The Fair UI databaseArticles on UI & bias
A collection of articles on the topic of UI and bias, gathered in connection with our work on the Fair UI project.
“Bias in Online Freelance Marketplaces: Evidence from TaskRabbit and Fiverr” — Anikó Hannák et al., 2017 (Conference paper)
TaskRabbit workers perceived to be female received 10% fewer reviews than their male counterparts with the same amount of experience. TaskRabbit workers perceived to be Black, especially men, received lower feedback scores than other workers with similar characteristics. Fiverr workers perceived to be Black men recieve 32% fewer reviews than other men. They also receive lower feedback scores. The only other group receiving worse scores are those without any profile picture. Reviews for Black female workers on Fiverr use less positive adjectives and reviews for Black workers, in general, include more negative adjectives. Workers perceived to be Asian, especially men, receive significantly higher rating scores than other workers.
“Designing Informative Rating Systems for Online Platforms: Evidence from Two Experiments” — Garg & Johari, 2018
The authors’ results show that the adjectives shown in a rating system can have an effect on rating distribution. The rating distribution obtained can be substantially more dispersed than under the “standard” star rating scale.
“Reputation Inflation” — Filippas & Golden, 2018
The authors demonstrate, through the analysis of data from a large online labour marketplace, supported by data from another 4 online marketplaces, that ratings become increasingly positively inflated over time.