Suppose a librarian receives an email from a man named Greg Walsh, wanting to become a cardholder and politely asking what he needs to do to make this happen (e.g., Does he need proof of address?) or simply inquiring about the open hours. Would the librarian reply? And, if so, would the reply be polite, including for instance some form of salutation, such as “Hello” or “Good morning”? Does your answer change if the guy is called Tyrone Washington? Is a librarian treating Jake Mueller differently from DeShawn Jackson?
Unfortunately, it turns out, the answer is yes.
Got answers? who’s asking?
Together with Corrado Giulietti and Michael Vlassopoulos, both at the University of Southampton in England, I sent out emails of this kind to approximately 4,900 libraries around the United States (as well as to several other public services, such as school districts and sheriff’s offices). We used the American Library Directory and contacted all libraries for which we could find a valid email address. We randomly assigned to each library one of the four fictitious names listed above, so that there are no systematic differences in the characteristics of libraries receiving an email from Greg (or Jake) rather than Tyrone (or DeShawn). In most cases, we used a general info address, e.g., “office@.” The results are reported in a study titled “Racial Discrimination in Local Public Services: A Field Experiment in the US,” which is forthcoming from the Journal of the European Economic Association, one of the most important academic journals in the field of economics.
What we found is that 69 percent of U.S. libraries sampled reply to requests from a person with a white-sounding name (we used Greg Walsh and Jake Mueller) while the response rate for those with a black-sounding name, either Tyrone Washington or DeShawn Jackson, was lower, at 65 percent. This difference of four percentage points is statistically significant: we can exclude with a high degree of confidence that it is owing to simple randomness.
It is in line with what we found for school districts and smaller than for sheriff’s offices, where the gap is seven percentage points. The gap for county treasurers is also around four percentage points, albeit probably not statistically significant owing to a smaller sample size, while for job centers and county clerks we find no evidence of a gap in the response rate.
Further analysis reveals that the gap is much stronger in libraries located in rural counties, with a whopping eight percentage points difference, while for libraries in urban counties the difference is only two percentage points. In our sample, only 30 percent of libraries are in rural areas. Looking at geographical areas in more detail is complicated, as the sample size becomes small, making the results less robust from a statistical perspective. With this caveat in mind, the analysis shows that the Midwest has a larger gap, six percentage points, compared to the Northeast, two percentage points, while the South and West are in line with the four percentage points.
TO WhoM Are we nice?
We also found a similar difference in terms of the politeness of the replies. While on average 72 percent of replies from librarians address the sender by name or contain some form of salutation, this is five percentage points less likely to happen if the reply is to a person with a black-sounding name. So, a person with a distinctively black name is not only less likely to receive a reply, if a reply is sent, it addresses the recipient in a less polite manner.
To check whether this gap was owing to discriminatory attitudes toward people from a low socioeconomic background rather than racial bias, in a second round of emails to the same libraries, we included in the signature “Real Estate Agent. Buy–Sell–Rent,” thus holding constant the profession for both black and white senders. This made no difference in the observed gaps, thus confirming that racism is the most likely reason for what we found.
Why emails matter
We believe that this measure captures a more general discriminatory attitude. A librarian not replying to requests for information coming from an African American may also treat African Americans differently in other aspects of library service.
A nationally representative survey conducted by the Pew Research Center in 2013 shows that library services are particularly important to “[w]omen, African Americans and Hispanics, adults who live in lower-income households, and adults with lower levels of educational attainment.” The type of behavior our study uncovered means, however, that libraries risk being part of the problem rather than the solution, and failing, or alienating, some of the users who need them most.
Mirco Tonin (@mircotonin) is Professor of Economic Policy at the Free University of Bozen-Bolzano in Italy.