Test construction competency components and pupils' mathematics outcomes in Sierra Leone

This study examined how specific components of mathematics teachers' test construction competency predict pupils' learning outcomes in senior secondary schools in Western Area Rural District, Sierra Leone. The article is derived from Objective 4 of the author's doctoral thesis on test construction competency and pupils' mathematics achievement. A convergent mixed-methods design was used, with the quantitative strand reported here through component-level multilevel modelling. The sample comprised 40 senior secondary schools, 80 mathematics teachers and 640 pupils. Teachers completed a Test Construction Competency Questionnaire covering planning/blueprinting, item writing, item review/moderation and post-test item analysis/evidence use. Teacher-made mathematics assessment artefacts were audited, and pupils completed a curriculum-aligned mathematics achievement test. Because pupils were nested within teachers and schools, three-level random-intercept models were estimated. Adding the four competency components significantly improved model fit over the controls-only model (Delta -2LL = 16.00, df = 4, p = .003). Planning/blueprinting significantly predicted pupil achievement (B = 4.10, SE = 1.40, 95% CI [1.36, 6.84], p = .004), as did item writing (B = 2.50, SE = 1.10, 95% CI [0.34, 4.66], p = .024). Item review/moderation and post-test item analysis/evidence use were positive but not statistically significant. The findings suggest that front-end assessment practices, especially planning and item writing, offer practical leverage points for strengthening mathematics assessment quality and improving pupils' achievement in similar resource-constrained settings.