Prediction of student achievement at A.N.U. from A.C.T. college assessments

Abstract

Relationships between measures of“ student performance in A.C.T. colleges and results achieved in first year at A.N.U. are investigated. The college variables are those produced by the A.C.T. Schools Accrediting Agency and made available to A.N.U. for admission purposes, namely Tertiary Entrance Score (TES), Australian Scholastic Aptitude Test (ASAT) and Principal's Recommendations. There are three ASAT variables, Verbal, Quantitative and Total, and three Principal's Recommendations, Academic, Personal and Overall. Sex and college attended are also considered as possible variables to predict A.N.U. results. Analyses were performed on three years of data; students entering A.N.U. from A.C.T. colleges in 1978, 1979 and 1980. There were 241, 258 and 272 such students, respectively, in these years. Results of all of these students in all first year units attempted were available, as well as actual marks obtained in the majority of cases. Correlations and regressions of marks in individual A.N.U. units against college variables showed that, for all faculties except Law, TES was the best predictor of mark obtained. Addition of other college variables significantly improved the regressions for some units, however not consistently over the three years. Sex was significant in a few instances. In the Faculty of Law, ASAT was generally a better predictor than TES. Grade scores were defined to summarise results obtained by students in large groups of units, based either on faculty or subject area. As for marks, TES was the best predictor of grade score, and addition of other variables significantly improved the relationship in very few cases. Sex was still significant for a few grade scores and ASAT was still a better predictor than TES in the Faculty of Law. The larger number of student results summarised by grade score meant that the effect of A.C.T. college attended could be tested. There were substantial differences between colleges in 1978 and 1979, when addition of college as a variable significantly improved regressions with grade score. However no such differences were obtained in 1980, and the inconsistency of results over the three years means no general inferences can be made on relative performances in relation to college attended. Implications of the results for A.N.U. admission procedures and possible further studies in this general area are discussed. Show more