PEPS-Y Core 1: Pilot Hypothesis Checks

1 Setup

hypothesis status proxy
H1 Ideology Tested now ideology_left_index
H2a Objective resources Tested now objective_resources_index
H2b Perceived capacity Tested now perceived_capacity_index
H3a Objective need/risk Tested now objective_need_index
H3b Perceived risk Tested now perceived_risk_index
H4 Norms and perceived contribution Tested now norms_legitimacy_index
H5a Generalized social trust Tested now social_trust_index
H5b Institutional trust Tested now institutional_trust_index
H5c Public-service evaluations Tested now service_quality_index
H6a Regional service quality Tested with EQI service-quality context EQI 2024 qualityp matched to survey regions
H6b Relative regional prosperity Tested with regional GDP context Eurostat 2023 NUTS2 GDP per capita, centered within country

The cleaned pilot sample contains 355 interviews across 7 countries. The main outcome is the preregistered territorial solidarity index, defined as the mean of the municipal, country, and EU willingness-to-pay items. The GISCO-based region match linked EQI and GDP information to 355 interviews.

2 Preregistered Hypotheses

  • H1 Ideology. “Individuals who are more economically left-leaning and egalitarian show higher political solidarity than individuals who are more economically right-leaning.”
  • H2a Objective resources. “Individuals with higher objective socioeconomic resources (income and wealth) show higher political solidarity.”
  • H2b Perceived capacity. “Net of objective resources, individuals who perceive higher personal capacity to contribute (financial security) show higher political solidarity.”
  • H3a Objective need/risk. “Individuals with higher exposure to social risks show higher political solidarity.”
  • H3b Perceived risk. “Net of objective indicators, individuals who perceive higher personal risk of deterioration and higher future need show higher political solidarity.”
  • H4 Norms and perceived contribution. “Individuals who believe that most others contribute fairly (pay taxes, do not abuse benefits, endorse solidarity norms) show higher political solidarity.”
  • H5a Generalized social trust. “Individuals with higher generalized social trust show higher political solidarity.”
  • H5b Institutional trust. “Individuals with higher institutional trust show higher political solidarity.”
  • H5c Public-service evaluations. “Individuals who are more satisfied with the quality and fairness of public services show higher political solidarity.”
  • H6a Regional service quality. “Residents of regions with higher public service quality show higher political solidarity.”
  • H6b Relative regional prosperity. “Holding absolute conditions constant, residents of relatively better-off regions within a country show lower political solidarity on average, consistent with a net-contributor logic in territorial redistribution.”

3 Testing Strategy

The pilot report tests the micro-level hypotheses in two stages and then combines them. Block A covers ideology, resources, perceived capacity, need, risk, and the preregistered risk-by-capacity interaction. Block B covers norms, generalized trust, institutional trust, and public-service evaluations. The full model combines both blocks with the same socio-demographic controls and country fixed effects. I use the territorial solidarity index as the main outcome, then check whether the same directional patterns appear in the ordered-logit item models and in the secondary solidarity outcomes.

For interpretation, I treat a hypothesis as confirmed in the pilot when the main full-model coefficient has the preregistered sign and its 95% confidence interval excludes zero. I label it mixed when the sign is in the expected direction but the estimate is imprecise, or when support appears only in some secondary outcomes. I label it not confirmed when the estimate is near zero or points the wrong way. The H6 context models are now estimated using the linked EQI and regional GDP files and should be read as contextual pilot checks rather than final multilevel evidence.

4 Visual Hypothesis Checks

The plots below use quintile-style bins for each preregistered micro-level predictor. They are descriptive checks rather than substitutes for the regression models.

5 Main OLS Models

In this section we implement the preregistered block-by-block regression strategy. The first model estimates the contribution of the individual-foundations variables only; the second does the same for the norms-and-legitimacy variables; and the full model asks which associations remain once both sets of predictors compete with each other in the same specification.

All substantive predictors are standardized, thus the coefficients can be read as the expected change in the territorial solidarity index associated with a one-standard-deviation increase in each predictor, holding the controls constant.

Block A and Block B are two separate OLS models fit on the same outcome but with different predictor sets. The full model is a third OLS model that combines both blocks.

model term_label estimate conf_int p_value
Block A H1 Left ideology 0.301*** [0.192, 0.409] <0.001
Block A H2b Perceived capacity 0.165* [0.038, 0.291] 0.011
Block A H3b Perceived risk 0.065 [-0.028, 0.158] 0.171
Block A H3a Objective need/risk 0.026 [-0.078, 0.129] 0.626
Block A H3 Risk × Capacity interaction -0.001 [-0.098, 0.095] 0.976
Block A H2a Objective resources -0.017 [-0.147, 0.113] 0.796
Block B H5b Institutional trust 0.214** [0.070, 0.359] 0.004
Block B H5a Social trust 0.180** [0.046, 0.314] 0.009
Block B H4 Norms and legitimacy 0.070 [-0.030, 0.170] 0.171
Block B H5c Service evaluations -0.013 [-0.145, 0.119] 0.845
Full H1 Left ideology 0.257*** [0.151, 0.362] <0.001
Full H5b Institutional trust 0.205** [0.077, 0.332] 0.002
Full H5a Social trust 0.192** [0.072, 0.311] 0.002
Full H2b Perceived capacity 0.142* [0.020, 0.263] 0.023
Full H3b Perceived risk 0.088 [-0.013, 0.189] 0.091
Full H4 Norms and legitimacy 0.054 [-0.047, 0.155] 0.293
Full H3a Objective need/risk 0.028 [-0.084, 0.139] 0.628
Full H3 Risk × Capacity interaction -0.037 [-0.132, 0.058] 0.442
Full H2a Objective resources -0.057 [-0.188, 0.073] 0.390
Full H5c Service evaluations -0.069 [-0.200, 0.062] 0.302

5.1 Regression Tables

Characteristic

Block A

Block B

Full model

Beta

1

SE

2

Beta

1

SE

2

Beta

1

SE

2
H1 Left ideology 0.301*** 0.048

0.257*** 0.047
H2a Objective resources -0.017 0.066

-0.057 0.063
H2b Perceived capacity 0.165** 0.058

0.142* 0.056
H3a Objective need/risk 0.026 0.055

0.028 0.053
H3b Perceived risk 0.065 0.050

0.088 0.048
H2b Perceived capacity * H3b Perceived risk -0.001 0.048

-0.037 0.047
H4 Norms and legitimacy

0.070 0.047 0.054 0.045
H5a Social trust

0.180** 0.056 0.192*** 0.055
H5b Institutional trust

0.214** 0.068 0.205** 0.065
H5c Service evaluations

-0.013 0.062 -0.069 0.060
1

p<0.05; p<0.01; p<0.001
2

SE = Standard Error

5.2 Model Fit Comparison

Both predictor blocks explain a meaningful share of variation in political solidarity in the pilot data, with Block A showing slightly stronger fit than Block B on its own. I also add a Context only model and a Full + Context model so the regional indicators can be evaluated both on their own and net of the individual-level blocks.

model n r_squared adj_r_squared
Block A 267 0.250 0.202
Block B 273 0.223 0.181
Full (A + B) 267 0.334 0.280
Context only 273 0.127 0.087
Full + Context 267 0.346 0.287

6 Hypothesis-by-Hypothesis Assessment

The table below matches hypotheses of preregs into explicit verdicts based on pilot data.

hypothesis hypothesis_text how_tested pilot_verdict evidence
H1 Individuals who are more economically left-leaning and egalitarian show higher political solidarity than individuals who are more economically right-leaning. Full territorial-solidarity model with standardized ideology index and full controls. Confirmed in the pilot beta = 0.257, 95% CI [0.151, 0.362], p = 0.000
H2a Individuals with higher objective socioeconomic resources (income and wealth) show higher political solidarity. Full model using the preregistered objective-resources index, net of perceived capacity and controls. Not confirmed in the pilot beta = -0.057, 95% CI [-0.188, 0.073], p = 0.390
H2b Net of objective resources, individuals who perceive higher personal capacity to contribute (financial security) show higher political solidarity. Full model using the perceived-capacity index, net of objective resources and the risk interaction. Confirmed in the pilot beta = 0.142, 95% CI [0.02, 0.263], p = 0.023
H3a Individuals with higher exposure to social risks show higher political solidarity. Full model using the objective-need/risk index, net of perceived risk and controls. Mixed pilot evidence beta = 0.028, 95% CI [-0.084, 0.139], p = 0.628
H3b Net of objective indicators, individuals who perceive higher personal risk of deterioration and higher future need show higher political solidarity. Full model using the perceived-risk index plus the preregistered perceived-risk × perceived-capacity interaction. Mixed pilot evidence Main effect: beta = 0.088, 95% CI [-0.013, 0.189], p = 0.091; interaction beta = -0.037, p = 0.442
H4 Individuals who believe that most others contribute fairly (pay taxes, do not abuse benefits, endorse solidarity norms) show higher political solidarity. Full model using the norms-and-legitimacy index; secondary outcomes checked whether the pattern was stronger for deserving or migrant-targeted solidarity. Mixed pilot evidence beta = 0.054, 95% CI [-0.047, 0.155], p = 0.293
H5a Individuals with higher generalized social trust show higher political solidarity. Full model using the social-trust index, with secondary outcomes used to check scope across deserving and migrant targets. Confirmed in the pilot beta = 0.192, 95% CI [0.072, 0.311], p = 0.002
H5b Individuals with higher institutional trust show higher political solidarity. Full model using the institutional-trust index, supplemented by secondary-outcome checks. Confirmed in the pilot beta = 0.205, 95% CI [0.077, 0.332], p = 0.002
H5c Individuals who are more satisfied with the quality and fairness of public services show higher political solidarity. Full model using the service-quality index, separated conceptually from institutional trust. Not confirmed in the pilot beta = -0.069, 95% CI [-0.2, 0.062], p = 0.302
H6a Residents of regions with higher public service quality show higher political solidarity. Full + context model using EQI 2024 `qualityp`, added on top of the individual-level predictor blocks with the same respondent controls and country fixed effects. Not confirmed in the pilot beta = -0.158, 95% CI [-0.35, 0.034], p = 0.107
H6b Holding absolute conditions constant, residents of relatively better-off regions within a country show lower political solidarity on average, consistent with a net-contributor logic in territorial redistribution. Full + context model using within-country centered NUTS2 GDP per capita from Eurostat 2023, added on top of the individual-level predictor blocks with the same respondent controls and country fixed effects. Not confirmed in the pilot beta = 0.076, 95% CI [-0.016, 0.168], p = 0.109

Substantively, the pilot already gives a fairly coherent picture at the individual level. The strongest and cleanest evidence points to ideology, generalized social trust, and institutional trust: all three are positive, reasonably precise, and stable in the full model. Perceived capacity also behaves in the preregistered direction. By contrast, the pilot offers little support for objective resources, objective need, or service-quality evaluations once the broader trust and governance measures are in the model. Perceived risk is directionally consistent with the preregistration, but the estimate remains imprecise and the preregistered interaction with perceived capacity is close to zero. The norms-and-legitimacy measure also looks more tentative: it is not clearly associated with the main territorial index, although some of the secondary outcomes suggest it may matter more for deserving-target solidarity than for solidarity in general. The new contextual checks do not support the preregistered H6 story in this pilot: EQI service quality points weakly negative, and relatively richer regions within countries are associated with higher rather than lower territorial solidarity.

7 Item-Level Ordered Logit Robustness

We need to check whether the pattern is actually being driven by only one territorial target. According to the preregistration we also do item-level checks.

The models below re-estimate the same predictor set separately for plsolmuni, plsolcntry, and plsoleu. We use ordered logistic regression rather than OLS.

outcome term_label estimate conf.low conf.high p.value
plsolcntry H1 Left ideology 0.559 0.276 0.848 0.000
plsolcntry H5b Institutional trust 0.487 0.120 0.856 0.009
plsolcntry H5a Social trust 0.480 0.158 0.805 0.004
plsolcntry H2b Perceived capacity 0.454 0.126 0.785 0.007
plsolcntry H3b Perceived risk 0.245 -0.031 0.526 0.083
plsolcntry H4 Norms and legitimacy 0.239 -0.014 0.490 0.062
plsolcntry H3a Objective need/risk 0.058 -0.247 0.364 0.708
plsolcntry H3 Risk × Capacity interaction -0.071 -0.338 0.193 0.600
plsolcntry H5c Service evaluations -0.246 -0.591 0.096 0.161
plsolcntry H2a Objective resources -0.321 -0.687 0.044 0.085
plsoleu H1 Left ideology 0.918 0.645 1.199 0.000
plsoleu H5b Institutional trust 0.542 0.197 0.893 0.002
plsoleu H2b Perceived capacity 0.449 0.145 0.756 0.004
plsoleu H5a Social trust 0.375 0.074 0.679 0.015
plsoleu H3b Perceived risk 0.263 0.003 0.528 0.049
plsoleu H3 Risk × Capacity interaction -0.001 -0.271 0.262 0.991
plsoleu H4 Norms and legitimacy -0.031 -0.270 0.210 0.799
plsoleu H3a Objective need/risk -0.054 -0.329 0.220 0.699
plsoleu H2a Objective resources -0.225 -0.571 0.117 0.198
plsoleu H5c Service evaluations -0.242 -0.572 0.086 0.148
plsolmuni H1 Left ideology 0.706 0.421 1.001 0.000
plsolmuni H5b Institutional trust 0.391 0.020 0.764 0.039
plsolmuni H5a Social trust 0.383 0.064 0.708 0.019
plsolmuni H4 Norms and legitimacy 0.303 0.039 0.566 0.024
plsolmuni H2b Perceived capacity 0.269 -0.052 0.593 0.101
plsolmuni H3a Objective need/risk 0.188 -0.114 0.493 0.224
plsolmuni H3b Perceived risk 0.107 -0.158 0.376 0.431
plsolmuni H2a Objective resources 0.013 -0.337 0.363 0.941
plsolmuni H5c Service evaluations -0.080 -0.416 0.256 0.642
plsolmuni H3 Risk × Capacity interaction -0.263 -0.526 -0.004 0.048

8 Optional Context Models

The GDP measure is centered within country to represent relative regional prosperity rather than absolute development.

model term_label estimate conf.low conf.high p.value
Context only H6a Regional service quality -0.179 -0.391 0.033 0.100
Context only H6b Relative regional prosperity 0.093 0.004 0.182 0.042
Full + Context H6a Regional service quality -0.158 -0.350 0.034 0.107
Full + Context H6b Relative regional prosperity 0.076 -0.016 0.168 0.109

9 Notes

  • The pilot file does not include design or post-stratification weights, so all models here are unweighted.
  • External contextual indicators are now attached automatically from EQI 2024 and Eurostat 2023 regional GDP files. T
  • The preregistration specifies random-effects and/or fixed-effects models to account for respondent clustering within regions and countries. Because the pilot sample is small and does not include the contextual variables that would benefit from a multilevel structure, this report uses OLS with country fixed effects and HC3 robust standard errors.

10 Secondary Ouftcome Models

The preregistration specifies repeating the main models for solidarity sub-dimensions: deserving targets, migrant-targeted solidarity, and welfare chauvinism.

outcome term_label estimate conf.low conf.high p.value
solidarity_deserving H1 Left ideology 0.226 0.145 0.307 0.000
solidarity_deserving H4 Norms and legitimacy 0.132 0.036 0.228 0.008
solidarity_deserving H5a Social trust 0.085 -0.021 0.192 0.119
solidarity_deserving H2a Objective resources 0.059 -0.067 0.185 0.358
solidarity_deserving H2b Perceived capacity 0.053 -0.058 0.165 0.351
solidarity_deserving H3b Perceived risk 0.048 -0.030 0.126 0.227
solidarity_deserving H5c Service evaluations 0.022 -0.096 0.139 0.719
solidarity_deserving H3a Objective need/risk -0.008 -0.119 0.102 0.884
solidarity_deserving H3 Risk × Capacity interaction -0.035 -0.123 0.054 0.445
solidarity_deserving H5b Institutional trust -0.056 -0.184 0.073 0.395
solidarity_migrant H1 Left ideology 0.567 0.439 0.695 0.000
solidarity_migrant H5a Social trust 0.210 0.070 0.350 0.004
solidarity_migrant H2b Perceived capacity 0.158 0.005 0.310 0.044
solidarity_migrant H5b Institutional trust 0.137 -0.025 0.299 0.099
solidarity_migrant H3b Perceived risk 0.132 0.005 0.258 0.043
solidarity_migrant H3a Objective need/risk 0.046 -0.113 0.204 0.574
solidarity_migrant H2a Objective resources -0.006 -0.182 0.170 0.946
solidarity_migrant H5c Service evaluations -0.037 -0.215 0.140 0.682
solidarity_migrant H4 Norms and legitimacy -0.040 -0.185 0.105 0.586
solidarity_migrant H3 Risk × Capacity interaction -0.129 -0.273 0.015 0.080
welfare_chauvinism H4 Norms and legitimacy 0.172 0.056 0.288 0.004
welfare_chauvinism H3 Risk × Capacity interaction 0.095 -0.024 0.213 0.118
welfare_chauvinism H2a Objective resources 0.065 -0.082 0.213 0.388
welfare_chauvinism H5c Service evaluations 0.059 -0.103 0.220 0.477
welfare_chauvinism H3a Objective need/risk -0.054 -0.199 0.091 0.467
welfare_chauvinism H3b Perceived risk -0.083 -0.201 0.034 0.164
welfare_chauvinism H2b Perceived capacity -0.105 -0.240 0.031 0.131
welfare_chauvinism H5a Social trust -0.124 -0.257 0.009 0.068
welfare_chauvinism H5b Institutional trust -0.193 -0.363 -0.023 0.027
welfare_chauvinism H1 Left ideology -0.341 -0.457 -0.226 0.000

11 Appendix

11.1 Outcome Distribution

11.2 Missingness Diagnostic

The preregistration specifies listwise deletion as the primary approach when missingness is limited. The table below reports the proportion of missing values for each key variable in the analysis, so the appendix preserves that information without interrupting the main argument.

variable n_total n_missing pct_missing
education_level 355 52 14.6
income_equiv 355 35 9.9
perceived_risk_index 355 9 2.5
perceived_capacity_index 355 4 1.1
gender_binary 355 1 0.3
migrant_background 355 1 0.3
solidarity_territorial 355 0 0.0
ideology_left_index 355 0 0.0
objective_resources_index 355 0 0.0
objective_need_index 355 0 0.0
norms_legitimacy_index 355 0 0.0
social_trust_index 355 0 0.0
institutional_trust_index 355 0 0.0
service_quality_index 355 0 0.0
age 355 0 0.0

After listwise deletion the full model retains 267 of 355 observations (75.2%). If overall loss exceeds ~20%, multiple imputation should be considered as a robustness check per the preregistration.