We found no gender differences in research funding rates
But a clean sheet doesn’t mean a fair system
By Vincent A. Traag (RoRI and the Centre for Science and Technology Studies, Leiden University)
Academia has some persistent gender differences. For instance, in the EU, only a little over 30% of the professors are women. Because securing research funding is vital to climbing the academic ladder, a critical question arises: Are gender disparities in funding rates driving this overall imbalance? Using data from 8 different funders around the globe covering over 70,000 applicants in 43 different funding programmes, we set out to answer that question.
We found that men and women had equal chances of success in being awarded funding. In other words, we found no gender differences in funding rates.
How about gender bias? Much of the literature on this topic discusses gender bias instead of gender differences, but these terms are rarely clearly defined. To avoid any such confusion in this project, we interpreted our results using definitions of gender bias, gender disparities and gender differences that Ludo Waltman and I had previously developed. In this context, we think of gender bias as an unjustified direct causal effect, and gender disparity as an unjustified (indirect) causal effect of gender on funding rates, while gender differences refer to any observed differences in funding rates between genders. But how does our finding of equal funding rates then relate to bias?
In general, gender differences do not imply gender bias, just as correlation does not imply causation. The reverse is also true: the absence of gender differences does not mean there is no gender bias.
Imagine a scenario where a university recruitment process is biased against women, forcing them to pass a much higher bar than men just to get hired. Because the women who survive this filter are, on average, more qualified, you would logically expect them to secure research grants at higher rates than men.
But what if a second gender bias exists against women at the funding stage? That funding bias would suppress women’s success rates, dragging them back down to match the men’s. In this case, the bias in funding may mask the bias in hiring and lead to an apparent gender equality in funding, even if there are still real biases at work at both hiring and funding stages.
When using observational data, we usually cannot discern whether such gender biases exist. Elusive concepts such as the “higher bar” or “quality” are typically not directly observable, and we cannot directly control for them. Attempts at controlling for “quality”, for instance by using proxies such as peer review or scientific impact, run into similar problems, because they may also be subject to gender biases themselves.
In short, we cannot draw strong conclusions about the absence of gender bias based on the absence of gender differences in funding rates that we observe. At the same time, we also cannot draw any firm conclusions about whether any gender bias is present. To claim that a hidden funding bias is actively masking a hiring bias requires making stronger assumptions. Based on our data we cannot judge whether a ‘masking effect’ is happening. Introducing direct corrective measures in funding allocations based on these results would therefore seem unwarranted
Yet, we are still left with our starting paradox: massive gender disparities across academia are very real. Even if funding rates remain equal, only 36% of the applicants in the funding programmes we analysed are women.
How do we get to the root of that disparity? If we concentrate future analyses on identifying where these gender differences emerge, we may be able to develop more effective interventions. For instance, it seems that family formation and gendered childcare duties, especially in early careers, play a large role in gender differences in academia.
By pivoting our attention here, institutions and funders could develop more effective levers for change. Providing robust parental leave, offering part-time appointments, and actively supporting work-life balance will do more than just address gender disparities—it has the power to transform and improve academic work culture for everyone.
This post is based on a new, large-scale study published by the Research on Research Institute (RoRI); you can read it here.