We conduct an information provision experiment to investigate the relevance of statistical information for economic attitudes towards immigration. Our experimental design is embedded into a large-scale representative online survey. We randomize the provision of information on the share and the unemployment rate of foreigners, representing facts about immigration related to the size and economic characteristics of the immigrant population, respectively. We aim to analyze the effect of information provision on two prominent economic channels of immigration attitudes: welfare state and labor market concerns about immigration. In addition, we examine whether biases in beliefs about immigration translate into immigration policy preferences and preferences for redistribution in host societies.
External Link(s)
Citation
Bareinz, Patrick and Silke Uebelmesser. 2020. "Biased Beliefs about Immigration, Economic Concerns, and Information Provision." AEA RCT Registry. November 30. https://doi.org/10.1257/rct.6819-2.0.
In our main experiment, participants are randomly assigned to one of four experimental groups which differ in terms of the type and quantity of information provided. Specifically, respondents in one of the three treatment arms receive information on either the share of foreigners in Germany, the unemployment rate of foreigners in Germany, or on both of these immigration statistics. We then aim to investigate whether information provision affects attitudes towards immigration and preferences for redistribution in host societies.
Our survey experiment also contains a fifth experimental group. This group is similar to the passive control group in that it receives no information. However, we randomize the order of a question block on the COVID-19 crisis between the control group and the fifth group in order to investigate whether there exist priming effects on our outcome variables related to the ongoing health and economic crisis.
Intervention Start Date
2020-11-27
Intervention End Date
2020-12-11
Primary Outcomes (end points)
Welfare state concerns about immigration
Labor market concerns about immigration
Immigration policy preferences
Preferences for Redistribution
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Our experimental design consists of four stages. In the first stage, we elicit respondents' prior beliefs about the share and the unemployment rate of foreigners. In addition, we elicit prior beliefs about the general unemployment rate, serving as a benchmark for respondents' overall beliefs about official statistics.
In the second stage, random subsets of respondents are provided with true information about the share and/or the unemployment rate of foreigners: - Treatment arm A: receives information on the share of foreigners (representing the size of the immigrant population)
- Treatment arm B: receives information on the unemployment rate of foreigners (representing economic characteristics of the immigrant population)
- Treatment arm C: receives information on the share and the unemployment rate of foreigners (representing a bundle of both types of information)
- Control group (passive): does not receive any information
The two types of statistical information distinguish between facts about the size of the immigrant population, and information on its economic characteristics, represented by the unemployment rate of foreigners. In addition, treatment arm C allows us to investigate the dimension of quantity in terms of a bundle of both types of information.
In the third stage, respondents are asked literature-based survey measures of our outcome variables: welfare state and labor market concerns about immigration (i.e. economic concerns), immigration policy preferences, and preferences for redistribution.
In the fourth stage, we elicit respondents' posterior beliefs about the share and/or the unemployment rate of foreigners for those experimental groups which receive information on the respective statistics.
The fifth experimental group allows for the examination of potential priming effects regarding the COVID-19 crisis and also represents an active control group with respect to our main experiment. While respondents in this group do not receive any information, they are primed by a question block on COVID-19 to think about the ongoing health and economic crisis before answering to our outcome measures. We then examine whether respondents in this group differ in terms of their immigration and policy attitudes when compared to the passive control group.
Experimental Design Details
Randomization Method
Randomization done by a computer
Randomization Unit
Individual
Was the treatment clustered?
No
Sample size: planned number of clusters
No clustering
Sample size: planned number of observations
3000 individuals
Sample size (or number of clusters) by treatment arms
Approximately 600 individuals for each experimental group
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)