The scientific novelty that underrepresented genders and races introduce gets devalued and discounted. We think it has to do with social structures and the dominant role of majority groups—in most cases, white men—in how scientific problems are framed and understood. Science is a communal effort, and it conforms to the pattern of its social structure.
It’s also possible that the fresh perspectives that women and nonwhite scholars bring are atypical and can sometimes be hard to grasp, so they get devalued by the majority.
How were you able to generate this evidence?
We analyzed a dataset of nearly all completed PhD theses in the United States between 1977 to 2015. That’s nearly 1.2 million dissertations. We inferred the students’ gender and race by matching this data to U.S. Census and Social Security Administration data. We also looked at how many were still publishing research five years later. For those who had graduated a PhD as a primary adviser in that time, we assumed they were tenured faculty at a respectable institution.
The complex part was identifying who introduced novel conceptual links and, of those, which ones proved over time to be influential in their field. We opt for ‘novelty’ over ‘innovation’ in the paper because there is an important difference between the two.
What’s the distinction between them?
Innovation is defined as something that is new and that people use often—but what is used often is perhaps decided by the majority, so it’s not necessarily innovative. If you think about Google Scholar citations, people will often evaluate researchers based on how often their work gets cited by others. But citations can underrepresent certain disciplines or can have a perpetuating effect in that the strong get stronger.
People who connect ideas in new ways introduce novelty, which is not necessarily innovation. Novelty is basically a prerequisite for innovation.
So, how did you identify novelty in the researchers’ work?
Basically, we looked at topics listed in dissertation abstracts and identified what appear to be substantively important concepts while ruling out a lot of noise. We wound up with concepts like ‘HIV’ or ‘social capital’ or the name of a chemical compound.
Once we had all the substantively important concepts from the texts, we started to identify the students who were able to connect them in novel ways. But we also wanted to measure which novel concepts had impact in their respective fields. To do that, we studied how often they were adopted in subsequent PhD dissertations.
For example, Lilian Bruch, a pioneer in HIV research, first linked monkeys to the virus in her 1987 doctoral thesis. Forty-seven PhD dissertations later adopted her conceptual link. That might not sound like a lot, but uptake of her novelty was nearly 70 times higher than the average.
What can researchers and leaders in academia do in response to your findings?
The first step to addressing bias and discrimination in academia is knowing when and where it occurs. The fact that we now have the data showing this is really exciting. So, awareness is key. We also need to be vigilant — to continuously evaluate and address biases in faculty hiring, grant writing and research evaluation.