Mental health and decisions under risk among refugees and the public in Lebanon

Lebanon hosts a large refugee population (approximately 1.5 million Syrians and 180,000 Palestinians), creating complex pressures on public services and on individuals’ decision-making under risk. Poor mental health has been linked to suboptimal decisions, and conflict-related trauma and displacement can affect mental health and risk preferences. Risk preferences may shift toward aversion following violence exposure or toward risk-seeking after large material losses. These shifts bear on health, employment, finance, and migration behaviors. The study aims to: (1) validate a 10-item multidimensional psychological well-being (MPWB) measure in Arabic for Lebanese nationals and refugees; (2) assess differences in well-being between Lebanese, Syrian, and Palestinian groups; (3) examine the relationship between subjective well-being and risk-taking; (4) explore associations with trust; and (5) test whether behavioral interventions—nudges (disclosure) and boosts—improve advantageous financial choices under risk. Hypotheses included similar psychometric properties for the MPWB as in other contexts, greater risk-taking in loss frames, positive effects of both interventions (with boosts especially effective), and that interventions would mitigate baseline risk preferences.

Prior work links poor mental health to impaired decision-making (e.g., depression/anxiety effects) and shows that trauma, displacement, and ongoing stressors worsen mental health and can contribute to mental disorders, with intergenerational effects. Risk preferences can change after conflict or catastrophic losses—violence exposure may increase risk aversion, while large losses can increase risk-seeking to recoup losses, consistent with loss aversion. WHO guidance emphasizes context-specific mental health policy design. Behavioral insights, notably nudges (altering choice environments without restricting options) and boosts (building competencies for autonomous decision-making), have shown promise in improving decisions; prior studies in Serbia and the US found boosts and disclosure of expected values can aid financial decisions. Lebanon presents relevant contextual parallels (demographics, economy) to prior settings, but refugees are heterogeneous, making competence-enhancing interventions potentially attractive. Overall, the literature motivates examining mental health, risk preferences, trust, and the impacts of nudges/boosts in this high-stress, low-trust setting.

Design: Cross-sectional experimental survey administered in Arabic via Qualtrics from October 11 to November 22, 2019. Total N=754 recruited. Inclusion: age ≥18, residence in Lebanon, nationality Lebanese, Syrian, or Palestinian. For analyses, Palestinians from Syria (n=12) and one other nationality were excluded, yielding N=741 (Lebanese n=510; Syrian n=119; Palestinian from Lebanon n=112). Recruitment: Lebanese participants online; refugees approached in person due to access constraints. Blinding: Participants were unaware of intervention testing; data collectors were blinded to allocation. Ethical approvals: Centre for Business Research, Judge Business School, University of Cambridge; permission from Lebanese Ministry of Public Health. Measures and procedure: Participants completed 61 items: 10-item MPWB well-being scale; five behavioral indicators linked to well-being; three baseline financial risk items; then random assignment (Qualtrics) to control, disclosure nudge, or boost intervention; followed by 16 experimental financial decision items presented in random order; demographics last (18 items). Financial decisions were two-option gambles varying by frame (gains vs. losses), magnitude, risk probability, and whether the risky or safe option had higher expected value (EV). Risky choice defined as lower-probability/higher-payout option; safe choice as higher/certain probability with lower payout. Advantageous choice defined as option with higher EV. Interventions: Control saw choices without EV information. Disclosure nudge displayed the EV next to each option along with a brief explanation that EV equals outcomes weighted by their probabilities. Boost intervention provided instructional content on calculating EV, required correct computation of EV on three practice items (calculator allowed) before proceeding. Outcomes: Baseline risky choice (gain, loss, neutral frames); post-intervention number of advantageous choices in gain and loss frames; MPWB composite score (sum 10–70); trust measures (general trust binary; trust across domains—family, known people, neighbors, strangers). Power: A priori f-test indicated N=552 needed for 90% power (alpha 0.05) to detect R^2=0.025 in regressions of well-being on demographics. Analysis: Confirmatory factor analysis (CFA) of MPWB separately by group; linear regressions for group differences in well-being and associations with trust and demographics; logistic regressions predicting baseline risky choices from MPWB and nationality (planned contrasts: Lebanese vs. refugees; Palestinian vs. Syrian); generalized linear/linear models for intervention effects on advantageous choice, including interactions with nationality and baseline choices. Items were randomized; analyses separated gain and loss frames where appropriate.

- Sample: Of 754 recruited, 741 analyzed (Lebanese 510; Syrian 119; Palestinian 112). Lebanese participants had higher education, income, and expenditures than refugees. - MPWB validity: CFA indicated good fit for a single-factor structure in all groups (CFI > 0.9). Factor scores correlated 0.99 with simple sum scores, supporting use of summed composite (10–70). Distributions were approximately normal with no floor/ceiling effects. - Group differences in well-being: No significant differences in average MPWB between Lebanese and refugees (absolute difference = -1.27 on a 60-point scale; 95% CI [-2.83, 0.29]; t=-1.66; p=0.10), nor between Syrian and Palestinian refugees (absolute difference = -1.53; 95% CI [-4.16, 1.08]; t=-1.18; p=0.24). - Baseline risk-taking: No significant nationality differences overall. Risk-taking higher in loss-framed baseline items across groups. - Well-being and risk-taking: Higher MPWB associated with higher risk-taking at baseline across frames—gain (β=0.04, SE=0.01, z=3.70, p<0.01), loss (β=0.03, SE=0.01, z=2.73, p<0.01), neutral (β=0.05, SE=0.01, z=4.44, p<0.001). This relationship was significantly weaker for Lebanese than for refugees (gain: β=-0.02, SE=0.01, z=-2.49, p<0.05; loss: β=-0.02, SE=0.01, z=-2.82, p<0.01; neutral: β=-0.03, SE=0.01, z=-4.25, p<0.001). Among refugees, the gain-frame association was stronger for Palestinians than Syrians (β=0.04, SE=0.01, z=-2.54, p<0.05). Grouped by mental health levels, flourishing refugees showed the highest risky choice proportions (e.g., gain: 48.5% combined; Palestinian 63.0%; Syrian 65.2%). - Trust: General trust was low, especially among Palestinians (most people can be trusted: Lebanese 12.4%, Syrian 13.4%, Palestinian 6.2%). Trust in family higher than in strangers across groups; refugees reported higher family trust than Lebanese. MPWB was not significantly related to trust (p=0.48). Trust modestly predicted risk-taking only for Lebanese in the gain frame (β=0.05, SE=0.02, z=2.18, p<0.05). - Intervention effects on advantageous choice: • Lebanese: Boost increased number of advantageous choices—gain frame (β=0.86, 95% CI [0.50, 1.22], t=4.75, p<0.01); loss frame (β=0.75, 95% CI [0.33, 1.17], t=3.62, p<0.01). Baseline advantageous choice interacted with boost in loss frame (β=0.61, 95% CI [0.01, 1.21], t=2.00, p<0.05). Disclosure nudge not significantly better than control. • Syrian: No significant effects of boost or disclosure in gain or loss frames; baseline choices did not predict or moderate effects. • Palestinian: Gain frame null; in loss frame, disclosure nudge led to fewer advantageous choices than control (β=-0.88, 95% CI [-1.44, -0.32], t=-3.17, p<0.01). Baseline advantageous choice attenuated this negative effect (β=0.72, 95% CI [-0.08, 1.52], t=1.76, p=0.08, not significant). • Between-nationality comparisons: Boosts were significantly more effective for Lebanese than for Syrians (gain: β=-1.13, 95% CI [-1.87, -0.39], t=-3.08, p<0.01; loss: β=-0.83, 95% CI [-1.65, 0.01], t=-2.00, p<0.05) and Palestinians (gain: β=-0.89, 95% CI [-1.67, -0.11], t=-2.26, p<0.05; loss: β=-0.94, 95% CI [-1.76, -0.12], t=-2.27, p<0.05). Disclosure in loss frame was worse for Palestinians relative to Lebanese (β=-1.08, 95% CI [-1.88, -0.28], t=-2.70, p<0.01). - Overall: Clear pattern of greater risk-taking among refugees with better subjective well-being; boosts improved advantageous choice for Lebanese only; nudges showed limited or negative effects for refugees.

The study demonstrates that the Arabic MPWB functions as a unidimensional measure across Lebanese, Syrian, and Palestinian groups and that average well-being levels are similar between host and refugee populations. Crucially, higher subjective well-being among refugees is associated with greater risk-taking across decision frames, whereas this relationship is attenuated or absent among Lebanese nationals. This suggests that flourishing refugees may be more willing to engage in higher-risk options, potentially reflecting openness to opportunities or attempts to improve circumstances, but also carrying potential for adverse outcomes in vulnerable contexts. Trust levels are low overall and are not linked to well-being; trust has minimal predictive value for risk-taking. Intervention findings indicate that competence-building boosts can significantly improve advantageous choices for the Lebanese group, consistent with effects observed in some prior contexts, but such benefits did not extend to refugee groups. Disclosure-type nudges did not improve and, in some cases (Palestinians in loss frames), worsened decision quality. These results imply that interventions effective in one context may not generalize across populations differing in education, numeracy, stress, or testing conditions, and that refugees may require more robust or tailored decision-support strategies. Policy relevance includes recognizing that not only poor mental health but also better-reported well-being among refugees can be linked to riskier choices, which may affect health-seeking, financial decisions, and vulnerability to exploitation. Health systems should consider these dynamics when designing integrated services and behavioral supports for both host and refugee populations.

This study contributes by validating a brief well-being measure in Arabic across Lebanese and refugee populations and by identifying a novel association: higher subjective well-being among refugees correlates with greater risk-taking. While mean well-being appears similar between host and refugee groups, behavioral interventions had heterogeneous effects—boosts improved advantageous financial choices for Lebanese participants but not for refugees; disclosure nudges were ineffective or detrimental for refugees under loss framing. These insights can guide policymakers toward context-sensitive, competence-building supports and highlight the need to anticipate risk behaviors across the well-being spectrum in vulnerable populations. Future research should: (1) replicate findings in larger, well-matched samples; (2) test ecologically valid risk domains relevant to health behaviors and service utilization; (3) assess numeracy/comprehension to tailor boosts; (4) explore mechanisms (e.g., stress, education, legal status) underlying differential intervention efficacy; and (5) evaluate integrated service models that address low trust and support autonomous, advantageous decision-making for refugees.

- Measurement: MPWB domains are single-item indicators within a multifaceted construct; hypothetical monetary gambles may not capture real-world risk preferences, particularly for health-related or high-stress decisions. - Administration differences: Refugees were tested in-person while Lebanese participated online, potentially introducing social desirability or context effects, though primary outcomes may be less susceptible. - Confounding: Associations between refugee status, well-being, and risk-taking may reflect unmeasured socio-economic, legal, structural, or cultural factors. Education differed across groups, potentially affecting understanding of probabilities/expected value. - Intervention comprehension: The study did not directly verify comprehension of EV among all participants; lower education among refugees may limit boost effectiveness. - Power for complex interactions: Sample size limited exploration of higher-order interactions and nuanced subgroup analyses (e.g., Palestinians from Syria excluded due to small n). - Temporal context: Data were collected before the Beirut explosion (August 2020) and COVID-19’s subsequent impacts, which may alter generalizability to current conditions.