Algorithm reasoning as a learning pathway: a PLS-SEM model linking computational understanding to conceptual explanation in South African mathematics
Résumé
Introduction This study contributes to applied structural equation modeling (SEM) research by examining the performance and interpretive boundaries of partial least squares–structural equation modeling (PLS-SEM) under analytically constrained yet empirically common conditions. Methods Using cross-sectional survey data from a small educational sample ( N = 71), the study estimates a hybrid SEM combining observed single-indicator variables and reflective latent constructs to model relationships among learner characteristics, cognitive processes, affective states, and multidimensional learning outcomes. Structural relationships were evaluated using bootstrapped standardised path coefficients, and hypothesised mediation effects were tested via bootstrapped indirect-effect estimation. Results Results indicate that educational level significantly predicts a latent computational understanding construct, which in turn strongly predicts conceptual understanding, while several demographic and affective predictors do not exhibit statistically significant effects. No indirect effects were supported. Measurement assessment revealed limited internal consistency for selected constructs and inadequate discriminant validity (HTMT >.90), alongside weak global fit indices (SRMR = 0.162; NFI = 0.307), constraining confirmatory inference. Discussion Rather than treating these outcomes as analytic failure, the study interprets them as informative boundary conditions for PLS-SEM inference under hybrid measurement and limited sample size. This study is positioned as an exploratory, theory-informed analysis of the relationship between computational and conceptual understanding under constrained data conditions ( N = 71). The PLS-SEM model is used to examine whether the proposed relationships are consistent with the theoretical rationale, while interpreting findings cautiously given sample size and measurement limitations.
Citer ce document
Accès au document
Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter
Voir l'article sur le site de la revueAuteur(s)
Statistiques
Consultations : 2
Téléchargements : 0