Neighbourhoods in the Netherlands with greater exposure to immigrants were associated with lower right-wing vote shares in the study's analysis of 2023 voting. The estimated exposure slope was -0.300 across 2,816 neighbourhoods, with p < 0.001. The result describes a relationship between places, not evidence that exposure caused residents to vote differently.
It is not a simple finding that more exposure always coincides with less opposition. The relationship was more negative in dense areas, while greater exposure in low-density areas was associated with more right-wing voting. In high-exposure neighbourhoods, poorer places recorded larger increases in right-wing vote share.
The study asks how exposure to immigrants relates to right-wing voting, using that vote as a proxy for anti-immigrant sentiment. The proxy combines votes for PVV, FvD and JA21. It is an electoral measure built from party support, not a direct reading of each resident's attitude.
The map behind the result
To build its exposure measure, the study used the 2023 version 2 500m-by-500m grid and aggregated it to CBS municipalities and neighbourhoods. The measure uses Euclidean distance between grid cells and defines immigrants as people born outside the Netherlands with a heritage outside Europe. LISS Politics and Values data were used to identify the right-wing anti-immigrant parties.
Main models covered 341 municipalities and 2,816 neighbourhoods. The authors used ordinary least squares regressions with controls for wealth, age and urbanisation. These are comparisons across aggregated areas, not an experiment.
The authors report that the negative association survived the listed robustness checks. Even so, it remains an aggregate association rather than a causal estimate.
Density changed the picture
The analysis tested whether density changed the exposure relationship. Its estimate for that density interaction was -0.400, with p < 0.01. Dense regions showed a stronger negative relationship between exposure and right-wing voting, while higher exposure in low-density regions was associated with more right-wing voting.
The curve bends at higher exposure
Another analysis examined the change in right-wing vote share from 2021 to 2023. Most neighbourhood differences were positive, and the strongest increases were in relatively low-exposure places. Comparable exposure data were not available for 2021, so this comparison does not estimate the effect of a measured change in exposure.
A quadratic model, which allows a straight-line relationship to bend, produced evidence of that curvature. In the COROP-fixed-effects model, the linear exposure coefficient was -0.243, with a standard error of 0.027, while the quadratic coefficient was positive at 0.273, with a standard error of 0.065. Both terms had p < 0.01 in 2,444 neighbourhoods. The pattern is consistent with a possible late reversal, but the study does not report the exact exposure level at which the turn occurs.
The sharpest shifts were local
To look at the upper end of exposure, the researchers used thresholds of 0.2, 0.25 and 0.3. Across all three restricted analyses, WOZ home value was the only variable significant at p < 0.01. Poorer neighbourhoods had larger increases in right-wing vote share.
Those restricted samples were small and omitted province and COROP fixed effects. The result should therefore be read as an association within selected high-exposure neighbourhoods, not as evidence that lower wealth caused the larger shifts.
Two neighbourhoods had increases above 20 percentage points: Brouwhuis, where the right-wing share rose by 22.6 points, and IJmuiden-Zuid, where it rose by 20.9 points. Each had a 2023 right-wing share of 44%.
The authors present acute concentration and visible reminders as a possible mechanism for these outliers. The examples are descriptive and do not establish that this mechanism caused the voting increases.
A result about places, not people
The central limit is the level of analysis. The data describe populations in municipalities and neighbourhoods, not individual life histories. Lifetime exposure is not modelled, and population aggregation can produce an ecological fallacy: a pattern across places may hide opposing responses among individuals. The study therefore cannot show how individual residents changed their attitudes.
Right-wing voting is also only a proxy for anti-immigrant sentiment, so it may reflect factors beyond immigration. The authors interpret the findings as a possible mix of contact-like negative associations, a threat or relative-deprivation pattern at high exposure and low wealth, and effects of acute visible reminders. They present these as proposed mechanisms rather than demonstrated causes.
The manuscript is an arXiv preprint, version 1, dated 25 August 2026. Its acknowledgements report support for Benedikt Meylahn and Tommaso Giommoni from the Dutch Institute for Emergent Phenomena cluster at the University of Amsterdam under the research priority area Emergent Phenomena in Society: Polarisation, Segregation, and Inequality.
Paper data and sources
Original title: The complex relationship between anti-immigrant sentiment and exposure in the Netherlands
Authors: Benedikt Meylahn, Tommaso Giommoni, Mike Lees, Alessandro Nai
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-25
DOI: Not available
Original paper · Full text