Enhancing test quality in Lesotho Basic Education through Classical Test and Item Response Theory Analyses

Abstract

 

This study developed and psychometrically evaluated a Grade 6 mathematics test for Lesotho’s Cambridge curriculum using Classical Test Theory (CTT) and Item Response Theory (IRT). A quantitative psychometric design involving 200 learners from Cambridge International Schools was employed. Data were analysed using jMetrik through CTT indices and 1PL and 2PL IRT models. Results revealed substantial psychometric weaknesses, including extreme item difficulty variation, weak and negative discrimination indices, low reliability coefficients (α = 0.1975), and several misfitting items. The 2PL model demonstrated a superior fit over the 1PL model based on deviance, AIC, and BIC statistics. While CTT identified global weaknesses in reliability and item discrimination, IRT provided richer diagnostic evidence regarding item functioning and conditional measurement precision across ability levels. The study concludes that integrating CTT and IRT strengthens assessment validity, reliability, fairness, and evidence-based test development in Lesotho basic education. The study recommends rigorous item piloting, routine use of CTT and IRT analyses, and enhanced psychometric training for teachers and assessment practitioners in Lesotho.