The predictive validity of the Academic IELTS test

Abstract Despite an abundance of studies that address the fundamental issue of the relationship between students’ IELTS results and their academic performance, findings remain inconclusive. One reason for this are variations in research design, notably in how researchers collect, analyse, and interpret data, and report their findings. The present study constitutes a critical methodological synthesis of 32 published IELTS predictive validity studies. The results revealed a number of efficacious design practices, and also some concerning tendencies, including: (a) dependency on linear models of causality, (b) small sampling frames situated in Anglo-Australian contexts, (c) a lack of thoroughness in describing contexts, samples, and approaches to academic performance measurement, and (d) where qualitative data was generated through interviewing and observation, a tendency to overlook ‘quality’ issues, such as confirmability, dependability, and trustworthiness. Recommendations are made for enhancing research designs along with suggestions for novel investigations to address gaps in the predictive validity literature.