Abstract
Structural Equation Modeling (SEM) is widely used to assess causal relationships among latent variables, yet its strict assumptions often limit empirical applications. Partial Least Squares SEM (PLS-SEM) offers greater flexibility, but the choice of weighting scheme remains a methodological challenge. This study introduces a PCA-based weighting scheme to improve the stability and accuracy of PLS estimation. Importance-Performance Analysis (IPA) is further integrated to identify high-impact but underperforming indicators. Applied to child malnutrition in East Java, the approach reveals that socio-economic conditions most strongly influence food security, parenting, and health-environment services. IPA highlights exclusive breastfeeding as a priority for intervention. The proposed methodological approach strengthens PLS estimation and yields actionable insights for prioritizing policy measures.