A Study To The Construct Validity Of Computer-based Cet 4 Test

Even though construct validity is acknowledged as the most important and essential one based on Messick's framework of"unitary concept of validity"in language testing, empirical studies on construct validity of language tests are rare among Chinese scholars. From previous construct validity studies of other large-scale standardized tests (TOEFL in particular), a combination of Exploratory Factor Analysis and Confirmatory Factor Analysis is a preferred solution to establish the corresponding language competence structure model. As we know that the Computer-based CET 4 (CET4-CBT) has been under development and applied to field tests gradually, it requires building a validity argument for the new test by gathering various types of empirical evidence. The present study contributes such evidence from the construct validity perspective.This empirical construct validity study investigated the language competence structure of the CET 4-CBT. Correlation Analyses, Exploratory Factor Analyses and Structural Equation Modeling were conducted for a field test completed by 289 participants in December, 2008. The computer-based test was proved to be rather comparable with the parallel paper-based test, with better internal correlation structure than the traditional test. Among three alternative models (a single general factor model, a correlated two-factor model, and a higher-order factor model), the higher-order factor model that included a single higher-order general factor (EFL ability) and two first-order factors ("Listening"and"Reading") was identified as the best representation of the test's language competence structure. The newly-introduced listening-based integrated tasks in CET 4-CBT were shown to load heavily on the Listening factor but correlate little with other language ability (such as reading or writing) or prompts in the Listening section. Some beneficial recommendations were also provided for the improvement of the on-going CET 4-CBT test.