Implicit biases and gender gaps in the assessment of job candidates

How to cite

Radbruch, Jonas; Schiprowski, Amelie. 2021. "Implicit biases and gender gaps in the assessment of job candidates." AEA RCT Registry. AEARCTR-0007072. https://doi.org/10.1257/rct.7072-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Jonas Radbruch* , Amelie Schiprowski

* Primary investigator

Registry ID
AEARCTR-0007072
DOI
https://doi.org/10.1257/rct.7072-1.0
First registered
Last updated
Status
On going
Trial dates
2020-01-01 – 2022-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/7072

Abstract

Researchers have documented gender gaps in many labor market outcomes, such as hiring and admission decisions or the evaluation of performance. This study aims at understanding the role of gender biases among interviewers as potential drivers of gender gaps in hiring outcomes. Moreover, we plan to analyze the impact of a ‘bias awareness’ intervention that aims at reducing the gender gap in the assessment of candidates applying to a large company.

Intervention

The intervention is a workshop module conducted with interviewers in the treatment group. The module targets gender biases regarding the cognitive and math abilities of female candidates. As part of the module, interviewers participate in a Gender-Science Implicit Association Test (Greenwald et al., 2009) and get direct feedback on their score. Moreover, interviewers in the treatment group receive short re-freshers before each interview phase. Due to the pandemic, the entire intervention takes place online.

Experimental design

The study has two goals: (1) estimate the relationship between an interviewer’s implicit gender attitudes and her assessment behavior; (2) evaluate the effect of an intervention targeting gender biases. The intervention includes a workshop module and regular refreshers (as described above). We will evaluate the effect of being assigned to an interviewer in the treatment group (interacted with candidate gender) on the assessment. We plan to evaluate the effect both through simple post-intervention differences and a diff-in-diff framework analyzing within-interviewer (team) variation over time.

Design details

Randomization method
We randomize at the interviewer team level. To balance interviewer team characteristics, we first build pairs of interviewer teams and then randomly assign one team per pair to treatment and the other to the control group, using the ‘nbpMatching’ R package.
Randomization unit
Interviewer Team
Clusters
11
Observations
1000-1500 interviews for the intervention period, depending on the number of applicants and conducted interviews (outside researchers’ control). About 7,000 interviews for the pre-intervention period.
Arms
The treatment group consists of 6 clusters, the control group of 5 clusters.
Primary outcomes
Assessment of candidate’s cognitive abilities (average cognitive score)
Secondary outcomes
Assessment of candidate’s non-cognitive skills (average non-cognitive score) Recommendation to hire candidate (0/1)

Cited by

No citation data available for this registration yet.

Keywords