Market Discrimination and Social Media

How to cite

Moritz, Raphael; Manger, Christian; Pull, Kerstin. 2021. "Market Discrimination and Social Media." AEA RCT Registry. AEARCTR-0007627. https://doi.org/10.1257/rct.7627-2.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Raphael Moritz* , Christian Manger , Kerstin Pull

* Primary investigator

Registry ID
AEARCTR-0007627
DOI
https://doi.org/10.1257/rct.7627-2.0
First registered
Last updated
Status
Completed
Trial dates
2021-01-01 – 2022-12-31
Countries
Germany ()
JEL codes
R23, J15, J16, C93
Registry record
http://www.socialscienceregistry.org/trials/7627

Abstract

We conduct a correspondence test in the market for shared housing to identify the magnitude of ethnic and gender discrimination and the role of personal information posted on a social media platform. About 4,900 fictitious applications with a randomly assigned Turkish-, or German-sounding female or male name will be sent to vacant room ads. In addition, we randomly add a link to a social media profile to determine the effect of such personal information on callback rates. We carefully constructed these fictional social media accounts on Instagram for over the course of two years to make them as realistic as possible, conducting several surveys among young people to ensure that the profiles are close to reality with a sufficient number of followers. Unlike previous research, we can approximate whether landlords or roommates actually visit these profiles and exploit this information using statistics on profile visits and page impressions.

Experimental design

We use a non-matched pair randomized controlled trial (correspondence test design) to test for ethnic and gender discrimination in the market for shared housing in Germany. To do this, we randomly vary the applicant’s name, which signals an ethnic minority or ethnic majority member, and we randomly vary gender. In addition, we vary the amount of information by randomly providing a link to a social media profile that contains additional information about the applicant.

Design details

Randomization method
Computer generated (pseudo) randomization using a Python program
Randomization unit
Individual level
Clusters
n/a
Observations
3,720
Arms
Assuming a 44% percent overall callback rate (from pre-study) we will send out approximately 620 applications for each of the eight treatment groups (immigrant/native*female/male*with/without social media info), resulting in 1,240 applications per treatment arm. Note that the initial sample size (see above) was higher in a previous version of this pre-registration because we decided to split the trial into two parts after the first trial was completed. We submitted a new pre-registration for the second study (see AEARCTR-0011322).
Primary outcomes
Callback rates conditional on ethnicity, gender and social media information

Cited by

No citation data available for this registration yet.

Keywords