Cybersegregation: Is Neil a More Desirable Tenant than Tyrone or Jorge?


Data for this study come from a primary data collection initiative conducted in Boston and Dallas over a 16-week period between January and early May of 2009. During that period, a systematic random sample of over 700 housing providers advertising rental housing units on Craigslist in each of these two metropolitan areas was selected, and e-mail inquiries about the advertised unit from equally matched prospective renters with white-, black-, and Hispanic-sounding names were sent. Audits or correspondence tests were conducted during the business week (i.e., Monday through Friday).

Several factors informed our choice of Boston and Dallas as the sites for our study. First, these metropolitan areas fall in very different regions of the country and, therefore, have different racial and ethnic compositions and histories. According to Census 2000, the Boston metropolitan area is 80 percent white, 7.3 percent black, and 5.9 percent Hispanic. In contrast, the Dallas metropolitan area is comprised of lower shares of whites (59.3 percent) and greater shares of minorities (13.6 percent black and 21.5 percent Hispanic). Second, both housing markets remain moderately-to-highly segregated, implying the persistence of discriminatory practices on the part of home providers. Finally, both areas are comprised of highly educated populations with relatively high levels of employment in professional, management, and related occupations, increasing the probability that residents have computers and use the internet to search for housing.

Why Craiglist? Craigslist is an ideal source of rental ads primarily because the first contact between housing provider and prospective renter can occur via e-mail. Although the ads on Craigslist often provide a phone number, they always provide an e-mail address, encouraging home seekers to contact the housing provider directly via e-mail. Other electronic search engines, such as those associated with the real estate sections of the on-line versions of the local newspapers in these areas, do not always insure direct e-mail contact between the housing provider and prospective renter. Another reason why Craigslist is an ideal source is because its format is similar in these two metropolitan areas. A final reason why Craigslist was chosen for this informal test is because of the geographic information often provided in the listings.

The white and black names that were used in the audit in responses to the electronic ads for rental housing come from the name frequency data used by Bertrand and Mullainathan (2004) in their job market study. Two sets of these names were used in audits to decrease the chances that housing providers might detect that audits were underway. For whites, the names used were Neil Baker and Matthew O’Brien, and for blacks the names were Tyrone Jackson and Tremayne Robinson, names that were selected in the Boston and Chicago employment study due to their high racial/ethnic identification. For Hispanics, we used the names Jorge Rodriguez and Pedro Gonzales because of they are easily recognizable as “Hispanic” and therefore have high face validity.

The key dependent variables in our analysis are derived from the outcomes generated by the e-mail inquiries responding to random samples of ads for rental housing from the Boston and Dallas Craigslist websites. Specifically, we recorded whether each auditor received a response or more than one response and if they were told the advertised units are available, were invited to inspect the unit, were advised to contact the housing provider, and were informed of additional units that are available.