| Summary: | 碩士 === 東海大學 === 外國語文學系 === 101 === This study examined the construct validity of the Tunghai English Placement Exam (TEPE) by comparing the test structure of the TEPE as intended by the test designers with the structure stemming from student performance on the test. Conducting research on the construct validity of the TEPE was to show the aptness of the decisions and inferences made by teachers with regards to the students’ performance on the test. A secondary objective was the stability of the factor structure over repeated exam administrations. The study used TEPE test data collected from 18,371 freshmen non-English majors enrolled in Tunghai University from 2006 to 2011. Exploratory and confirmatory factor analysis (EFA and CFA) through the use of SPSS determined the construct validity of the TEPE. It was hoped that by investigating the construct validity of the TEPE, that more knowledge can be shed about the usefulness of the test with regards to student assessment, if there were any changes of the characteristics of the test-taking population, and about the multidimensionality of English language ability. EFA and CFA results favored a three-factor model, which mirrored the test designers’ intended test structure for the most part. One significant change however, with the model in the study was that one of the grammar sections of the TEPE grouped more closely with the reading factor, instead of the grammar factor as anticipated by the test designers. Nonetheless, the test structure itself remained constant for all years, as indicated by the minor variations in factor loadings and model fit indices. Lastly, the study indicated a conflicting view on the topic of multidimensionality of language, as the results showed that grammar and reading were more interconnected than anticipated. The research demonstrated the factor structure of the TEPE gave support for the construct validity of the exam. However, because of an anomaly with one of the grammar sections, any interpretations of the TEPE scores must be more closely examined when distinguishing between grammar and reading scores.
|
|---|
