From 2015 to 2016, close to one million refugees arrived in Germany, largely from countries with a pre-dominantly Muslim background. Muslims are the social group in Europe in general and Germany in particular that currently faces the strongest negative attitudes. We study the prevalence and dynamics of taste-based and statistical discrimination in a representative sample of the German population by using a two-wave online survey. Respondents choose between two other participants who to participate with in two-person coordination games. In addition, we elicit respondents’ behavior and beliefs with respect to the coordination games and measure relative prosocial preference with respect to the other participants in separate allocation games. Respondents know about the other participants whether they currently live in refugee shelter homes in Germany or not and see pictures of their faces. We exploit two sources of exogenous variation: (i) Pictures reveal whether headscarves are worn or not. (ii) Respondents do or do not receive information between waves—right after Wave 1 and/or right before Wave 2—on group level behavior of refugee and non-refugee participants in previous coordination games.
External Link(s)
Citation
Schwerter, Frederik, Matthias Sutter and Christian Unkelbach. 2017. "A representative investigation of discrimination against Muslims in Germany." AEA RCT Registry. October 23. https://doi.org/10.1257/rct.2382-1.0.
1.: Refugee or non-refugee subject is chosen to play the coordination game with
2.: How main subjects choose to behave in the coordination games
3.: Beliefs of how refugee and non-refugee subjects played in previous coordination games
4.: Budget allocations between refugee and non-refugee subjects
Primary Outcomes (explanation)
Secondary Outcomes (end points)
1. Political Preferences 2. Moral Preferences
Secondary Outcomes (explanation)
1.1.: What party would/did subjects vote for in previous/current federal elections
1.2.: Using likert scales, how much do subjects like immigration
1.3.: Do subjects approve of Angela Merkel's immigration policy
1.4.: Do subjects agree with Germany' former president that Islam belongs to Germany
2.: Using the "moral foundations theory" questionnaire
Experimental Design
There are three groups of subjects:
1) Main participants (MP)
2) Residents of a German refugee shelter home (R)
3) Non-refugee residents of the region in Germany where the shelter home of 2) is located (C)
Wave 1 and Wave 2
MPs are asked whether they would prefer to participate in two-person coordination games with one of two other participants that are presented to them. One of them is a subject from group R and the other is from group C. MPs know this and, additionally, see a picture of each of them.
MPs know that their choice will be implemented with a chance of 90%, but that with 10% their choice will be overruled and their other participant will be determined by a coin flip. Before knowing who their other participant is, MPs state their behavior in the game for each potential other participant.
In addition, MPs were told that all subjects from groups R and C played the same coordination games with each other in a previous experiment. We measure MPs’ beliefs of how subjects from group R played the game when paired together, how two subjects from group C played the game when paired together, and how one subject from group C and one subject from group R played the game when paired together. MPs also allocated budgets between the three pairs of subjects—one pair of subjects from group R, one pair of subjects from group C and one mixed pair—in three separate allocation games.
Coordination Game Wave 1
Subjects choose one among 5 options. If they choose the same option, they have a chance of winning a price. Otherwise they have no such chance. The choice options are labelled: “15min too early,” “5min too early,” “On time,” “5min too early,” and “15min too late.”
Coordination Game Wave 2
Subjects choose one among 5 options. If they choose the same option, they have a chance of winning a price. Otherwise they have no such chance. The choice options are labelled:
“Lemonade,” “Bier,” “Tea,” “Hot Lemon,” and “Pineapple Juice.”
Treatment Dimension 1
The pictures of group R subjects show that they wear headscarves or not.
Treatment Dimension 2
Right after Wave 1 and/or right before Wave 2 subjects receive information on how all subjects from groups R and C behaved in Coordination Game Wave 1 when playing it previously among each other.
Experimental Design Details
Randomization Method
Randomization done by a computer
Randomization Unit
Individual
Was the treatment clustered?
No
Sample size: planned number of clusters
3000 participants in Wave 1 and as many as possible (up to 3000) in Wave 2
Sample size: planned number of observations
3000 participants in Wave 1 and as many as possible (up to 3000) in Wave 2
Sample size (or number of clusters) by treatment arms
Roughly 1500 participants will see pictures of Refugees wearing headscarfs and the remaining participants will see pictures of Refugees that do not show headscarfs.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)