Data-Driven Evaluation of Mathematics Teacher Development: Within-Teacher Change and Territorial Heterogeneity in Panama

Abstract

Large-scale professional-development programmes generate longitudinal educational data that can be used to examine both within-teacher change and territorial heterogeneity. This study presents a data-driven assessment of perceived professional learning across six dimensions of didactic suitability among 420 Panamanian in-service mathematics teachers enrolled in the Estrategias Didácticas para la Enseñanza de la Matemática (EDEM) programme in 2017, 2018, 2020, and 2021. A total of 409 teachers contributed linkable pre- and post-training self-assessments. A 45-item common core was harmonised across questionnaire versions, creating a reproducible longitudinal dataset analysed through reliability estimation, paired Wilcoxon tests, rank-biserial effect sizes, multiplicity-adjusted subgroup comparisons, and distribution-sensitive visual analytics. Self-ratings increased across all six dimensions. Affective (Δ = 0.74; r = .739) and interactional (Δ = 0.69; r = .765) changes were largest, whereas mediational change was small (Δ = 0.10; r = .134; Holm-adjusted p = .031). Affective, mediational, and interactional change varied across adequately represented provinces, while gender differences did not survive multiplicity adjustment. The results show how longitudinal, data-driven analysis can reveal patterns of professional learning and territorial heterogeneity that would be obscured by aggregate post-training scores alone. The findings are interpreted as changes in teacher self-perception rather than causal programme effects because the study lacks a comparison group and direct observation of classroom practice.