Revisiting test stability: further evidence relating to the measurement of difference in performance on a test of academic literacy

Abstract

In several earlier analyses of two tests of academic literacy – the Test of Academic Literacy Levels (TALL) and its Afrikaans counterpart, the Toets vir Akademiese Geletterdheidsvlakke (TAG) – we have adopted an approach to the problem that tests may be abused (and therefore used to harm people) by discussing various antidotes to this, so as to ensure fairness and consistency in the tests we use, as well as by demonstrating how the process of test development and implementation might be made more transparent. We have pointed out that a true measure of the stability of such tests may only become apparent when we have data stretching across a number of years. This article reports on an investigation of differing measures of performance on the two tests in question across several of their administrations between 2005 and 2008. We point to a number of ways in which empirical measures may be used to achieve benign, enabling effects on those tested.

Southern African Linguistics and Applied Language Studies 2009, 27(3): 253–263