Reconceptualising Learning Approach Adoption through Assessment Perceptions: Evidence from Ghanaian Colleges of Education

Abstract

Background. Fewer research examines how assessment perceptions shape learning approaches in teacher education, particularly the perceptual alignment between lecturers and pre-service teachers in developing contexts. This study tested the Assessment Perception Scale (APS) dimensions as predictors of deep learning approaches across Ghana's Colleges of Education.
Aims. Three aims: (1) examine lecturer-trainee perceptual divergence across five APS dimensions; (2) test APS predictive power for deep learning approaches; (3) explore group interactions.
Sample. Nationally representative N = 3,874 (354 lecturers, 3,520 pre-service teachers) across 24 public Ghanaian Colleges of Education; 92.3% response rate.
Methods. Multistage sampling followed Krejcie & Morgan (1970). One-way MANOVA tested group differences; hierarchical multiple regression (demographics, APS, and interactions) predicted deep learning. APS α = .92-.96.
Results. Table 2 reveals significant perceptual divergence across all APS dimensions, Pillai's Trace = .988, F(5, 3,514) = 57,428.64, and p < .001; task authenticity showed the largest effect (ηp² = .145). Table 3 demonstrated that APS explained 62% of the variance (R² = .62) vs. demographics 8% (R² = .08); transparency was strongest (β = .42). Group against Task Authenticity interaction was significant (ΔR² = .02, p = .032).
Conclusions. APS dimensions powerfully predict deep learning, substantially outperforming demographics. Trainees' task authenticity perceptions drive deep approaches more than lecturers. National coverage across 46 institutions mandated CoE reform prioritizing transparency/authenticity training, informing GTEC policy on Colleges of Education in Ghana.