Sex-Disaggregated Systematics in Canadian Time Allocation Committee Telescope Proposal Reviews

TL;DR

This study reveals significant gender bias in Canadian telescope proposal reviews, with women’s proposals rated worse than men’s, and suggests that implicit bias may influence peer review outcomes in astronomy.

Contribution

It provides the first systematic analysis of sex-based bias in Canadian telescope proposal reviews using statistical methods and proposes measures to mitigate this bias.

Findings

Women’s proposals rated significantly worse than men’s.

PI sex is the only significant predictor of proposal scores.

Bias persists even after controlling for other factors.

Abstract

Recent studies have shown that the proposal peer review processes employed by a variety of organizations to allocate astronomical telescope time produce outcomes that are systematically biased depending on whether proposal's principal investigator (PI) is a man or a woman. Using Canada-France-Hawaii Telescope (CFHT) and Gemini Observatory proposal statistics from Canada over 10 recent proposal cycles, we assess whether or not the mean proposal scores assigned by the National Research Council's (NRC's) Canadian Time Allocation Committee (CanTAC) also correlate significantly with PI sex. Classical t-tests, bootstrap and jackknife replications show that proposals submitted by women were rated significantly worse than those submitted by men. We subdivide the data in order to investigate sex-disaggregated statistics in relation to PI career stage (faculty vs. non-faculty), telescope requested, scientific review panel, observing semester, and the PhD year of faculty PIs. Consistent with the bivariate results, a multivariate regression analysis controlling for other covariates confirmed that PI sex is the only significant predictor of proposal rating scores for the sample as a whole, although differences emerge for proposals submitted by faculty and non-faculty PIs. While further research is needed to explain our results, it is possible that implicit social cognition is at work. NRC and CanTAC have taken steps to mitigate this possibility by altering proposal author lists in order to conceal the PI's identity among co-investigators. We recommend that the impact of this measure on mitigating bias in future observing semesters be quantitatively assessed using statistical techniques such as those employed here.