Subjective socioeconomic status and income inequality are associated with self-reported morality across 67 countries

The study investigates whether and how subjective experiences of economic scarcity relate to human morality. Subjective scarcity, operationalized as low subjective socioeconomic status (SES), may differ from objective scarcity and has been linked to shifts in cognition and decision-making. Prior literature offers conflicting predictions: some studies suggest scarcity fosters greed, dishonesty, and reduced prosociality, while others argue it promotes empathy, cooperation, and prosocial behavior. Existing research has limitations including reliance on single-country samples (often the United States), underpowered experiments, and narrow measurement of morality. This paper aims to provide a comprehensive, cross-national, well-powered assessment using multiple morality-related measures and to test whether national income inequality moderates individual-level associations.

Two competing paradigms characterize prior findings. One predicts negative effects of scarcity on morality, linking lower SES or deprivation to greed, dishonesty, lower prosocial intentions, and reduced charitable giving. Another suggests that lower perceived social class fosters contextual orientation, empathy, compassion, and prosociality, with some evidence of greater generosity and trust among lower-class individuals. Replications are mixed and effects may be context dependent, potentially moderated by macro-level inequality. The literature also suffers from limited cross-cultural generalizability, statistical conclusion validity due to small underpowered studies, and restricted measurement validity from focusing on single morality indicators. These gaps motivate a large-scale, cross-national test with multiple morality measures and examination of national inequality as a moderator.

Design and data: Pre-registered analysis using the International Collaboration on Social & Moral Psychology of COVID-19 (ICSMP) dataset. Final analytic sample: N = 50,396 adults from 67 countries (28 nationally representative for age and gender). Data collected online during April–May 2020 with forward-backward translated surveys; ethics approval from the University of Kent IRB; informed consent obtained. Exclusions included incomplete responses, out-of-age-range participants, failed attention checks, and those indicating gender as Other (to avoid unstable estimates for rare categories in multilevel models). Mean age 43; 52% female.

Key measures:

• Subjective SES: MacArthur ladder (0–10), indexing perceived social standing in one’s country.
• National inequality: GINI index (0–100) from the World Bank (supplemented for a few countries from other sources).
• Morality outcomes:
  1. Moral Identity (10 items, 0–10; aggregated internalization and symbolization; Cronbach’s α = 0.729; multiple-group factor alignment indicated acceptable invariance).
  2. Morality-as-Cooperation (7 items tapping cooperative moral concerns; 0–10; α = 0.732; acceptable invariance).
  3. Moral Circle (single item, 16 levels from immediate family to all things in existence).
  4. Prosocial Intentions (sum of intended donations to national and international charities from a hypothetical daily median income).
• Covariates: Age and gender (female vs male).

Analytic strategy:

• Multilevel linear mixed-effects models (country random intercept), standardized coefficients, Wald 95% CIs, REML estimation, two-tailed α = 0.05. Pairwise deletion of missingness; robustness via Random Forest imputation (Supplementary Table S4).
• Decomposition of subjective SES effects into within-country and between-country components (grand-mean, within-country, and between-country centering).
• Nested country-level OLS regressions to estimate country-specific SES–morality associations.
• Robustness checks using adjusted disposable net-income (objective national indicator) as an alternative macro-level covariate.
• Cross-validation: 10-fold CV with 200 repetitions to assess predictive robustness (lowest RMSE models retained).

Main multilevel results (controlling for age and gender):

• Individual-level subjective SES (lower SES = higher morality):

  • Moral Identity: β = -0.14, t(45832) = -5.54, p < 0.001, 95% CI [-0.15, -0.13].
  • Morality-as-Cooperation: β = -0.07, t(45946) = -0.36, p < 0.001, 95% CI [-0.08, -0.06].
  • Prosocial Intentions: β = -0.08, t(45649) = -2.57, p = 0.010, 95% CI [-0.09, -0.07].
  • Moral Circle: β = 0.01, t(46646) = 1.09, p = 0.003, 95% CI [0.00, 0.02] (negligible magnitude; heterogeneity across countries). Results were similar or slightly stronger in nationally representative subsamples.

• Country-level inequality (GINI):

  • Moral Identity: β = 0.13, t(45832) = 2.98, p = 0.003, 95% CI [0.04, 0.21].
  • Moral Circle: β = 0.08, t(46646) = 2.94, p = 0.003, 95% CI [0.03, 0.13].
  • Morality-as-Cooperation: β = 0.03, t(45946) = 1.53, p = 0.479 (ns).
  • Prosocial Intentions: β = 0.06, t(45649) = 1.78, p = 0.118 (ns).

• Moderation: SES × GINI interactions were generally not meaningful; income inequality did not significantly moderate SES–morality associations in the full sample.

• Variance components: Most variance at the individual level (87.7% Moral Identity; 90.4% Morality-as-Cooperation; 95.5% Moral Circle; 89.3% Prosocial Intentions). ICCs ranged ~0.04–0.12.

• SES–GINI association: Higher GINI correlated with lower subjective SES (t(47315) = -14.79, p < 0.001, r = -0.07). Lower quantile of SES predicted higher inequality (t(46306) = 11.02, p < 0.001, β = 0.13).

• Within vs between decomposition of SES:

  • Within-country SES negatively predicted Moral Identity (β = -0.13, p < 0.001), Morality-as-Cooperation (β = -0.07, p < 0.001), and Prosocial Intentions (β = -0.08, p < 0.001). For Moral Circle, within-country SES showed a very small positive association (β = 0.01, p = 0.002), whereas between-country SES was negatively associated with Moral Circle (β = -0.07, p = 0.004).

• Contextual differences: Country-level SES–prosocial intention associations varied (e.g., India β = -0.26, p < 0.001; South Africa β = -0.15, p = 0.006; Sweden β = -0.08, p = 0.003). Regionally, associations were larger in the Americas (r = -0.12) than in Europe (r = -0.08).

• Robustness: Results held with imputation, cross-validation (10-fold, 200 reps), and when using adjusted disposable net-income (which was negatively associated with Moral Identity, Morality-as-Cooperation, and Prosocial Intentions; not with Moral Circle).

Contrary to hypotheses predicting moral depletion under scarcity, individuals reporting lower subjective SES showed higher moral identity, stronger endorsement of cooperation-related moral principles, and greater prosocial donation intentions. At the macro level, higher national income inequality was associated with stronger moral identity and larger moral circles. These findings align with theories that lower social class heightens contextual orientation and interdependence, encouraging prosociality and cooperation as adaptive strategies to improve future outcomes. The moral circle findings are nuanced: a tiny within-country positive link between lower SES and circle size, but a between-country pattern where higher average SES corresponds to smaller circles, suggesting contextual sensitivity. Although effect sizes are small, they may be practically meaningful at scale and over time. The results highlight that certain morality dimensions may become more salient under perceived scarcity, without implying that individuals with higher SES are less moral overall; rather, moral expression may depend on context and the specific moral domain measured.

This large cross-national study shows that subjective SES and national income inequality relate to multiple dimensions of self-reported morality. Lower subjective SES is associated with higher moral identity, greater morality-as-cooperation, and stronger prosocial intentions; higher national inequality relates to stronger moral identity and larger moral circles. The work improves cross-cultural generalizability, statistical power, and measurement breadth in this domain. Future research should integrate objective and subjective scarcity measures, employ behavioral and field methods to test causality, refine macro-inequality metrics beyond GINI, examine moderators (e.g., culture, personality), and assess these relationships beyond pandemic contexts.

• Self-reported measures may not fully capture real behavior, though prior work shows moderate correspondence for donations and dishonesty.
• Cross-sectional observational design limits causal inference.
• Small effect sizes and modest model explanatory power; potential common-method variance inflating individual-level variance.
• Lack of individual-level objective income data (household income not available in ICSMP); reliance on national adjusted net-income for robustness.
• GINI is a crude, one-parameter inequality measure and an indirect proxy for perceived scarcity; cannot localize inequality within the distribution.
• Heterogeneity across countries and regions; moral circle results especially context sensitive.
• Data collected during the COVID-19 pandemic, which may strengthen associations relative to non-crisis periods.
• Exclusion of participants identifying gender as Other may limit generalizability for gender-diverse populations.