In a 2001–2002 correspondence experiment, Bertrand and Mullainathan sent 4,870 fictitious résumés to selected newspaper job ads in Boston and Chicago, randomly assigning White-sounding or African-American-sounding names. Interview callback rates were 9.65% and 6.45%, respectively: a 3.20 percentage-point gap, approximately 50% relative to 6.45%. Randomization supports attributing the sampled contrast to assigned name signals; actual applicants’ racial identities were not manipulated. The outcome was a callback—not a completed interview, job offer, hiring, or earnings. Selected ads, occupational categories, cities, and period limit generalization; no current nationwide rate is established.

Source: https://sendhil.org/wp-content/uploads/2019/08/Publication-52.pdf