Addressing gender discrimination in cognitive assessment using the English Comprehension Test

Abstract

Orientation: The empirically designed English Comprehension Test (ECT) is theorised to measure verbal reasoning and is currently undergoing validation. The test development produced two versions of the ECT, namely, ECT version 1.2 and ECT version 1.3. This study focuses on the latest test version, ECT version 1.3. Research purpose: The purpose of this study was to statistically explore the performance of men and women who were assessed by the empirically designed ECT. Motivation for the study: Cognitive assessment has often been used as a discriminatory tool against gender, race and/or languages. The discrimination against race and gender were the consequences of a patriarchal system and Apartheid in South Africa, as black men and women were deemed to be subordinate to white men. With the demise of Apartheid, measures have been put in place to guard against unfair assessment practices. In addition, legislation was developed to ensure that test developers and test users employed assessments that did not unfairly prejudice individuals based on their race, gender and language. These measures are imperative to ensure fairness and equal opportunities for men and women across race and language groups. Research design, approach and method: This study used a quantitative cross-sectional design. The ECT was administered to a non-probability convenience sample of 881 individuals. The data were analysed by differential test functioning (DTF) in Winsteps and analysis of variance (ANOVA) in the Statistical Product and Service Solutions (SPSS) package. Main findings: The results indicated that the majority of the test items did not present any bias, but five possibly biased items were identified across gender groups in the test. These five items that were possibly biased appear to be affected by language and not gendered knowledge, and this, however, necessitates further investigation. The ANOVA results only indicated statistically significant differences across the different language groups, thereby confirming the DTF results. Practical/managerial implications: A major limitation of this study is the restriction of range and lack of generalisability. Contribution/value-add: This study promotes the use of DTF and ANOVA as a means of ensuring fairness in assessment practices across gender groups. Moreover, it contributes to cross-cultural test development and validation research in South Africa.