Risk Without Reward

Diverse voices and perspectives bring innovation to their organizations, but the people offering them don’t necessarily get credit or support for doing so.

Business scholars have been studying this phenomenon for years and making the case that investing in diverse talent boosts an institution’s bottom line.

A new study makes the same argument about academe: diversity is good for business. But instead of profit, the metric of choice here is research innovation.

The findings are at once promising and sobering. Scholars from underrepresented groups, as a whole, achieve higher rates of scientific novelty, the study says. Yet novel contributions by gender and racial minorities are less likely to be taken up by their peers than are novel contributions by those in the majority.

Contributions by gender and racial minorities are also less likely to result in successful scientific careers.

“We reveal a stratified system where underrepresented groups have to innovate at higher levels to have similar levels of career success,” the paper says. And too often their careers end “prematurely, despite their crucial role in generating novel conceptual discoveries and innovation.”

Which “trailblazers,” the study asks, “has science missed out on as a consequence?”

The precise answer to that question is unknowable, but the possibilities are concerning. To address these inequities and stem the loss of knowledge, the authors urge institutions to study and combat biases in faculty hiring, research evaluation and publication practices.

“The Diversity-Innovation Paradox in Science,” published this week in Proceedings of the National Academy of Sciences, drew on a massive database of about 1.2 million Ph.D. recipients and their dissertations, across fields (including the humanities and social sciences), from the last three decades. The researchers then matched information from that database, ProQuest, with other data from the Web of Science citation database, to see who published at least one paper within five years of obtaining their degree, as proxy for career progression.

Using U.S. Census and other data, the authors tried to predict the Ph.D.s’ gender and race based on their names. The three racial categories considered were white, Asian and underrepresented. Nonbinary gender was difficult to discern due to the male-female method of predicting gender based on first names.

Researchers then determined whether the Ph.D.s were minorities or majorities, or both, within their respective fields, based on those fields' demographics.

As for scientific novelty, the paper defines it three ways: general novelty, or the number of new links between ideas it makes; impactful novelty, or uptake by peers in consequent years (by mentions, not citations); and distal novelty, or linking ideas and combining them in totally new ways, such as metaphorically.

Instead of reading millions of documents, the authors used machine learning, an application of artificial intelligence, and textual analysis of the Ph.D.s' publications to assess their level of novelty.

Scientific development, the paper theorizes, is “the process where concepts are added to the ever-growing ‘constellation’ -- i.e., our accumulating corpus of texts -- in new combinations, or the introduction of new links between scientific concepts.”

Co-lead author Bas Hofstra, a postdoctoral fellow in computational sociology at Stanford University, said that other scientists are taking up the novelty generated by certain groups. Students who are underrepresented by gender in their fields find less adoption of their novel ideas compared to those who are overrepresented, for example.

Women and nonwhite scholars also experience less uptake by their peers. Hofstra attributed this, in part, to reluctance toward adopting distal novelty, which connects ideas in “such new ways that are harder to parse, and more difficult to place and understand for others in science.”

Which ideas are considered scientifically useful or worthy of further research may be in itself a function of position and bias, Hofstra added. He cautioned, however, that “we need more research to specifically unravel each of these mechanisms.”

In the interim, Hofstra, like the study, stressed the “crucial importance” of evaluating and addressing biases in faculty hiring and evaluation.

Monica Cox, chair of engineering education at Ohio State University and author of Demystifying the Engineering Ph.D., said that “it shouldn’t be dangerous to be innovative.”

And yet it often is, she said, as academe remains a fundamentally conservative enterprise. Faculty search committees often prioritize “safe” hires, in terms of candidates’ research portfolios (think connections to confirmed successes and disciplines), institutional pedigrees, advising lineages and even appearance.

In many cases, Cox said, “people are not ready for the innovation in the scholarship or in the representation.”

As a black woman working in an interdisciplinary field, Cox said she could imagine the dynamics in Hofstra's paper affecting a career path like her own. Yet Cox said she’d been successful, despite various challenges, “because I’ve had advocates in the field who have sponsored me even though my work was different.”

Cox said she’d seen other underrepresented scholars overcome barriers by promoting their work on social media -- a kind of “leveling of the playing field” -- or by becoming engaged in technology commercialization.

Beyond these examples of mentorship and self-advocacy, Cox said it’s imperative that institutions have dedicated diversity proponents who “are consciously reviewing applications and making sure that biases don’t come into conversations about somebody’s work,” for instance. And on tenure and promotion, she said, “How do you evaluate someone’s contribution if it does not fit into a box?”

Fostering diversity comes down to “policies and presence,” she advised. Otherwise, “higher ed will go for what is safe, conservative and known.”

Hofstra and his colleagues completed their study prior to the COVID-19 outbreak, but it’s impossible to discuss faculty careers now without thinking about how they’ll be thrown off by virus-related hiring freezes and cuts. Some institutions are in the middle of multiyear diversity-based hiring campaigns, for example.

Even if the gender-racial distribution of hires stays the same as it is now -- which Hofstra questioned -- lower numbers of hires over all may make it more difficult for underrepresented groups “to place their ideas.”

There could be less of a “critical mass of similar peers” that generate, adopt and champion ideas moving forward, he warned.