PEPS-Y Income: Resources and Political Solidarity

1 Setup

This report follows preregs/income_prereg and the theory draft in papers/Income wealth and inequality as drivers of political solidarity 20260218.docx. The current pilot file is rich enough for the preregistered individual-level analyses, interaction checks, secondary outcomes, and item-level ordinal robustness checks. The weighted specification is still out of scope because the pilot .sav does not contain design weights.

The cleaned pilot sample contains 355 interviews across 7 country samples. The main outcome is the preregistered territorial solidarity index, defined as the mean of the local, country, and EU willingness-to-pay items.

2 Preregistered Hypotheses

hypothesis status proxy
H1 Resources and solidarity Tested now Income rank within country, financial resilience, subjective household finances
H2 Ideology moderation Tested now Resource × right-ideology interactions
H3 Dependants / insurance moderation Tested now Resource × dependants interactions
H4 Economic insecurity and welfare chauvinism Tested now Resource terms in the welfare-chauvinism model
H5 Inequality as context Not testable from available files Would require a country-level inequality file such as a harmonised Gini
  • H1 Resources and solidarity. Respondents with higher economic resources should report higher political solidarity.
  • H2 Ideology moderation. The positive resource-solidarity association should be stronger among left-leaning respondents and weaker among right-leaning respondents.
  • H3 Dependants / insurance moderation. Resource effects should be stronger among respondents with dependants, consistent with an insurance logic.
  • H4 Economic insecurity and welfare chauvinism. Lower resources should predict stronger support for deserving groups relative to migrants.

3 Testing Strategy

The report keeps the preregistration’s structure but presents it in a compact pilot format. The main model predicts territorial solidarity from the three resource measures, right ideology, dependants, and the preregistered interaction terms. I then repeat the same specification for the secondary outcomes, apply an FDR correction across the preregistered outcome family for the main resource terms, and re-estimate the item-level models with ordered logit.

For interpretation, I use the same verdict language as in the core1 and core2 pilot reports. A hypothesis is confirmed in the pilot when the key estimate has the preregistered sign and is reasonably precise. It is mixed when the direction fits the expectation but the evidence is limited to one indicator or one secondary outcome. It is not confirmed when the estimates are near zero, inconsistent, or point in the wrong direction.

4 Outcome and Construct Checks

The theory text treats territorial and group solidarity as related but not identical. Here I check that in two simple ways. First, Cronbach’s alpha asks whether the items within each block move together closely enough to be treated as one scale. Second, the eigenvalue plot is a quick way to ask how many broader dimensions sit behind all eight items at once. If all items were basically measuring one thing, the first eigenvalue would dominate and the later ones would drop off sharply. In this pilot, the territorial items hang together reasonably well, and the group items do too, but the full eight-item battery does not behave like one perfectly unified measure. In plain language: people who are solidaristic in one domain also tend to be solidaristic in the other, but the two domains are not interchangeable, so it makes sense to keep reporting them separately.

scale alpha
Territorial solidarity 0.794
Group solidarity 0.849
All solidarity items 0.889

5 Visual Checks

These descriptive plots are not substitutes for the regression models, but they help show whether the main resource measures move with territorial solidarity in any obvious way. Financial resilience is the five-point item about how easily the household could deal with an unexpected financial shock, with higher values meaning more resilience. Income rank is not raw income but the respondent’s percentile rank within their own country sample, so higher values mean relatively higher income within country. Subjective finances is the four-point self-assessment of how comfortably the household is managing financially, again coded so higher values mean a more comfortable situation.

The Subjective finances panel is less smooth than the others because that variable only has a few response categories, so several observations pile up at the same x-values. That is a feature of the pilot data rather than a coding problem.

6 Main OLS Model

The dependent variable here is the territorial-solidarity index: the simple mean of the three willingness-to-pay items for the local level, the country level, and the EU level, after recoding them so higher values always mean more solidarity. In other words, each respondent gets one summary score for how willing they are to support redistribution across territorial targets.

All substantive predictors are standardized before entering the model. That means the coefficients can be read on a common scale: each one shows the expected change in the territorial-solidarity index associated with a one-standard-deviation increase in that predictor, holding the controls constant. This makes it easier to compare the relative size of income rank, financial resilience, subjective finances, ideology, and the interaction terms in the same table.

term_label estimate conf_int p_value
Subjective finances × dependants 0.395** [0.156, 0.634] 0.001
Subjective household finances 0.045 [-0.120, 0.209] 0.597
Income × ideology 0.037 [-0.061, 0.135] 0.460
Income rank within country 0.032 [-0.119, 0.182] 0.680
Subjective finances × ideology 0.028 [-0.102, 0.159] 0.672
Income × dependants 0.009 [-0.201, 0.219] 0.934
Resilience × dependants 0.007 [-0.233, 0.246] 0.957
Dependants in household -0.014 [-0.247, 0.220] 0.908
Financial resilience -0.071 [-0.248, 0.106] 0.430
Resilience × ideology -0.083 [-0.202, 0.035] 0.169
Right ideology -0.306*** [-0.418, -0.193] <0.001

The main model does not show a broad positive resource gradient in solidarity. None of the three resource measures is clearly associated with the territorial index once ideology, dependants, and the interaction terms are included. The clearest and most stable coefficient is not about resources at all: right ideology is strongly associated with lower solidarity. The only preregistered interaction that stands out is the positive subjective finances × dependants term, which suggests that feeling financially secure matters more for respondents with dependants than for others.

7 Interaction Plots

The first plot shows how the predicted territorial-solidarity score changes with income rank across the ideological spectrum. The second shows the strongest preregistered interaction in the pilot: subjective finances matter more when respondents have dependants.

8 Hypothesis Assessment

hypothesis hypothesis_text how_tested pilot_verdict evidence
H1 Respondents with higher economic resources report higher political solidarity. Main territorial-solidarity model plus the preregistered group-solidarity model using income rank, financial resilience, and subjective household finances. Not confirmed in the pilot Territorial solidarity: income beta = 0.032, 95% CI [-0.119, 0.182], p = 0.680; resilience beta = -0.071, 95% CI [-0.248, 0.106], p = 0.430; subjective finances beta = 0.045, 95% CI [-0.12, 0.209], p = 0.597. Group solidarity: income beta = -0.002, 95% CI [-0.141, 0.138], p = 0.983; resilience beta = -0.045, 95% CI [-0.199, 0.11], p = 0.574; subjective finances beta = 0.065, 95% CI [-0.067, 0.197], p = 0.335.
H2 The positive resource-solidarity link is stronger among left-leaning respondents and weaker among right-leaning respondents. Interactions between each resource indicator and right ideology in the territorial-solidarity model. Not confirmed in the pilot Income × ideology: beta = 0.037, 95% CI [-0.061, 0.135], p = 0.460; resilience × ideology: beta = -0.083, 95% CI [-0.202, 0.035], p = 0.169; subjective finances × ideology: beta = 0.028, 95% CI [-0.102, 0.159], p = 0.672.
H3 The resource-solidarity link is stronger among respondents with dependants. Resource × dependants interactions in the territorial-solidarity model, checked again on the group and deserving-target outcomes. Mixed pilot evidence Territorial model: income × dependants beta = 0.009, 95% CI [-0.201, 0.219], p = 0.934; resilience × dependants beta = 0.007, 95% CI [-0.233, 0.246], p = 0.957; subjective finances × dependants beta = 0.395, 95% CI [0.156, 0.634], p = 0.001. The strongest secondary signal appears for subjective finances among the group and deserving outcomes.
H4 Economic insecurity predicts higher welfare chauvinism. Welfare-chauvinism model using the three resource indicators as the preregistered insecurity proxies. Not confirmed in the pilot Income rank: beta = -0.052, 95% CI [-0.246, 0.142], p = 0.600; resilience: beta = 0.069, 95% CI [-0.127, 0.265], p = 0.491; subjective finances: beta = -0.087, 95% CI [-0.248, 0.074], p = 0.290.
H5 Higher country-level inequality lowers solidarity and raises welfare chauvinism. Contextual inequality model listed in the preregistration. Not testable in the pilot file No country-level inequality file is available in the repo, so the pilot cannot estimate this contextual step.

Substantively, the pilot suggests that economic resources are not doing the bulk of the work in this dataset. The dominant and most consistent pattern is ideological: right-leaning respondents report lower solidarity across the outcome family, especially toward migrants. The preregistered resource interactions also do not support a broad self-interest story. Instead, the only recurring moderation signal comes from subjective household finances among respondents with dependants, which fits the paper’s insurance logic more than a simple income-gradient argument. At the same time, the welfare-chauvinism hypothesis is not clearly supported in the pilot.

9 Secondary Outcomes and FDR Correction

This section keeps the preregistered distinction between the main outcome and the broader family of related outcomes. The reason for doing this is simple: once we test the same predictors across several outcomes, some “significant” results can appear by chance even when there is no stable pattern. The Benjamini-Hochberg false-discovery-rate correction is a way to be more cautious without being as severe as a Bonferroni correction. It asks: after taking the whole family of tests into account, which signals still look noteworthy?

The tables below therefore do two things at once. They show whether the resource terms behave similarly across the main territorial outcome, the group-based outcomes, and welfare chauvinism; and they show how much confidence we should have in those signals once multiple testing is acknowledged.

9.1 Territorial Solidarity

outcome_label term_label estimate conf.low conf.high p.value p.value.fdr
Territorial solidarity index Financial resilience -0.071 -0.248 0.106 0.430 0.717
Territorial solidarity index Income rank within country 0.032 -0.119 0.182 0.680 0.977
Territorial solidarity index Subjective household finances 0.045 -0.120 0.209 0.597 0.607

9.2 Group-Solidarity Outcomes

outcome_label term_label estimate conf.low conf.high p.value p.value.fdr
Deserving-target solidarity Financial resilience -0.017 -0.158 0.124 0.814 0.814
Deserving-target solidarity Income rank within country -0.022 -0.158 0.113 0.747 0.977
Deserving-target solidarity Subjective household finances 0.030 -0.085 0.145 0.607 0.607
Migrant-target solidarity Financial resilience -0.086 -0.316 0.144 0.465 0.717
Migrant-target solidarity Income rank within country 0.030 -0.179 0.239 0.782 0.977
Migrant-target solidarity Subjective household finances 0.117 -0.081 0.315 0.247 0.559
Overall group-solidarity index Financial resilience -0.045 -0.199 0.110 0.574 0.717
Overall group-solidarity index Income rank within country -0.002 -0.141 0.138 0.983 0.983
Overall group-solidarity index Subjective household finances 0.065 -0.067 0.197 0.335 0.559

9.3 Welfare Chauvinism

outcome_label term_label estimate conf.low conf.high p.value p.value.fdr
Welfare chauvinism Financial resilience 0.069 -0.127 0.265 0.491 0.717
Welfare chauvinism Income rank within country -0.052 -0.246 0.142 0.600 0.977
Welfare chauvinism Subjective household finances -0.087 -0.248 0.074 0.290 0.559

10 Item-Level Ordered Logit Robustness

The item-level checks ask whether the main patterns are being driven by only one territorial or group target. They also use an estimator that matches the ordinal response scales more closely than OLS.

10.1 Territorial Targets

outcome_label term_label estimate conf.low conf.high p.value
Local community Subjective household finances 0.688 0.357 1.030 0.000
Local community Income rank within country 0.101 -0.279 0.480 0.603
Local community Dependants in household -0.134 -0.799 0.531 0.693
Local community Financial resilience -0.292 -0.677 0.092 0.136
Local community Right ideology -0.671 -0.978 -0.375 0.000
Other EU countries Subjective household finances 0.250 -0.052 0.554 0.105
Other EU countries Financial resilience 0.094 -0.258 0.445 0.600
Other EU countries Income rank within country 0.007 -0.357 0.373 0.969
Other EU countries Dependants in household -0.085 -0.718 0.551 0.791
Other EU countries Right ideology -0.821 -1.107 -0.543 0.000
Own country Subjective household finances 0.665 0.329 1.010 0.000
Own country Dependants in household 0.056 -0.612 0.727 0.869
Own country Income rank within country 0.001 -0.382 0.382 0.995
Own country Financial resilience -0.216 -0.597 0.164 0.264
Own country Right ideology -0.563 -0.866 -0.268 0.000

10.2 Deserving Targets

outcome_label term_label estimate conf.low conf.high p.value
Children Subjective household finances 0.273 -0.048 0.599 0.097
Children Dependants in household 0.272 -0.398 0.947 0.426
Children Income rank within country -0.029 -0.421 0.363 0.885
Children Financial resilience -0.088 -0.463 0.287 0.644
Children Right ideology -0.197 -0.488 0.089 0.181
Older people Income rank within country NA NA NA NA
Older people Financial resilience NA NA NA NA
Older people Subjective household finances NA NA NA NA
Older people Right ideology NA NA NA NA
Older people Dependants in household NA NA NA NA
Poor people Subjective household finances 0.579 0.256 0.911 0.001
Poor people Financial resilience 0.007 -0.368 0.382 0.969
Poor people Dependants in household -0.071 -0.747 0.603 0.835
Poor people Income rank within country -0.274 -0.679 0.126 0.181
Poor people Right ideology -0.672 -0.988 -0.368 0.000

10.3 Migrant Targets

outcome_label term_label estimate conf.low conf.high p.value
EU migrants in country Subjective household finances 0.387 0.080 0.697 0.014
EU migrants in country Income rank within country 0.183 -0.182 0.551 0.328
EU migrants in country Dependants in household -0.191 -0.818 0.435 0.549
EU migrants in country Financial resilience -0.224 -0.577 0.128 0.213
EU migrants in country Right ideology -0.985 -1.279 -0.701 0.000
Non-EU migrants in country Subjective household finances 0.406 0.101 0.717 0.010
Non-EU migrants in country Income rank within country 0.080 -0.290 0.456 0.672
Non-EU migrants in country Financial resilience -0.104 -0.467 0.259 0.575
Non-EU migrants in country Dependants in household -0.123 -0.761 0.516 0.705
Non-EU migrants in country Right ideology -1.301 -1.625 -0.995 0.000

11 Notes

  • The pilot file does not contain design weights, so all results here are unweighted.
  • income_rank_country is built as a within-country percentile rank to keep the resource measure comparable across country samples.
  • The current report is best read as a prereg-consistent pilot analysis rather than the final confirmatory paper model set.

12 Appendix

12.1 Missingness Diagnostic

variable n_total n_missing pct_missing
dependants 355 69 19.4
education_level 355 52 14.6
income_rank_country 355 35 9.9
fresil 355 29 8.2
ecohhld 355 9 2.5
ideology_right 355 7 2.0
employment_status 355 2 0.6
gender_binary 355 1 0.3
solidarity_territorial 355 0 0.0
age 355 0 0.0
model n pct_of_total
Main territorial model 210 59.2
Group-solidarity model 210 59.2
Welfare-chauvinism model 210 59.2