Shedding light on the predictive validity of English proficiency tests in predicting academic success

Abstract

This embedded mixed method case study aims at shedding light on the use of English proficiency exams as placement tests and their viability as predictors of future academic performance. Most predictive validity studies achieve correlation coefficients in the range of 0.20-0.30 (In 1988, Davies suggested that 0.30 was an acceptable correlation for predictive validity studies.) when such exam results are compared with final course marks and/or GPAs, meaning that the results of language proficiency exams tend to have limited usefulness in admissions decisions. The Common Educational Proficiency Assessment (CEPA) is the focus of this research project. The results of the CEPA form a major part of admissions criteria for tertiary institutions in the United Arab Emirates (UAE). The CEPA has been reported to have achieved very high correlations (i.e. 0.699 in 2007) when compared to final first semester marks. This study examined this phenomenon at a large, vocational college. In addition, through an exhaustive exploration of college practices, and the input and opinions of a wide variety of stakeholders, creating a vivid picture of the context within which the CEPA operated, this study identified variables which may contribute to the success of the UAE CEPA as a placement instrument. The efficacy of using regionally- produced exams rather than internationally produced ones to not only gauge linguistic competence, but also to predict future success in an English readiness programme (required before matriculating) is considered. It is hoped that the results of the study may lead to improved predictive validity for regionally or locally produced placement tests at other institutions.