To test whether people with names associated with a specific religion were treated differently when applying for a house or flatshare, the Guardian sent almost 1,900 requests to properties listed in five locations – London, Aberdeen, Devon, Sunderland and Leicester – within a two-day period in the last week of October.
The wording of the two requests were different but similar in tone and length. To avoid any possibility of being rejected on a first-come-first-served basis, the requests were sent within a few minutes of one another and were alternated, meaning that in 50% of cases “David” made first contact and in 50% “Muhammad” did.
The sample size was 948 ads in total. The responses were categorised into positive/neutral responses (those who opened a conversation and/or offered a viewing), negative responses (for example a message saying the room was gone) and no response (those who had not responded to the initial message within a week of the initial contact). A positive or neutral response was assigned a value of one and a negative or no response was assigned a value of zero.
David received 449 negative responses and Muhammed 546 negative responses to the 948 requests each sent.
We tested the statistical significance of the findings using a Z-test on the entire dataset where each row relates to a separate ad and a McNemar test on the aggregated results.
The Guardian did not record any details of any of the advertisers who responded. All associated email addresses, profiles and accounts were deleted in their entirety after the completion of the project.
