A dual pathway for understanding the relation between wellbeing and resilience

The study addresses how individuals in Indonesian urban slums adapt to adversity, particularly water-related risks such as floods and poor access to clean water. Indonesia’s rapid urbanization has concentrated low-income populations in flood-prone, infrastructure-poor settlements (Bima, Manado, Pontianak). The research focuses on how relational wellbeing relates to individual resilience, theorizing that social relationships and context shape wellbeing and, in turn, resilience. The authors adopt White’s relational wellbeing framework (material, relational/social, subjective/human dimensions) and Benard’s perspective on individual resilience and its protective factors (family and community). The central research question is: To what extent can the relation between relational wellbeing dimensions and individual resilience be explained by family and community protective factors among slum dwellers in Indonesia? The study hypothesizes that relational wellbeing dimensions positively relate to individual resilience (H1), and that family (H2) and community (H3) protective factors mediate these relations.

The theoretical framework synthesizes: (1) Individual resilience as adaptive functioning in adversity, moving from trait views to multisystem perspectives where resilience manifests through dynamic systems (Masten, Southwick). Benard’s view highlights social competence, problem-solving, autonomy, and purpose/future as core attributes, shaped by environmental protective factors. (2) Protective factors: Families and communities provide caring relationships, high expectations, and opportunities to participate/contribute, which buffer risks and foster adaptation. Secure attachment, family closeness, and support enhance resilience; communities offering warmth, expectations, and participatory opportunities can also promote resilience. (3) Relational wellbeing (White) conceptualizes wellbeing as interdependent material, relational, and subjective dimensions, with relationships constitutive of subjectivity. Operational properties include income satisfaction (material), safety and access to services (relational), and self-concept/religiosity (subjective). (4) Community resilience is multidimensional (physical, social, economic, institutional, natural) and is expected to relate to individual resilience. (5) Hypotheses: Relational wellbeing dimensions—family/community participation, security/competition, material satisfaction, social/facility satisfaction, religiosity, self-concept—are posited to positively relate to individual resilience (H1). Family (H2) and community (H3) protective factors are hypothesized to mediate these relationships.

Design and data: Secondary analysis of the RISE (Resilient Indonesian Slums Envisioned) project dataset on water management, wellbeing, and resilience. Cross-sectional household survey conducted November 2021–February 2022 in slum settlements of three medium-sized Indonesian cities facing flood risks: Bima (fluvial/coastal), Manado (pluvial), Pontianak (pluvial/fluvial/coastal). Setting and site selection: Purposive selection of cities based on World Bank flood risk typologies. Within cities, districts and sub-districts designated as slums by mayoral decrees were chosen, emphasizing higher exposure to water-related problems. Sampling and participants: Due to incomplete population registries, a random-walk sampling method was used: starting near local government offices, dice-determined household intervals, nearest-birthdate selection within households, inclusion criteria age ≥18 and ≥5 years residency. Of 920 approached, 700 completed interviews (response rate 76.42%): Pontianak n=300, Bima n=200, Manado n=200. Table 2 provides descriptive statistics across measures. Mean age ranged approx. 41–46 years; gender distribution varied by city. Measures:

• Individual resilience: Based on Benard’s conceptualization, initially 16 items across four attributes (social competence, problem-solving, autonomy, sense of purpose/future). CFA supported a two-factor structure after removing two items: (1) social competence and problem-solving (9 items), (2) autonomy and sense of purpose/future (5 items). Fit indices: CFI=0.92, RMSEA=0.09, SRMR=0.05; factor loadings 0.49–0.65; AVE=0.31 (factor 1), 0.67 (factor 2); CR=0.80 and 0.91; α=0.91 for both. Composite score computed as mean across 14 items.
• Relational wellbeing: Developed from White’s framework; initial 58 items reduced via factor analysis to six dimensions: (1) family & community participation (7 items), (2) security & competition (4), (3) subjective material wellbeing (4), (4) public facility/social satisfaction (5), (5) religiosity (3), (6) self-concept (4). CFA: CFI=0.91, RMSEA=0.08, SRMR=0.07; loadings 0.46–1.17; reliability α=0.80–0.95.
• Protective factors (Benard): Family and community protective factors capturing caring/expectations and opportunities/participation. CFA favored a four-factor model merging high expectations with caring at both family and community levels. Fit: CFI=0.90, RMSEA=0.09, SRMR=0.04; reliability α=0.87–0.93. Composite scores computed for family and community protective factors (higher indicates stronger protection).
• Community resilience (CDRI-guided): Five dimensions—physical (facility access, utilities), social (education, poverty), economic (income, savings, assets), institutional (interactions with local figures), natural (hazard exposure/expectations). Dimension scores computed as means; higher indicates greater resilience.
• Individual characteristics: Age, gender, education, income. Analytic strategy: Confirmatory factor analyses (R lavaan) for validation (ML estimation; CFI>0.90, SRMR<0.08 considered acceptable). Bivariate correlations reported. Primary hypothesis testing via parallel (multiple) mediation using lavaan to estimate direct (c′) and indirect effects through family and community protective factors, controlling for age, gender, and community resilience dimensions. Effect sizes for indirect effects via MBESS; interpretation via Cohen’s benchmarks.

• City differences: Individual resilience differed significantly across cities, F(2,697)=304.10, p<0.05, with Pontianak highest and Manado lowest. Significant between-city differences also observed across most measures (Table 2).
• H1 (direct relations): Partially supported. Only family & community participation (b=0.18, p<0.001) and public facility/social satisfaction (b=0.07, p=0.01) were positively associated with individual resilience in models including mediators. Other relational wellbeing dimensions showed weaker or non-significant direct paths.
• H2 (family mediator): Predominantly supported. Family protective factor positively mediated most relations between relational wellbeing dimensions and individual resilience (small effect sizes r≈0.01–0.06). Religiosity’s indirect effect via family was near zero and not significant (b≈-0.01, p=0.06).
• H3 (community mediator): Not supported. Community protective factor showed negative (inverted) mediation for all relations between relational wellbeing dimensions and individual resilience, except a positive indirect effect for religiosity (b=0.04, p<0.001). The community protective factor was negatively associated with individual resilience overall (b=-0.20, p<0.001), with small-to-medium inverse effect size.
• Moderation-related note: A negative interaction was observed between religiosity and community protective factor on individual resilience (b=-0.11, p<0.001), indicating the positive relation between religiosity and resilience weakens as the community protective factor increases.
• Controls: Gender was not significantly associated with resilience. Age had a statistically significant but negligible coefficient (b≈0.00, p=0.003). Among community resilience dimensions, economic (b=0.16, p<0.001) and natural (b=0.38, p<0.001) dimensions were positively associated with individual resilience.
• Model performance: The mediation model explained R^2≈0.40 of variance in individual resilience; R^2 for mediators ≈0.36 (Table 4). Psychometric validation across measures showed acceptable to good fit and reliability.

Findings indicate dual and contrasting pathways from relational wellbeing to individual resilience through family and community protective factors. Positive assessments of relational wellbeing (participation, perceived security/competition, material satisfaction, satisfaction with public facilities, self-concept) are associated with stronger family protective factors, which in turn bolster individual resilience. This aligns with theories that caring relationships, high expectations, and opportunities within families meet psychological needs (belonging, competence, autonomy) and support adaptive functioning. Conversely, community protective factors exhibited an unexpected negative mediation to resilience, except in the religiosity pathway. The authors posit several explanations: (1) Protective functions may be domain-specific—family support may better foster problem-solving and emotional regulation than generalized community support, while involuntary or instrumental participation may not benefit mental health. (2) Potential confounding by collective efficacy; without fostering a sense of collective agency, community support might not translate into proactive resilience and could engender a false sense of security. (3) Non-empowering or dependency-oriented help may inadvertently undermine competence and agency, counteracting resilience. Religiosity appears to positively relate to resilience but may also encourage reliance on spiritual coping (e.g., surrendering to God), potentially reducing engagement with community structures for collective action in this context. Overall, the results suggest that strengthening family-based protective processes is a more consistent pathway to enhancing individual resilience, while community supports must be designed to be empowering and efficacy-enhancing to avoid negative or suppressor-like effects.

The study advances understanding of how relational wellbeing relates to individual resilience among slum dwellers through two protective pathways. It shows that family protective factors largely and positively mediate these relations, whereas community protective factors often exert inverted mediation, except in the religiosity pathway. These insights nuance resilience-building strategies by emphasizing family-oriented interventions and rethinking community support to avoid dependency and to build efficacy and agency. Practically, policies should foster caring, high-expectation, participatory family environments and community programs that empower residents with skills, knowledge, and collective efficacy for hazard mitigation and daily challenges. Future research should test additional mediators (e.g., collective efficacy), examine different types and voluntariness of community participation, and assess dynamics across normal and hazard periods.

• Urban-only sample from three Indonesian cities; generalizability to rural contexts is limited.
• Community activities were not specified, precluding comparisons between types of participation and their links to resilience.
• Many significant coefficients were small (approx. -0.02 to 0.39), indicating modest practical significance; other unmeasured mediators may be stronger.
• Cross-sectional design limits causal inference and cannot capture shifts between normal times and flooding periods; longitudinal or event-based data during flood episodes are needed.
• Potential confounding variables (e.g., collective efficacy) not directly measured may influence the observed negative community mediation effects.