Experimental Design
We run controlled experiments in which decision-makers with either a native German or French background (which we refer to as `spectators') are faced with a situation in which two people, called worker A and worker B in the experiment, have earned an amount of 20 Euro in total for performing a task. One worker has the same majority background as the spectator and the other has a minority background. More specifically, German (French) spectators are matched with minority background workers with Turkish (Maghrebi) origins. Turks and Maghrebi represents the largest non-Western immigrant groups in Germany and France, respectively.
The spectators are explained that worker A and worker B have been recruited via an online platform and that they both successfully completed a task. Spectators are also informed about the workers' gender and age range. Furthermore, they are told how the workers are called by their first name, which is a signal of their ethnic background. In fact, we do not reveal the actual first names of the workers but use a nick-name that provides a correct signal of the workers' ethnic background and, at the same time, preserves workers' anonymity. For each of the four different groups of workers (i.e. native Germans and French, Turkish and Maghrebi immigrants), and for each gender, we put together a list of 10 first names that are common among people belonging to the respective groups. Every worker is then randomly assigned a name from the list of nick-names that is relevant for his/her group, and this name is communicated to the spectator.
Spectators are informed that workers' earnings for performing the task were determined according to a randomly drawn payment scheme out of the following three schemes:
- Worker A and worker B earn 10 Euro each. (condition Baseline)
-Worker A earns 15 Euro and worker B earns 5 Euro. (condition Majority Lucky)
-Worker A earns 5 Euro and worker B earns 15 Euro. (condition Minority Lucky)
The task of the spectators is to decide how much money, if any, they choose to redistribute between worker A and worker B. Spectators are paid a fixed fee for their decision and are informed that at the end of the study, the redistribution decision of 10% randomly selected spectators will be implemented for real. Workers receive a fixed fee to complete the task and a variable fee, which depends on the condition they are allocated to and on the redistribution decision of the matched spectator. The complete set of instructions given to the workers and spectators are provided in a separate document. At the end of the experiment, spectators answer a survey with questions on a number of socio-demographic characteristics and traits, such as political orientation, views on redistribution and attitudes to ethnic diversity in society. The instructions document includes the survey questions.
The spectators and the workers in the experiment are recruited by survey company Ipsos. We aim at recruiting a sample of 6000 spectators, 3000 with a native German and 3000 with a native French background. Each spectator is randomly allocated to only one of the three experimental conditions. We recruit 600 workers in total, 100 for each country/condition combination. Each worker completes two surveys, the payment for each survey is determined by the decision of a randomly matched spectator (i.e. two spectators are matched with each worker). Workers are selected on the basis of their country of birth and their parents' country of birth. For example, minority workers in the German experiment are either born in Turkey or have parents who are both born there.