Investigation of rater, gender, and age biases to test formats:FTFI and COT

초록

Korean Journal of English Language and Linguistics 8-1, 1-20. The employment of computers for language test delivery by academic institutions and companies is increasing. But the changing mode of test delivery does bring certain risks. The primary risk is that the new test format could diminish the validity of what the test is meant to measure. For a test to have genuine validity, test results obtained should not be affected by factors other than test takers' language ability. In the computerized test the test format itself may influence test takers' performances or raters' scoring, with potentially serious results. Thus, to evaluate test fairness and validity, the current study investigated not only the severity of raters and the difficulty of a Computerized Oral Test (COT) compared with a Face-To-Face Interview (FTFI), but also potential rater, gender, and age biases to the test formats with multi-faceted Rasch (bias) analyses. The results of the study revealed significant differences in the severity of the raters and the difficulty of the two tests, but found no meaningful gender and age biases to the test formats. Significant rater bias to the test formats was also found in this study, but it should be noticed that only two raters were employed in this study. In a larger and more diverse sample, a more convincing or even different result might have been arrived at. Thus, further research on this issue needs to be undertaken.

키워드