We study whether and to what extent social media information can reduce (ethnic) discrimination in a two-sided market characterized by asymmetric information. We analyze whether information that breaks with prevailing ethnic stereotypes might induce the uninformed side of the market to update its probabilistic beliefs on a desired, but hidden quality of an ethnic minority applicant. We create eight social media profiles (male and female) and apply for 3,676 vacant room ads for shared housing on a two-sided platform. The profiles are each identical within one gender, except for the different names assigned to them: two profiles of each gender are assigned a Turkish-sounding name, two a German-sounding name. To each application, we randomly assign one of the eight names and whether it contains a link to the corresponding social media profile. When an application includes such a link, the otherwise substantial discrimination against applicants with Turkish-sounding names is not only significantly reduced, but almost eliminated –- hinting at the potential of social media information that breaks with prevailing stereotypes to reduce (statistical) discrimination. (PsycInfo Database Record (c) 2023 APA, all rights reserved)