Psychometric network analysis of questionnaire data in applied linguistics

Questionnaire research in applied linguistics commonly uses reliability analysis, factor analysis, structural equation modeling, regression, and person-centered methods to investigate multidimensional constructs. These procedures represent items as indicators of latent factors, components of composite scores, or variables for classifying respondents. Psychometric network analysis provides a complementary representation in which observed items are modeled as interconnected components and edges express conditional associations after the remaining items have been controlled. This methodological article explains the conceptual and analytical foundations of psychometric network analysis and demonstrates an item-level workflow using PERMA-based well-being responses from 206 Iranian EFL teachers. A 23-node EBIC-GLASSO network was examined in terms of edge structure, strength, expected influence, bridge strength, empirical communities, edge accuracy, and case-dropping stability. The analysis identified strong within-domain connections, a cross-domain association between perceived life value and excitement or interest, highly connected positive and negative components, and communities that only partly corresponded to the intended PERMA domains. The article shows how network analysis can complement established psychometric procedures by locating conditional item relations and cross-domain connections while retaining explicit checks on redundancy, estimation accuracy, and centrality stability. Practical guidance is provided for conceptualization, estimation, robustness assessment, and interpretation in applied linguistics questionnaire research.