Authoritarian Preferences

In the main experiment, participants read news articles (Part 1 of the experiment) and are subsequently matched to four groups with different compositions of members of majority and minority. We define minority as those individuals with a name most ethnic Germans would identify belonging to Muslims, namely but not exclusively, Turkish, Arabic, Persian or some names originating from the Balkans. We define majority as individuals with an ethnic German name. The four groups vary in their composition comprising 0, 1, 4, or 9 minority members out of 10 group members. The group members will have their first name displayed during the main experiment. See below for the construction of the reference sample.
Participants in the main experiment will distribute money between themselves and each of these groups separately (Part 2 of the experiment). They will subsequently state their preference for decision-making rules that determine the distribution of money (Part 3 of the experiment).
Our main treatment variation is the type of news articles participants read before they make distribution decisions. They will either read 10 neutral news items without person identifiers or 10 articles that constitute a power threat. Our main power threat treatment comprises news articles of success stories of minority members. Each of these articles provide information about one minority group member living in Germany who either set up a successful firm or made important societal contributions. In a second power threat treatment, we introduce ethnic German success stories. Articles are balanced with respect to gender with 5 female success stories and 5 male success stories. Neutral news articles are constructed on the basis of power threat articles. Here, individuals’ information is replaced with sectoral information or a focus on the innovation itself (if relevant).
Participants receive a fixed payoff for their participation. Their allocation decisions are incentivized as follows: At the end of the experiment, one participant and one allocation part is chosen per treatment for additional payment. Part 2 is chosen with 90 percent probability and Part 3 with 10 percent probability. With Part 2 being chosen, one of the groups is further with equal probability randomly chosen, and the participant’s allocation decision when facing that group determines his own additional payoff as well as the additional payoff of that group’s members. If Part 3 is chosen, the participant’s choices determine the decision-making rule that decides on the allocation of payoffs across the four groups. The participant here is making a decision behind the veil of ignorance, and thus, his or her payoff is equal to one randomly selected person out of the four groups.

After collecting data for our main experiment as listed in its original pre-registration, we observed a differential treatment effect in West and East Germany. Since our sample was aimed to be representative with respect to German state residency, we have relatively few observations from East Germany (formerly German Democratic Republic). We therefore collect more data from these to test whether indeed the treatment effect is only present in East Germany.

Once the East German data collection has been completed, we observed meaningful treatment effects based on location, but some of these effects were marginally significant. We therefore collect further data both in West and East Germany, so that per treatment, there are about 300 participants. This gives us 80 percent power to detect the effect size derived from the prior collected data at 5 percent significance level.

Finally, we will introduce a second power threat treatment, where ethnic German success stories are depicted. We will run this study only in East Germany to see whether the effect we observe in the data is due to success stories in general triggering discrimination of minorities, or this only happens with migrant success stories.

2023-06-21

2025-04-30

The main outcome variables comprise the allocation decisions in Part 2 and Part 3.
In Part 2, the outcome variable is the amount of money allocated to individuals based on their minority or majority membership.
The main outcome variable from Part 3 is the points allocated to authoritarian decision-making rules, decision rules 2 and 3. Further outcome variables are the points allocated to groups and individuals per group membership in decision rules 2 and 3, respectively.
Primary outcome variables will further be separately analysed based on whether participants are from former West or East Germany.

Emotions, authoritarianism measures, attitudes towards migration and the measure of power threat we employ. We will aggregate the answers to these measures and conduct separate analyses on them.
Exploratory analysis: Prior authoritarianism measures and social dominance orientation.

The experiment consists of four parts.
Part 1 consists of reading 10 news articles.
In Part 2, participants are matched with four groups, and each group separately. They are provided the first names of group members in each group. When facing each group, they decide how to allocate 330 euro across themselves and group members.
In Part 3, participants are informed that the members of the reference sample made decisions under three different decision rules. For all decision rules, each member of the four groups made an allocation decision for members of the four groups –including herself– separately.
- Decision rule 1: majority rule
- Decision rule 2: one group rule
- Decision rule 3: one person rule
Next, participants are asked how much they prefer each of the three rules.
In Part 4, participants fill in
1. Discrete emotions questionnaire (Harmon-Jones, Bastian, and Harmon-Jones 2016),
2. Their demographics: their place of birth, gender, age, the state they live in, monthly net household income, marital status, number of children, where their parents are born, highest education level achieved, their employment status,
3. Political and religious identification: political position (left and right), the party they would vote for if elections were to be held this weekend, whether they voted in the last federal elections, religion
4. Power threat: how much they agree that powerful positions both politically or in leading firms should be occupied by different types of groups, and how much they think the government should take into account the ethnic or religious background on those living in Germany.
5. Authoritarianism measure questions based on Altemeyer (1998), Costello et al. (2020) and MacWilliams (2016).
6. Social dominance orientation, the short version based on Ho et al. (2015).
7. Support for immigration, support for redistribution, attitudes towards immigrants and beliefs about immigrants’ poverty and unemployment based on Alesina, Miano, and Stantcheva (2022).

The reference sample is constructed via a laboratory experiment.

The experiment consists of four parts.
Part 1
Part 1 consists of reading the 10 news articles. The articles are presented in a random order. Participants then answer some questions pertaining to the news articles they have read to ensure that they pay attention to the news articles. Correct answers will be monetarily incentivized with a small payment.
Part 2
Participants are matched with each group separately. They are provided the first names of group members in each group. When facing each group, they decide how to allocate 330 euro across themselves and group members. This part is implemented for one participant per treatment with 90 percent probability.
Part 3
Participants are informed that the members of the reference sample made decisions under three different decision rules. For all decision rules, each member of the four groups made an allocation decision for members of the four groups –including herself– separately.
- Decision rule 1: the payoff of each individual is the average allocation to that person by members of the four groups.
- Decision rule 2: the payoff of each individual is the average allocation to that person by members of one group.
- Decision rule 3: the payoff of each individual is equal to the allocation to that person by one group member.
Next, participants are given 100 points, where each point translates into 1 percent probability. They are asked to allocate the 100 points across the three decision-making rules. They are also informed that they will subsequently decide which group or individual to choose for decision rules 2 and 3.
In the next screen, they make the same allocation decision for decision rule 2. Participants are again provided with 100 points to allocate to the four groups. Finally, for decision rule 3, participants are provided with 100 points to allocate to individuals.
At the end of the experiment, one participant from the main experiment and one part will determine the final allocation outcomes. Should Part 3 be chosen for payment, the final decision rule that allocates money across the 40 group members will be chosen with probabilities equal to those assigned by the randomly chosen participant from the main experiment. If this process leads to decision rule 2 to being chosen, the group that will determine the final allocation decision will be chosen based on the probabilities assigned to groups by that participant. Likewise, if the process leads to decision rule 3 to being chosen, the individual that will determine the final allocation decision will be chosen based on the probabilities assigned to individuals by that participant.
Part 4
Participants fill in
1. Discrete emotions questionnaire (Harmon-Jones, Bastian, and Harmon-Jones 2016),
2. Their demographics: their place of birth, gender, age, the state they live in, monthly net household income, marital status, number of children, where their parents are born, highest education level achieved, their employment status,
3. Political and religious identification: political position (left and right), the party they would vote for if elections were to be held this weekend, whether they voted in the last federal elections, religion
4. Power threat: how much they agree that powerful positions both politically or in leading firms should be occupied by i) ethnic Germans, ii) western Europeans, iii) Europeans, iv) Christians, v) people born in Germany, or vi) anyone living in Germany, and how much they think the government should take into account the ethnic or religious background on those living in Germany.
5. Authoritarianism measure questions based on Altemeyer (1998), Costello et al. (2020) and MacWilliams (2016).
6. Social dominance orientation, the short version based on Ho et al. (2015).
7. Support for immigration, support for redistribution, attitudes towards immigrants and beliefs about immigrants’ poverty and unemployment based on Alesina, Miano, and Stantcheva (2022).

Construction of the reference sample:
The reference sample is constructed via a laboratory experiment. We will screen the registered pool to pick minority (as well as majority) names and invite them to the experiment to generate the exact group compositions with respect to minority and majority members. At the beginning of the experiment, participants enter their first name and are told that these names are seen by current session participants as well as future participants of the study. Afterwards, they are assigned to one of four groups. They then decide to distribute 1200 euro among 40 people whose names are displayed. They make this decision three times: First, under majority voting where all 40 decisions per participants are averaged. Second, under one-group rule where the average decision of their own group determines the payments. Third, as a dictator. They are further informed that one of these decision rules may be implemented depending on the choices of future participants. In the reference sample, we also elicit participants’ preferences for the three different decision rules with the same method as in Part 3 above. We make it clear that these decisions have no payoff consequences. We also administer the same questionnaire at the end.

Computer randomisation

Experimental session

No

First data collection: 400
Second data collection: 200 participants
Third data collection: 990 participants

1590

600 control (about 300 in West and East Germany each), 600 power threat-minority (about 300 in West and East Germany), and 390 power threat-majority from East Germany.

Based on GPower 3.1, in a linear regression with three predictors and a sample size of 400, with α=0.05, and power at 0.80, the minimum detectable effect size f2 is 0.019. Amendment for second data collection: Based on Stata 16, with a t-test comparing two means in the main predictors with α=0.05 and power=0.80, 200 new observations from East Germany allows us to detect the observed treatment difference in the main sample for points allocated to the dictator rule and combined points allocated to groups with 0 and 1 person with a migration background in the one-group rule. Amendment for third data collection: We have previously collected in control, 190, and 148 participants in West and East Germany, and in power threat-minority, 192, and 146, respectively. Based on the realised effect sizes in this data, increasing the sample size gives us at least 80 percent power with α=0.05 to detect the following difference-in-differences according to G*Power 3.1: 1) the change in the percentage of participants who support an authoritarian rule in power threat-minority in West and East Germany, 2) the change in the amount of money allocated to participants controlling for minority status in power threat-minority in West and East Germany, 3) the change in the amount of points allocated to participants in the dictator rule controlling for minority status in power threat-minority in West and East Germany. In a similar vein, the power threat-majority has at least 80 percent power to detect effects similar to power threat-minority. Alternatively, if it has no effect as we hypothesize, the difference between power threat-minority and power threat-majority within East Germany can be detected with 80 percent power.