Worldwide divergence of values

The study investigates whether social values across nations have been converging or diverging in recent decades. While modernization theories predicted convergence toward liberal, individualizing values characteristic of Western democracies, alternative perspectives proposed persistence or divergence of cultural values, potentially along regional, religious, or civilizational lines. The authors emphasize the importance of this question for understanding international conflict, economic climate, law, and the generalizability of psychological research. They formulate competing hypotheses: convergence globally due to globalization; conditional convergence tied to economic development (post-materialist thesis); or global divergence with possible regional convergence or clustering (multiple modernities and civilizations theses). The paper aims to provide a comprehensive, item- and country-level assessment of value change using World Values Survey data from 1981–2022.

The paper reviews several theoretical traditions and prior empirical findings. Modernization theorists (inspired by Marx and Hegel) posited a trajectory toward a universal civilization with liberal, individualist values, driven by industrialization and globalization. Inglehart’s post-materialist thesis and Welzel’s human empowerment sequence argue that economic development shifts priorities from group obedience to self-expression, anticipating convergence toward emancipative values with rising prosperity. In contrast, Eisenstadt’s multiple modernities suggests diverse, culture-specific modernization pathways; Tomlinson argues globalization can strengthen distinct national identities; Huntington predicts renewed cultural divides along religious and linguistic civilizational lines. Evolutionary and ecological models link cultural values to socioecological pressures (subsistence, resource scarcity, climate, pathogens), allowing for both slow and rapid (evoked) value changes. Media-based accounts consider Western mass media as a potential force for convergence, yet evidence is mixed: some states regulate Western content, audiences often prefer local media, and media interventions typically require culture-specific production. Empirically, the World Values Survey (WVS) has been central for studying value change. Prior studies report global increases in individualism, diffusion of emancipative values faster in liberal democracies, and regional patterns (e.g., EU-related convergence with divergence from Central Asia). However, most past work focused on mean shifts in select values or limited country/wave coverage, leaving open the overarching question of global convergence versus divergence across many countries and values.

Data: WVS time-series (1981–2021/22) covering 105 countries total; analyses restricted to 76 countries with at least two waves; n = 406,185 individuals across waves. Forty WVS items present in all seven waves and interpretable as social values were selected. Sampling: country-level samples are probability/stratified samples of adults 18+ in private residences; weights provided by WVS. Data cleaning: For the child qualities battery, respondents selecting >5 qualities were excluded (n = 20,380; 5%); results robust to including them. Measures and normalization: Items with heterogeneous scales were min–max normalized to 0–1 across the full dataset. Primary outcomes: (1) Value variation (item level): standard deviation of country means at each timepoint per item; increasing SD over time indicates divergence. (2) Value distinctiveness (country level): for each country and item at each wave, compute absolute deviation from the global median; aggregate across items to yield a country’s distinctiveness score—higher values mean greater dissimilarity from the global center. (3) Within-country heterogeneity: analogous computation at the individual level within countries (absolute deviation of individual responses from country means), capturing internal value dispersion. Analytic strategy: Mixed-effects models tested temporal trends in value variation and distinctiveness, nesting observations within items, countries, and continents; additional models controlled for spatial autocorrelation (addressing Galton’s problem). Cohort vs. longitudinal effects were separated by centering to distinguish WVS sample composition changes from within-country changes over time. Robustness checks included subsamples with reduced turnover (countries present in ≥3, ≥4, ≥5 waves) and a fixed-country panel (33 countries with data in the 1990s, 2000s, 2010s). Exogenous country-level predictors over time included GDP per capita (World Bank), Gini (WID), KOF Globalization index, Freedom House Political Rights (recoded so higher = more rights), and distance from equator; religious distance for pairwise analyses used population shares of major religions. Clustering/value space: A PCA on the 40-item correlation matrix identified two principal components (PC1 = 65.4%, PC2 = 15.3%; total 80.7%). Countries were projected into a 2D value space using item loadings and country scores. PC1 was strongly associated with emancipative and secular values. Pairwise regressions (cross-classified mixed effects) related Euclidean distance in value space to pairwise differences in GDP per capita, geographic distance, religion, Gini, and political rights across waves. All predictors were standardized for comparability. Supplemental analyses considered alternative normalizations and weighting; results were robust.

Item-level divergence: Mixed-effects model showed timepoint significantly predicts higher value variation across items (b = 0.004, SE = 0.0007, t(239) = 5.12, p < 0.001, β = 0.17, 95% CI [0.002, 0.005]). Item-wise correlations revealed 27/40 values diverged over time; median r(timepoint, value variation) = 0.28, t(39) = 3.30, p = 0.002, 95% CI [0.11, 0.45]. Divergence was strongest for items loading on emancipative vs. obedient values (r with divergence = 0.54, p < 0.001), not for sacred–secular (r = 0.19, p = 0.237). The seven most divergent items: justifiability of homosexuality, euthanasia, importance of child obedience (reverse-emancipative), divorce, prostitution, suicide, and abortion. From the first to last wave, value variation rose by 141% (homosexuality), 94% (prostitution), 61% (euthanasia), and 42% (child obedience). Regional trends show increasing endorsement of emancipative values in Oceania, Europe, North and South America, with relative stability in Asia and Africa. Country-level distinctiveness: Value distinctiveness increased over time (b = 0.003, SE = 0.0005, t(9290) = 4.98, p < 0.001, β = 0.05, 95% CI [0.002, 0.004]); replicated controlling for spatial autocorrelation (b = 0.003, SE = 0.0005, t(9351) = 5.18, p < 0.001, β = 0.05). Longitudinal within-country change was significant (b = 0.003, SE = 0.0006, t(10,520) = 4.92, p < 0.001), cohort effect non-significant (b = 0.002, SE = 0.003, p = 0.503). Effects replicated in subsamples: ≥3 waves (b = 0.003, SE = 0.0006, t(8599) = 4.82, p < 0.001), ≥4 waves (b = 0.003, SE = 0.0006, t(6145) = 4.12, p < 0.001), ≥5 waves (b = 0.003, SE = 0.0007, t(4059) = 3.73, p < 0.001). In 33-country fixed panel, distinctiveness was higher in 2000s (b = 0.01, SE = 0.003, t(78) = 4.05, p < 0.001, β = 0.29) and 2010s (b = 0.01, SE = 0.003, t(78) = 2.66, p = 0.010, β = 0.19) vs. 1990s; no 2000s–2010s difference (b = −0.004, p = 0.167), suggesting slowing divergence. Predictors of distinctiveness: In multivariate models, GDP per capita positively predicted value distinctiveness (b = 0.08, SE = 0.01, t(600.59) = 8.05, p < 0.001, β = 0.18). Gini, Globalization, Political Rights, and Distance from Equator were not significant. The GDP–distinctiveness link was strong in Europe (b = 0.08, SE = 0.02, p = 0.002) but weaker/non-significant in Asia and Africa. Within-country heterogeneity: Higher within-country heterogeneity was associated with lower value distinctiveness (b = −0.06, SE = 0.01, t(8032) = −4.45, p < 0.001, β = −0.07), controlling for GDP per capita, Gini, globalization, political rights, and latitude. Temporal trends in within-country heterogeneity were mixed across nations. Value similarity and clustering: PCA-derived value space showed countries cluster by wealth and, increasingly, by geography and religion. Across waves, similarity in GDP per capita had the strongest association with value similarity (standardized β up to 0.59 at Time 7). Geographic proximity’s association rose over time (β from 0.16 at Time 1 to 0.35 at Time 7), indicating regional convergence. Religious similarity was a robust predictor (e.g., β = 0.35 at Time 2; 0.11 at Time 7), whereas Gini and political rights were weaker correlates by the final wave. High-income Western countries have become particularly distinctive due to higher endorsement of emancipative values.

The findings directly address the core question by demonstrating that, from 1981 to 2022, national social values have diverged globally, with the sharpest divergence on emancipative versus obedient values. This divergence is driven largely by increasing differences between high-income Western countries and the rest of the world. At the same time, values have converged regionally, as indicated by rising importance of geographic proximity for value similarity. The results partially align with multiple theoretical frameworks: they contradict simple unilineal modernization and the expectation of widespread global convergence; they support aspects of the post-materialist thesis (wealth linked to emancipative values) primarily in Western contexts; and they are consistent with multiple modernities and Huntington’s civilizations thesis through emerging regional and religious clustering. The negative association between within-country heterogeneity and cross-national distinctiveness suggests that domestic consensus accompanies greater differentiation from global norms. These patterns have implications for understanding geopolitical tensions, policy divergence, and the external validity of research that relies heavily on WEIRD samples, which appear increasingly atypical in value profiles.

This paper contributes a general framework for measuring cross-national convergence versus divergence of values using two complementary metrics (value variation and value distinctiveness) and a comprehensive WVS-based analysis spanning 76 countries and 40 consistently measured items. The results show robust global divergence of values, concentrated on emancipative issues, coupled with regional convergence and clustering by wealth, geography, and religion. High-income Western nations have become especially distinctive. Future research should: (1) replicate these analyses using other cross-national datasets and broader value items; (2) investigate causal mechanisms linking wealth, regional context, and political institutions to emancipative value change; (3) examine the roles of migration, media ecosystems, and political elites in shaping divergence; (4) test non-linear dynamics over longer time horizons; and (5) assess policy implications for international cooperation amid value divergence.

Key limitations include: (1) the analysis covers 40 items consistently present across WVS waves, which is not an exhaustive set of human values—results may differ with other measures; (2) despite WVS efforts, samples may imperfectly represent national populations and may underrepresent certain groups (e.g., indigenous populations); (3) the global median benchmark is contingent on countries present in each wave, introducing dependence on WVS composition; (4) observational design limits causal inference about drivers of divergence, including the role of wealth and regional factors; and (5) some continent-specific analyses rely on smaller subsamples, warranting cautious interpretation.