PSYCHOMETRIC NETWORK ANALYSIS OF LOGICAL-MATHEMATICAL THINKING IN STUDENTS WITH DYSLEXIA

Abstract

This article proposes and demonstrates an analytical framework for deconstructing the impact of the Singapore model on the logical-mathematical thinking architecture of students with dyslexia, comparing it with that of peers without dyslexia. Using network psychometrics, cognitive, affective, and symbolic components) are modeled. Multigroup networks with mixed data and regularization (EBICglasso) are estimated, and structural contrasts are performed using the Network Comparison Test (NCT). EGA/bootEGA and network loadings are used to assess modularity and dimensionality; expected influence centrality/bridge centrality prioritizes "lever nodes." Additionally, SEM with sequential mediation (bootstrap) is proposed to examine plausible latent paths. As a technical demonstration, a simulated baseline (random group/time assignment) was used, which showed no pre-intervention differences between groups and pre-post in dyslexia, confirming the specificity of the NCT under null conditions. In an empirical study, we expect to observe reinforcement of visuospatial→performance edges, a decrease in the centrality of math anxiety, and consolidation of a "visual math reasoning" module, consistent with the literature.The central contribution is methodological and applied: a replicable pipeline that moves from averages to structural mechanisms, allowing prioritizing instructional targets and guiding pedagogical decisions based on the propagation of effects in the network.