Educational Research Journal

Abstract

This paper examines the construct validity of a cloze test using two sources of evidence. The first derives from an analysis of the relationships among test scores obtained from four versions of a cloze test. Item analysis, analysis of variance and confirmatory factor analysis were performed on the test data. The second relates to the examinees' oral accounts of their thinking processes as they answer the cloze items. These two sources of evidence together show that examinees first tried to answer the cloze item making use of wellrehearsed or automatic processes. If this was not successful 1992, they then used analytic or controlled processes to attempt the items. Items which were best at assessing examinees' automatic processes were constructed by deleting prepositions from the passage whereas those items best for assessing controlled processes were constructed by deleting content words. These two types of items also differed in their level of difficulty and assessed different aspects of English language ability.

_{[ Chinese version | Index | Educational Research Journal | Other Journals | Subscription form | Enquiry ]}

Mail any comments and suggestions to hkier-journal@cuhk.edu.hk.