Affiliate Latanya Sweeney Examines Racial Bias in Online Ads

February 8, 2013
sweeney6crop.jpg

While it is commonly known that advertisements on search engines such as Google are customized depending on the user and activity, it would appear that some aspects of that customization might be questionable. Latanya Sweeney, Professor of Government and Technology in Residence and founding director of the IQSS Data Privacy Lab, has found in her recent research with racially-associated names that web searches for traditionally black-associated names are more likely to produce online ads suggestive of criminal records.

The Boston Globe has reported this week on Sweeney's research, in which names that are predominantly identified with African-Americans were run through Google and Reuters web searches; a set of predominantly "white-identifying" names were also run through the same search engines, and the web advertisements that each set produced were compared. Sweeney found that the online ads that accompanied black-identifying name searches were significantly more likely to suggest arrest or criminal records. The reason for these results might be attributable to a number of factors. Along with the Globe, the MIT Technology Review's report on the study examines how Google Adsense might produce such search results.

Latanya Sweeney's paper on the study is available for download online: "Discrimination in Online Ad Delivery." The data used in the study can be found online at FOREVERdata.