Getik, DemidDemidGetikIslam, MarcoMarcoIslamSamahita, MargaretMargaretSamahita2021-05-312021-05-312021 the A2021-05202112http://hdl.handle.net/10197/12226We study the origins of support for gender-related affirmative action (AA) in two pre-registered online experiments (N = 1, 700). Participants act as employers who decide whether to use AA in hiring job candidates. We implement three treatments to disentangle the preference for AA stemming from i) perceived gender differences in productivity, ii) beliefs about AA effects on productivity, or iii) other non-material motives. To test i), we provide information to employers that there is no gender gap in productivity. To test ii), we inform the candidates about the hiring rule ex-ante, allowing us to observe how AA is expected to affect productivity. To test iii), we remove the payment to the employers based on the chosen candidates’ productivity, thus making AA cheaper. We do not find significant differences in AA support across treatments, despite successfully altering beliefs about expected productivity differences. Our results suggest that AA choice reflects a more intrinsic and inelastic preference for advancing female candidates.enAffirmative actionBeliefsGenderInformationInstitutionC91D02D83J38J71Working Paper147https://creativecommons.org/licenses/by-nc-nd/3.0/ie/