National Study on Ethnic Discrimination in the Housing Market

Today I’m giving you the first national field experiment on ethnic discrimination in the housing market. Financed by the Swiss Office for Housing and the NCCR on the move, we examined to what extent one’s name affects the likelihood to be invited to view an apartment. We covered the entire country, across language regions and across urban and rural areas.

Between March and October 2018, our diligent research assistants sent more than 11,000 enquiries to over 5,700 landlords in all parts of Switzerland. We varied the name of the person sending an enquiry (stimulus sampling) along with other features such as politeness or the family situation. Overall over 70% of the enquiries were answered positively in the sense of an invitation to view the apartment or steps in this direction.

We find no clear differences between commercial and private landlords. The response rate for women was around 1 percent higher, while highly qualified people had a 2 percent higher response rate, especially academics who use their doctoral title (we dind’t expect this to make such a big difference when we designed the study). As previous field experiments have shown, the quality of the message we sent affected the probability of a response: Compared to a standard text, the response rate for friendlier queries is about 5 percent higher, while queries with the default text from online portals show a 10 percent lower response rate.

We find evidence of ethnic discrimination in the sense of unequal treatment based on the name. Enquiries with names from neighbouring countries (Germany, Italy, France) were even invited somewhat more frequently to view apartments than those from Switzerland, but people with Kosovar (response rate just under 3 percent lower) or Turkish names (response rate about 5 percent lower) have significantly fewer chances of being invited for a viewing. Whether those interested were naturalised with foreign-sounding names or stated that they had a permanent residence permit was hardly a factor. The rate of discrimination we observe is similar in order of magnitude to that found in comparable studies in other Western countries.

With the national coverage, we can also observe variation in responses by local context where the property is located. In municipalities with higher rental prices, the positive response rate is higher for everyone, and a higher vacancy rate in the municipality is associated with a higher response rate, except for people with Kosovar names. In urban areas the probability of discriminating against people with foreign names is lower. We also find that people with foreign-sounding names are less likely to be invited in municipalities with restrictive political attitudes towards immigration (as measured in the results of popular initiatives and referendums).

Auer, Daniel, Julie Lacroix, Didier Ruedin, and Eva Zschirnt. 2019. ‘Ethnische Diskriminierung auf dem Schweizer Wohnungsmarkt’. Grenchen: BWO. https://www.bwo.admin.ch/bwo/de/home/Wohnungsmarkt/studien-und-publikationen/diskriminierung-auf-der-schweizer-wohnungsmarkt.html.

Audit Studies — The Book

There’s a new book edited by S. Michael Gaddis on audit studies. The subtitle promises to go behind the scenes with theory, method, and nuance — and this is what the book provides. As such, the book is a much needed contribution to the literature, where we typically see the results and little how we got there. With (not so) recent concerns around researcher degrees of freedom, the tour behind the scenes offered by the various chapters are an excellent way to make visible and apparent the ‘undisclosed flexibility’ as Simmons et al. called it in 2011. It’s one thing to discuss this in abstract terms, and it’s another thing to sit down with actual research and reflect on the many choices we have as researchers. Indeed, public reflection on research practices may be relatively rare in itself when it comes to quantitative research.

The book comes with a dedicated support webpage: http://auditstudies.com/ (do me the favour to update the “coming soon” banner). On this website, several chapters can be downloaded as pre-prints, though it’s not all the contents if someone is looking for a free book. I hope the authors will make their code available on the website as promised in several places in the book, because this will be another greatly helpful resource for those new to audit studies or looking for new directions.

I greatly enjoyed to read the reflections by other researchers doing audit studies, and would definitely recommend the book to anyone thinking of doing an audit study. At times there were passages that seemed a bit redundant to me, but all the chapters are written in such an accessible way that this didn’t bother me much. Where I think the book falls a bit short is on two fronts. First, it is very US-centric. In itself this is not an issue, but there are several instances where the authors don’t reflect that perhaps in other countries the markets are not organized the same way. In my view, a comparison to other countries and continents would have been fruitful to underline some of these assumptions — I’ve tried to just this on attitudes to immigrants. Second, the book is not a guidebook. I know, it doesn’t claim to be one, but the book asks so many (justified) questions and offers comparatively few concrete guidelines like Vuolo et al. offer it on statistical power. In this sense, the book will stimulate readers to think about their own research design and not provide a template. And this is actually a good thing, because as the chapters make apparent without normally saying so, there is no universal approach that suits different markets in different places and at different times.

So, should you buy the book? Yes if you want to carry out your own audit study, yes if you want to better understand and qualify the results of audit studies, and yes if you’re looking for guidelines — because the book will make you realize that you’re largely on your own. What would probably useful, though, would be a checklist of things to consider, something readers will have to create themselves on the basis of chapters 4 (Joanna Lahey and Ryan Beasley), 5 (Charles Crabtree), and 6 (Mike Vuolo, Christopher Uggen, and Sarah Lageson).

Gaddis, S. Michael, ed. 2018. Audit Studies: Behind the Scenes with Theory, Method, and Nuance. Methodos 14. New York: Springer. https://www.springer.com/cn/book/9783319711522

Ruedin, Didier. 2018. ‘Attitudes to Immigrants in South Africa: Personality and Vulnerability’. Journal of Ethnic and Migration Studies. https://doi.org/10.1080/1369183X.2018.1428086.

Simmons, Joseph P., Leif D. Nelson, and Uri Simonsohn. 2011. ‘False-Positive Psychology: Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant’. Psychological Science 22 (11): 1359–66. https://doi.org/10.1177/0956797611417632.

Vuolo, Mike, Christopher Uggen, and Sarah Lageson. 2016. ‘Statistical Power in Experimental Audit Studies: Cautions and Calculations for Matched Tests With Nominal Outcomes’. Sociological Methods & Research, 1–44. https://doi.org/10.1177/0049124115570066.

Zschirnt, Eva, and Didier Ruedin. 2016. ‘Ethnic Discrimination in Hiring Decisions: A Meta-Analysis of Correspondence Tests 1990–2015’. Journal of Ethnic and Migration Studies 42 (7): 1115–34. https://doi.org/10.1080/1369183X.2015.1133279.>/small>

Switzerland’s Cantons as a Most-Similar Design?

From time to time I come across research that claims that Switzerland is the ideal place for many comparative studies. The argument is that the 26 cantons of Switzerland offer similar institutional settings, yet have plenty of autonomy in other regards. The intuition is that the federalism inherent in the Swiss system allows to hold constant many aspects while testing for other differences: a most-similar research design.

While I like the idea of working with sub-national data, I’d suggest a more careful look: in some cases the situation may indeed allow most-similar design, but in others this does not seem to be the case. Here I simply list a few reason why the Swiss cantons may not be the ideal place for comparative studies. Whether this matters will depend on the individual study.

Let’s start with the institutional setting. In broad terms, it’s the same across Switzerland: three levels of government (municipal, cantonal, national). When it comes to details, however, we can see that autonomy means something in Switzerland. Let’s take just two examples. Each canton has an authority to deal with immigration issues (on its own or attached to another service), except for the Canton of Bern which has four of them. The electoral system for the lower chamber is the same in all cantons, for the upper chamber it’s up to the canton to decide — and they make use of this.

Even though there are efforts to harmonize the educational system in Switzerland (a bit over half the cantons now harmonize their systems), there’s still much variety.

More profound differences may exist between regions. The difference between language regions is often highlighted, but there’s more than the share of French-speakers in a canton. The traditionally dominant religion in the different cantons varies and cuts across the language regions. If we look at distance to traditional trade routes through Switzerland (think contact hypothesis), we get a different picture still. And then there is history: different parts of Switzerland have become part of what is present-day Switzerland in quite different circumstances (something many accounts apparently manage quite well to gloss over).

Don’t care about culture? Let’s consider the economy. While Switzerland overall is a rich country, there’s a big difference between economic centres like Zurich, Geneva, or Basel, and other places. We could consider the degree of urbanization here, or the importance of agriculture, average incomes, average rents, and other economic factors.

The point here is not to dismiss the claim of a most-similar research design, but to highlight that the situation is not as simple as often implied: there are significant institutional, socio-economic, and cultural differences within. Whether they matter will depend on the research question.