If you have a foreign name and you are looking for an apartment, you will significantly less likely to be contacted by landlords than a German applicant. In particular, people with Arab and Turkish names are severely discriminated against. This is shown by a large-scale experiment conducted by the data journalists of Bayerischer Rundfunk and SPIEGEL. To prove that this kind of discrimination was not coincidence but systematic disadvantage, we sent out 20,000 apartment inquiries in an automated process over several weeks with fictitious German and non-German profiles. The ads we responded to appeared on the two biggest German online housing portals. We received around 8,000 answers to our inquiries – and their evaluation proves that people with foreign names are clearly discriminated on the rental market.
What makes this project innovative?
We set up an automated process that allowed us to conduct a large scale undercover research. The dataset that formed the basis for our story consisted of numerous e-mails and was generated by ourselves. Like this, for the first time, a German media outlet was able to prove that the systematic discrimination on the rental housing market against strangers really existed. Before this phenomenon only existed as a gut feeling. To supervise the technical process and herewith provide a consistent dataset, we developed a dashboard that showed a realtime analysis of the e-mail dumps.
What was the impact of your project? How did you measure it?
The results of our research were primarily published in a planned campaign on TV (Das Erste), radio (ARD), print (Der Spiegel) and online (Tagesschau.de, Spiegel.de, BR.de). Media all over Germany reported on the outcomes of our research. Immediate reactions came from high-ranking politicians such as Martin Schulz (back then German candidate for chancellor), German Minister of Justice Heiko Maas and the Bavarian Home Secretary Joachim Hermann. They all pointed out that discrimination based on the origin of an applicant is illegal. Scientists and students from several countries have contacted us to know more about our results and our methodology. We also got dozens of mails from affected people that had experienced exactly this kind of discrimination and who told us their story. The Berlin Anti Discrimiantion Office is planning to adopt our method.
Source and methodology
For detailled information on our methodology, please visit the following page that we published with our story:https://www.hanna-und-ismail.de/english/methodology.html
We used selenium browser automation to fill out the forms on the housing portals automatically. The automation programm was running on a web server with a VPN client and changing IP adresses to avoid being blocked by the housing portals. The e-mails we received as confirmation to our requests and answers from the landlords were exported on a daily basis into a mysql database and analyzed almost in realtime by a script in R. The descriptive analysis and the calculation of the discrimination rates were also made in R. All the code used for the calculations was published on our Github profile. In several thousand cases we could not match answer e-mails from landlords to the corresponding requests automatically. To make manual matching easier we imported the data into Microsoft Access and built an interface that made the matching by hand much easier and faster.
BR Data: Ulrike Koeppen, Steffen Kühne, Oliver Schnuck, Robert Schoeffel;SPIEGEL ONLINE: Christina Elmer, Patrick Stotz, Achim Tack;