Analysis Of The Predictive Validity Of The College Entrance Examination In English Mock Exam

Predictive validity is a kind of validity, which is a relatively new topic in the study of language testing. Since the 1950s, it has attracted more and more attention of language test researchers. Beginning with a review of the relationship between language teaching and testing, based on the theory of predictive validity produced by Hughes, Bachman & Weir, the purpose of the study is to investigate predictive validity on language testing.Based on the theory of predictive validity, the study aims to investigate whether the simulated NMET (National Matriculation English Test) for the senior three students plays a role of a good predicting, by analyzing the correlation between three scores on the simulated NMET and one score on the NMET about one senior high school. The study results will be of significant value to improve the simulation test quality and to improve senior English teaching and learning.The study will answer two questions:1) How well do the English scores on the simulated NMET predict the English scores on the NMET?2) What factors influence the accurate or inaccurate prediction of the simulated NMET?To answer the above questions, the author collects data from the simulated NMET taken by senior three students in the first senior high school in Wuyang, Luohe, all of whom graduated in July,2010. The study results are also analyzed and explained.The main findings are listed as follows:First, the English scores on simulated NMET can predict the students'English performance on the NMET. The correlation coefficients of the three simulated NMET and the NMET are statistically significant. The correlation coefficients of the art students on the three simulated NMET and NMET is different from that of the science students, which results mainly from the bad test quality of stimulated NMET.Second, a highly qualified stimulated NMET is very important, but the study results indicate that the quality of the simulated NMET test has a much room to improve, for the reasons are as following. Firstly, the total words of the reading comprehension are too long, which makes the students read more words in the same time. Secondly, the readability of the reading comprehension is too high or too low, which makes some passages are easy to do, while some are difficult.Third, the results of data analysis indicate that students'score on NMET is of great improvement, which is much higher than that on stimulated NMET.Four, the factors which influence simulated NMET are in many aspects, such as the high quality of the test, the short interval of the test, proper testing content can help the test predict well. In contrast, the bad quality of the test as well as some personal factors can lead to the inaccurate prediction.The study uses literature analysis, data analysis and paper analysis.