How stereotypes impair women’s careers in science

Hiring “on ability” leaves open what information the decision maker actually receives. Reuben and colleagues separate choices made without performance data, after candidates’ self-assessments, and with verified past performance. Bias can operate through whose predictions are believed and whose confidence counts as evidence, without an explicit statement excluding women.

Distinguishing performance from its presentation

Candidates completed arithmetic tasks, and employers selected someone for a subsequent task. Information conditions included no performance information, candidates’ forecasts of their own results, and actual previous scores. Because performance could be observed, a hiring preference did not have to stand in for evidence of ability.

Without performance information, choices favored men. Hearing self-predictions did not automatically correct the imbalance: men’s greater overstatement could become an advantage when employers failed to discount it sufficiently. Verified past performance reduced bias without eliminating it. The paper also examines associations between implicit gender stereotypes, expectations and the updating of beliefs.

The institutional question is broader than whether women should learn to advertise themselves more forcefully. Treating confidence as a dependable signal of potential can reward a form of presentation rather than the relevant work. Asking underestimated applicants to promote themselves harder may preserve the standard while transferring the cost of adaptation to them.

A laboratory choice is not an entire scientific career

This is a controlled hiring experiment, not a longitudinal study of recruitment to actual research jobs. Arithmetic does not exhaust the abilities needed in science, and the results do not establish general intelligence differences. The information conditions help identify mechanisms within this setting; they do not measure the size of discrimination in every occupation.

The experiment involving science faculty and identical applications offers a comparison between verified scores and résumé evaluation. Acker’s account of gendered organizations widens the question: even where assessors intend to be fair, do hiring procedures reward particular forms of self-presentation and then describe that reward as neutral selection on merit?