Identity Heterogeneity and Demand for Redistribution

We conduct an online survey experiment among members of the LISS Panel, a representative household panel in the Netherlands. The survey experiment will be offered to a random sample of the household panel, upon which panel members can opt-in the survey. With our survey experiment, we study the effect of heterogeneity in national identity on preferences for (re)distribution. Specifically, we consider that people can identify with the nation state in more ethnic or civic terms. We capture redistribution preferences by an incentivised allocation decision, in which a decision-maker is asked to distribute money over themselves and a recipient. The survey experiment has three treatment conditions and one control condition. In each condition, we vary the identity of the recipient, in order to activate in-group/out-group distinctions based on different national identity characteristics. In the first treatment, decision-makers know only that the recipient is Dutch, a national identity characteristic. In the second treatment, decision-makers know only that the recipient respects the Dutch political system and law, a civic national identity characteristic. In the third treatment condition, decision-makers know only that the recipient is born in the Netherlands, an ethnic national identity characteristic. In the control condition, the recipient is anonymous. We are interested in how the identity of the recipient affects the redistribution preferences of the decision-maker and whether this is dependent on their conception of national identity (measured prior to treatment).

2025-10-06

2025-10-28

Allocation decision (self vs. recipient); incentivised

4 statements capturing redistributive preferences (reducing income differences, the role of merit in relation to income differences, the role of the government in reducing income differences, and reducing differences among social classes); not incentivised

We conduct an online survey experiment among members of the LISS Panel, a representative household panel in the Netherlands, in which we study the effect of heterogeneity in national identity on preferences for redistribution. We capture redistribution preferences by an incentivised allocation decision, in which a decision-maker is asked to distribute money over themselves and a recipient. The survey experiment has three treatment conditions and one control condition. In each treatment condition, we vary the identity of the recipient. In the control condition, the recipient is anonymous.

Stage 1: Elicitation of prior beliefs
Measure of identification on different levels (i.e., neighbourhood, municipality, nation, Europe)
Measure of civic/ethnic national identity type
Measure of importance of organised life
Measure of importance of hobbies

Stage 2: Treatments & elicitation of outcome
Respondent is asked to allocate €150 over themselves and another respondent (recipient). For each treatment condition, we vary the identity of the recipient.

National Identity Treatment
Allocation decision (self vs. recipient that is Dutch); incentivised

Civic Identity Treatment
Allocation decision (self vs. recipient that respects the Dutch political system and law); incentivised

Ethnic Identity Treatment
Allocation decision (self vs. recipient that is born in the Netherlands); incentivised

Control Treatment
Allocation decision (self vs. anonymous recipient); incentivised

Stage 3: Elicitation of beliefs and background
Redistributive preferences, measured by 4 statements on reducing income differences, the role of merit, the role of the government in redistribution, and reducing differences among social classes
Anti-immigration beliefs, measured by 2 statements on the effect of immigration on the welfare state and on adapting to the destination’s culture and traditions
Measure of closeness to the recipient (differs by treatment)
Measures of assumed identity of the recipient (differs by treatment)
Measure of decision-maker’s civic/ethnic national identification
Measure of nationality
Measure of respect for Dutch political system and law

The randomization is done by computer.

The unit of randomization is the individual.

No

2800 individual respondents.

2800 individual respondents.

700 individuals in Treatment A, 700 participants in Treatment B, 700 individuals in Treatment C, 700 individuals in control condition.

Based on a power analysis conducted with Optimal Design with alpha at 5% and beta at 80% and assuming a standardized effect size of 0.15, we need approximately 700 respondents in each treatment condition. With 4 conditions, this results in a final sample of 2800 respondents.