A Study Of Test Bias In English Reading Test

The issue of test bias has been an active area of research for decades. Growing interest has been in the factors that might give rise to bias. The investigation of the causes of bias helps to identify whether the differences in test performance between two population groups are the result of genuine differences in ability or the characteristics of test that have systematically different effects on persons in different groups. This paper aims at making a study on the potential bias in English reading test for Tibetan students. Reading serves as a crucial part to evaluate students'language competence. If the reading test is contaminated by combined factors that are irrelevant to the construct of the test, the scores of the reading test can not be meaningfully interpreted as a manifestation of the test taker's language ability. The type of response of the reading test and the language used to assess reading comprehension in a foreign language are considered as two of the factors that may work to inflate score differences between two population groups. In view of the skepticism towards the validity of the multiple choice questions (MCQs) test format, the present study adopted a new test format—MCQs plus summarization to test reading comprehension. As to the language aspect of the reading test, Tibetan ethnic group provides a perfect example to verify whether indigenous language is one of the factors that creates test bias, because one striking difference between Tibetan nationality and Han majority is the language.Instead of being confined to the analyses of testing instruments that the previous researches had focused on, this study has incorporated the psychometric approaches as well as the socio-cultural approaches to touch upon the complex issue of test bias within a broad multilingual and multicultural context. From psychometric approaches, both quantitative and qualitative methods were employed. Two reading tests which were composed of the original passages and translated reading comprehension questions of PETS reading subtests, the questionnaires and the interviews were conducted to detect the artifacts of bias. From socio-cultural approaches, further investigation of China's language policy, minority education and practice for Tibetan nationality has helped to explain the reasons for the results of the tests, questionnaire survey and interviews.Based on a comprehensive way of probing into the issue, the conclusions that the present study has reached might be able to furnish the researchers with an enriched understanding of language testing for ethnic groups. Although test bias in English reading test has not been found arising from Tibetan students'indigenous language, this study has contributed to the discussion of test bias and it has taken us one step closer to a better understanding of the complexity of this issue. It is also hoped that a case study on Tibetan nationality will give voice to the issues related to the ethnic minority in China. Yet in the light of the limitations of present study, it is less powerful in generalization. Therefore, the future studies are expected to develop theories and methodologies to enrich the understanding of test bias within the broad Chinese minority education context.