<![CDATA[Purpose – This study examines how environmental stress is associated with different psychosocial pathway configurations affecting household well-being in climate-sensitive rural systems, addressing limited attention to how stress is transmitted through resource, livelihood, health, economic, and psychological processes in agriculture-dependent households. The comparison is explicitly bounded to two context-specific sets of within-model associations and does not treat hazard regime as a statistically isolated cause of regional differences. Design/methodology/approach – The study uses comparative survey data from 1,080 households across chronic-drought areas in East Nusa Tenggara and multi-hazard areas in Sumatera. Separate Partial Least Squares Structural Equation Modeling (PLS-SEM) models were estimated for NTT (n = 540) and Sumatera (n = 540), and theory-specified direct paths were evaluated with 5,000 bootstrap subsamples. Because the regional models contain non-equivalent constructs, indicators, paths, and Subjective Well-Being measures, the cross-context comparison is descriptive and conceptual; no formal multi-group coefficient test, measurement-equivalence claim, or cross-regional mediation test is made. Findings/Results – In drought-prone NTT, significant direct paths form a sequential modeled chain linking perceived drought chronicity, water insecurity, food security, mental health stress, and the NTT well-being profile; the descriptive arithmetic product for the livelihood-resilience/economic-vulnerability chain is smaller within the NTT model. In Sumatera, significant direct paths form pollution-health and supply-chain-economic modeled chains whose descriptive arithmetic products are close in magnitude within that model. These values are descriptive multi-step products, not bootstrapped specific indirect effects, and they do not establish mediation, causal ordering, or a hazard-regime effect. Originality/Value – The study contributes empirically grounded, context-sensitive archetypes that clarify conditions under which chronic attrition or cascading disruption may become more prominent in agriculture-dependent systems. It also demonstrates a transparent way to compare non-equivalent model architectures descriptively without treating their coefficients or well-being endpoints as directly interchangeable. The archetypes are presented as mechanism-oriented propositions for future longitudinal, measurement-invariant, and objectively validated research.]]>
