Validating the highest performance standard of a test of academic literacy at a South African university

Abstract

Poor graduation rates are a serious concern worldwide. In South Africa, this concern has escalated in the post-apartheid era wherein a democratic constitution has widened access to higher education for school-leavers. The socioeconomic and school backgrounds of the majority of these learners still hamper their timely completion at university. In order to combat this, local universities have implemented some necessary interventions. Such interventions are geared towards dealing with the academic language needs of incoming students. For the last two decades or so, standardised tests of academic language ability, now commonly known as academic literacy, have been used to determine these needs. Given the expected impact of these interventions on student completion rates, the importance of the validity of these tests cannot be overemphasised. The aim of this article is to investigate the validity of the highest performance standard set for one of the tests currently used to assess levels of academic literacy. Using 14 610 scores obtained on that test by first-year students at a South African university, in tandem with their average scores on completion of their first year, sensitivity and specificity statistics were computed to realise this aim. The results revealed that the performance standard investigated was valid 61% of the time.