An examination of the psychometric properties of a game-based assessment of cognitive ability
Date
2024-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch University
Abstract
South Africa has had a controversial political history concerning psychometric assessments, and this has prompted legislation to establish clear guidelines for its application. Assessments must be reliable, valid, unbiased and fair. It is imperative to ensure that the assessment being utilised accurately measures the intended construct. At the same time, given the country's rich cultural diversity, preventing unfair discrimination is of utmost importance. The initial step in achieving this is through validating the assessment. Therefore, it is paramount that new assessments are validated in the South African context to avoid unfair discriminatory practices and adverse impacts in the selection processes. This would reduce detrimental effects on individuals’ lives. This study sought to validate a new game-based cognitive ability assessment called Cognify in the South African context. The examination was specifically undertaken to ascertain Cognify’s efficacy in predicting academic performance within a student population. Cognify is a game-based assessment that measures general cognitive ability with the intended purpose of predicting job performance. Recent research has suggested using Cognify for other purposes such as student performance. Thus, this validation process used a cross-section research approach involving n = 219 Stellenbosch University students. The psychometric properties, reliability and validity, more specifically Construct, Convergent and Predictive validity, were analysed, and Cognify was compared to the Raven’s Progressive Matrices. It was found that Cognify demonstrates adequate reliability, strong construct validity, and fit for a single factor. Furthermore, it reveals moderate convergent validity and acceptable predictive validity. The findings collectively support the use of Cognify as a reliable and valid cognitive assessment for use in Stellenbosch University students. Given Cognify’s success in other countries, it was not surprising that the psychometric properties of this measure are sound in the South African sample of university students and that it is a moderate predictor of academic performance. However, it is recommended, given the homogeneity of the researched university, that further research is required. This study addresses the need for validation of new assessments in the South African context and provides evidence for its use amongst university students. The implications, limitations and suggestions for future research are also discussed.
Description
Thesis (MCom)--Stellenbosch University, 2024.