item type have on test-takers with higher reading abilities and test-takers with lower reading abilities respectively? (4) What reading subskills is each item type effective to measure? In order to answer those four questions, data for validity, for formal experimental reading test scores and for reading ability are collected. This study based on the procedures for validation evaluation proposed by Alderson et al. applies the descriptive analysis, Pearson correlation, one-way ANOVA and two-way ANOVA by making use of SPSS 13.0. It also makes item difficulty analysis of the reading test papers and the inter-rater reliability analysis. To ensure the reliability and validity of the test papers, a pretest is carried out in the control group before the test papers are adopted.Experimental results show that different item types do have influence on the subjects'performance. Item type effect on subjects with lower reading abilities is greater than on the ones with higher reading abilities. MCQ has the most face validity, and it is is the most effective item type to measure different reading subskills to identify facts and details and to guess the meanings of unknown words and phrases; T/F has the most response validity, and it is the best item format to assess subskills to grasp the main ideas and gist; SC has the most concurrent and predictive validity, it has the most reading validity in a whole, and it is the most appropriate method to test the subskills to make inference through the contextual clues and to understand the logical reference.This study provides some suggestions on ELS reading testing and language teaching and learning. This study hopes to find out a new way to improve the accuracy of measuring reading subskills and provide some suggestions for the reform of the reading comprehension in TEM4 and TEM8 in future.