Gender Systematics in the NRAO Proposal Review System

TL;DR

This study examines gender-based biases in NRAO proposal reviews, showing a shift from male favoritism to a more balanced or female-favoring trend after intervention measures, and proposes dual-anonymous review to further reduce bias.

Contribution

It provides empirical evidence of gender bias reversal in NRAO proposal reviews and suggests implementing dual-anonymous review to mitigate bias.

Findings

Gender bias favored males before 17A semester.

Bias shifted towards females after 17A semester.

Dual-anonymous review proposed to reduce bias.

Abstract

Several recent investigations indicate the existence of gender-related systematic trends in the peer review of proposals for observations on astronomical facilities. This includes the National Radio Astronomy Observatory (NRAO) where there is evidence of a gender imbalance in the rank of proposals with male principal investigators (PIs) favored over female PIs. Since semester 2017A (17A), the NRAO has taken the following steps: (1) inform science review panels (SRPs) and the telescope time allocation committee (TAC) about the gender imbalance; and (2) increase the female representation on SRPs and the TAC to reflect the community demographics. Here we analyze SRP normalized rank-ordered scores, or linear ranks, by PI gender for NRAO observing proposals from semesters 12A-21A. We use bootstrap resampling to generate modeled distributions and the Anderson-Darling (AD) test to evaluate the probability that the linear rank distributions for male and female PIs are drawn from the same parent sample. We find that between semesters 12A-17A that male PIs are favored over female PIs (AD p-value 0.0084), whereas between semesters 17B-21A female PIs are favored over male PIs, but at a lower significance (AD p-value 0.11). Therefore the gender imbalance is currently being ameliorated, but this imbalance may have been reversed. Regardless, we plan to adopt a dual-anonymous approach to proposal review to reduce the possibility of bias to occur.