Economic segregation is associated with reduced concerns about economic inequality

The paper addresses why public concern about economic inequality is often muted despite rising disparities. Prior work emphasizes ideology, lay beliefs, and misperceptions, as well as exposure to inequality. The authors propose a situational mechanism: economic segregation—the separation of people by economic means—reduces the salience of cross-class contrasts and thus dampens economic comparisons that typically heighten concern about inequality. Drawing an analogy to the Mach Bands illusion, the authors argue that juxtaposition (functional closeness) of wealth and poverty makes disparities more psychologically salient, whereas segregation obscures contrasts, impedes comparisons, and lowers concern. Crucially, they hypothesize that segregation reduces concern even when overall inequality and exposure to its magnitude are held constant, implying a psychological effect of segregation per se beyond distorted perceptions of inequality.

The study builds on literature showing that attitudes toward inequality are shaped by ideology, system justification, and beliefs (e.g., Waldfogel et al.; Goudarzi et al.), as well as by misperceptions of inequality (Hauser & Norton; Gimpelson & Treisman). Situational research links local exposure and social sampling to attitudes about redistribution (Dawtry et al.; Sands; Sands & de Kadt), while work on income and residential segregation documents class isolation across neighborhoods, schools, workplaces, and institutions (Reardon & Bischoff; Pew Research Center; Chetty et al.; Mijs & Roe). Social comparison theory and related work emphasize how salient reference groups and juxtaposition influence judgments of fairness and policy support (Festinger; Gerber et al.; Frank; Jackson & Payne; Baldwin & Mussweiler). The authors extend this literature by isolating the role of economic segregation itself—separate from inequality’s magnitude—in reducing concern via fewer economic comparisons.

The research combines an archival analysis (Study 1) with preregistered experiments (Studies 2a–2d, 3, 4, 5a–5b). IRB approval obtained; informed consent collected. Materials, data for Studies 2–5, and code for Study 1 are available on OSF. Study 1: Merged economic segregation measures (from Opportunity Insights and 2000 US Census) across 741 US Commuter Zones with World Values Survey (1995–2017) attitudes toward income equality. Segregation measured three ways (overall; high-income vs others; low-income vs others). Inequality measured three ways (overall Gini; bottom 99% Gini; top 1% income share). Controls included population, income per capita, racial composition, racial segregation, and 2020 Democratic vote share. All variables standardized. Estimated 918 specifications (3×3×2 with leave-one-state-out and full sample), clustering SEs within state and year. Outcome: single WVS item (reverse-scored) indicating concern for income equality. Studies 2a–2d: Hypothetical organizations manipulating segregation vs integration while holding overall wage inequality constant. - 2a (N=299 MTurk; analyzed N=290): between-participants; 80 employees across 4 offices; Economic Segregation (inequality between offices) vs Economic Integration (inequality within offices). Measured perceived within-office inequality (manipulation check), perceived overall inequality, fairness, concern, and issue salience/policy support. - 2b (N=100 Prolific; analyzed N=92) and 2c (lab; N=117; analyzed N=111): within-participant; same 20 employees grouped two ways (segregated vs integrated); counterbalanced order; measured within-office inequality, overall inequality, fairness, and concern. - 2d (N=602 MTurk; analyzed N=584): between-participants 2×4 design manipulating segregation (between vs within offices) and overall inequality level (Low, Medium, High, Very High; Gini 0.22/0.32/0.42/0.52). Manipulation checks and same outcome measures as 2a. Study 3 (N=200 Prolific; analyzed N=190): Amazon case. Participants saw 25 Amazon roles with salaries (from Glassdoor). Segregation vs integration of job locations while holding overall inequality constant. Measured within-quintet inequality, overall concern, fairness, and familiarity with roles (as potential moderator). Study 4 (N=503 MTurk; analyzed N=472): South Africa images (Johannesburg, Cape Town, Pietermaritzburg, Durban). Economic Integration condition juxtaposed high- and low-income neighborhoods side-by-side; Economic Segregation condition presented the same neighborhoods separated by income across screens. Measured perceived inequality per image (manipulation), and overall concern, fairness, and tolerance (Support for Economic Inequality Scale) for South Africa. Study 5a (N=303 MTurk; analyzed N=288) and 5b (N=301 MTurk; analyzed N=290): Country-level cityscapes with 12 incomes per city (4 cities; 48 individuals total). Manipulated segregation across cities (inequality between cities) vs integration within cities, holding overall inequality constant. Outcomes: perceived fairness, concern; mechanism measures: frequency of economic comparisons (thinking about own and others’ income; r=0.64); 5b adds attributions (external vs effort, 0–100). Mediation analyses tested whether reduced comparisons account for lower concern/fairness evaluations under segregation. Statistical analyses: independent- and paired-samples t-tests for planned contrasts; multiple regressions/ANOVAs for 2d; mediation with indirect effects and CIs; no multiple-comparison corrections given planned tests.

• Study 1 (archival, WVS + US segregation): Living in more economically segregated areas is associated with lower concern about inequality, controlling for inequality levels. Model 1 (controls inequality): β = −0.061, SE = 0.028, p = 0.029, 95% CI [−0.116, −0.006]. Model 2 (additional controls): β = −0.269, SE = 0.093, p = 0.004, 95% CI [−0.451, −0.087]. Specification curve (918 models): average β = −0.150; 100% negative; 98.1% statistically significant. - Study 2a (between-participants, hypothetical firm): Manipulation check confirmed lower within-office inequality perception under segregation (M=2.79) vs integration (M=5.09); t(269.89)=14.26, p<0.001, d=1.681. Despite identical overall inequality, segregation reduced perceived overall inequality (Mdiff=0.72, t=3.54, p<0.001, d=0.414), increased fairness (Mdiff=-0.88, t=-4.01, p<0.001, d=0.471), reduced concern (Mdiff=1.11, t=4.79, p<0.001, d=0.562), and lowered issue salience/policy support (Issue big: Mdiff=1.04, t=4.90, p<0.001, d=0.574; Need to decrease inequality: Mdiff=0.59, t=2.76, p<0.001, d=0.325). - Studies 2b (within, N=92) and 2c (within, N=111): Manipulation checks strongly supported segregation effect on within-office inequality (ts>16, ps<0.001, ds≈1.67–1.74). Perceived overall inequality did not differ (as expected), but segregation reduced concern (2b: Mdiff=-0.82, t=4.54, p<0.001, d=0.473; 2c: Mdiff=-0.97, t=5.00, p<0.001, d=0.474) and increased fairness (2b: Mdiff=0.51, t=2.87, p=0.005, d=0.299; 2c: Mdiff=0.64, t=3.30, p=0.001, d=0.312). - Study 2d (between, 2×4): Manipulation check: lower perceived within-office inequality in segregation (M=2.79) vs integration (M=4.65); t=17.33, p<0.001, d=1.455. Main effects of segregation on perceived overall inequality (Mdiff=0.91, t=6.47, p<0.001, d=0.522), fairness (Mdiff=-0.64, t=-4.28, p<0.001, d=0.354), concern (Mdiff=0.63, t=3.95, p<0.001, d=0.336), and issue salience/policy support (Issue big: Mdiff=0.85, t=5.46, p<0.001, d=0.448; Need to decrease: Mdiff=0.49, t=3.26, p<0.001, d=0.273). No significant interaction with overall inequality level, indicating effects across Low to Very High Gini conditions. - Study 3 (Amazon roles): Manipulation check significant (t(188)=12.34, p<0.001, d=1.791). Segregation reduced concern (Mdiff=1.18, t=5.16, p<0.001, d=0.748) and increased fairness (Mdiff=-0.92, t=-4.41, p<0.001, d=0.639). No moderation by role familiarity. - Study 4 (South Africa neighborhoods): Manipulation check significant (t(435.24)=17.78, p<0.001, d=1.626). Segregation reduced concern (Mdiff=0.96, t(470)=6.27, p<0.001, d=0.577), increased fairness (Mdiff=-0.55, t=-3.77, p<0.001, d=0.346), and increased tolerance of inequality (Mdiff=0.51, t=4.90, p<0.001, d=0.451), despite identical exposure to neighborhoods. - Study 5a (country cities; mechanism): Manipulation check significant (t(286)=14.54, p<0.001, d=1.714). Segregation reduced concern (Mdiff=1.12, t=5.03, p<0.001, d=0.591) and increased fairness (Mdiff=-0.67, t=3.14, p=0.002, d=0.370). Participants anticipated fewer economic comparisons under segregation (Mdiff=0.54, t=3.26, p=0.001, d=0.383). Mediation: comparisons partially mediated effects (Concern indirect β=0.49, SE=0.15, 95% CI [0.20, 0.79]; direct β=0.63, SE=0.17, 95% CI [0.30, 0.96]. Fairness indirect β=−0.28, SE=0.09, 95% CI [−0.46, −0.10]; direct β=−0.40, SE=0.20, 95% CI [−0.79, 0.00]). - Study 5b (replication; mechanism and attribution): Segregation reduced concern (Mdiff=0.55, t=2.49, p=0.013, d=0.292) and increased fairness (Mdiff=−0.54, t=−2.58, p=0.010, d=0.303). Fewer anticipated comparisons under segregation (Mdiff=0.70, t=4.40, p<0.001, d=0.516); mediation supported (indirect β=0.41, SE=0.11, 95% CI [0.20, 0.61]; direct β=0.15, SE=0.21, 95% CI [−0.26, 0.56]). Alternative account refuted: segregation did not increase internal attributions; it decreased them (Mdiff=−8.62, t=2.82, p=0.005, d=0.332). Overall: Across archival and experimental studies, economic segregation consistently reduced concern about inequality and increased perceived fairness/tolerance, even when actual and perceived overall inequality were held constant; reduced economic comparisons are a key mechanism.

Findings support the hypothesis that economic segregation dampens concerns about inequality by obscuring direct juxtapositions between wealth and poverty and reducing economic comparisons. The effect emerges across domains (organizations, cities, countries), samples (online, lab), and designs (between- and within-participants), and persists when overall inequality and exposure are held constant, indicating a psychological impact of segregation per se. The archival study shows real-world associations across US regions, while experiments demonstrate causality and identify reduced comparison frequency as a mechanism, independent of misperceptions of inequality. The results suggest two pathways by which segregation shapes attitudes: by distorting perceptions of inequality’s magnitude and by impeding social comparisons, each potentially reinforcing the other. The work implies that physical and functional distribution of socioeconomic groups is pivotal for public concern and policy support regarding inequality.

The paper demonstrates that economic segregation lowers concern about economic inequality by reducing the salience of cross-class contrasts and economic comparisons, even at identical levels of inequality. This effect is robust across a wide range of contexts and analytic specifications and is partially mediated by anticipated frequency of economic comparisons. The research advances understanding of how spatial and institutional arrangements shape inequality attitudes beyond ideological or informational factors. Future research should map the boundary conditions of comparison-based mechanisms, explore additional factors that reduce juxtaposition (e.g., segregated schools, workplaces, media consumption, transportation), test interactions with other forms of segregation (racial, ethnic), examine cross-national contexts, and analyze downstream behavioral and policy implications across the income distribution.

• Study 1 is correlational; although robustness checks and extensive specifications were used, causal inference is limited. - Potential self-selection: individuals less concerned about inequality may prefer more segregated areas, creating bidirectional relationships. - Generalizability: lab and online experiments may not capture complex real-world dynamics where segregation interacts with additional factors. - Context limitations: primary archival focus on US data due to cross-country data constraints; effects in other cultural contexts warrant testing. - Heterogeneity by income: no statistically significant moderation by participant income was detected, but effects on attitudes may not translate uniformly to behaviors or policy support across the socioeconomic spectrum.