Auditing race and gender discrimination in online housing markets

While researchers have developed rigorous practices for offline housing audits to enforce the Fair Housing Act, the online world lacks similar practices. In this work we lay out principles for developing an online fairness audit and demonstrate two examples; gender- and race-based discrimination in online housing advertisements, and personalized recommendation ordering. We employ a controlled sock-puppet audit technique to build online profiles associated with a specific demographic profile or intersection of profiles, and describe the requirements to train and verify profiles of other demographics. We also describe the process used to collect data for the two audits using these sock-puppet profiles. In the first we collect ads served on several sites in order to determine whether the number of housing- related ads served is dependent on the perceived race or gender of the profile. The second compares the ordering of personalized recommendations on major housing and real-estate sites. Using statistical tests, we examine whether the results seen in these areas exhibit indirect discrimination: whether there is correlation between the content served and users’ protected features, even if the system does not know or use these features explicitly. We believe this framework provides a compelling foundation for further exploration of housing fairness online.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2019

Author dc:creator

• Asplund, Joshua T.

Contributors dc:contributor

• Karahalios, Karrie

• computational audit
• algorithmic audit
• discrimination
• fair housing
• advertising

Statement dc:rights

• Copyright 2019 Joshua Asplund

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/104905

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/104905