Uncovering Latent Dimensions of Environmental Stress in Urban Residents: A PCA Simulation Study with Sensitivity Analysis and the Dynamic Environmental Resilience Signature Framework

Abstract

Environmental stressors — extreme heat, air pollution, water and power scarcity, and flood-related disruption — increasingly co-occur with psychological distress in ways that conventional single-variable analysis fails to capture. This simulation study applies principal component analysis (PCA) to a synthetic eight-variable environmental-stress dataset generated from a known three-factor latent structure — Environmental Pressure, Psychological Depletion, and Future Dread — and quantifies PCA’s predictive and structural advantages over raw-variable regression. The base scenario (n = 245) recovers the three-component structure with a Tucker’s congruence coefficient of φ = 0.991 against a large-sample reference solution, explains 85.3% of total variance, reduces mean variance inflation factor from 3.71 to 1.00, and improves the Akaike Information Criterion (AIC) by 12.0 units over raw-variable regression. A systematic sensitivity analysis across eight sample sizes (n = 40 to 400) and seven inter-variable correlation levels shows that component recovery crosses the conventional stability threshold (φ ≥ 0.95) at approximately n = 80–120 and stabilizes above n = 200, while PCA’s AIC advantage persists across all tested conditions (ΔAIC = 9.5 to 16.4). At the smallest sample size tested (n = 40), component recovery is markedly less reliable (mean φ = 0.942, stable solutions in 76% of replications), establishing a practical lower bound for applied deployment. Extending beyond the cross-sectional design, this paper proposes the Dynamic Environmental Resilience Signature: an individual’s characteristic sequence of dominant stress components across a seasonal cycle (pre-monsoon, monsoon/flood period, post-monsoon dry season), recoverable through longitudinal score projection without re-fitting the PCA model. Together, the simulation design, sensitivity boundaries, and longitudinal framework constitute a methodological guide for researchers intending to deploy PCA on environmental-stress survey data in Pakistani and comparable urban contexts.