When Diversity Delivers—and When It Doesn’t

IEEE Spectrum’s Q&A with Scott Page, author of The Diversity Bonus

4 min read

photo of diverse group meeting
Photo: Getty Images

photo of diverse group meetingPhoto: Getty Images

One of the responses that often comes up when diversity is discussed is the assertion that for best results the focus of employers should simply be on finding the best possible individuals, regardless of any other consideration one way or the other. In his new book, The Diversity Bonus: How Great Teams Pay Off in the Knowledge Economy (Princeton University Press), Scott Page investigates under what circumstances this assertion is valid, and when it can actually result in worse performance. IEEE Spectrum senior editor Stephen Cass talked to Page—the Leonid Hurwicz Collegiate Professor of complex systems, political science, and economics at the University of Michigan—about his work:

Stephen Cass: What is a “diversity bonus”?

Photo of Scott PagePhoto: Princeton University Press

Scott Page: The central premise of the book is if you have routine work, like chopping wood or filing papers or typing up forms, then the total product of a team is the sum of the total production of each individual. It’s additive. There’s no diversity bonus. But once you have a problem that’s difficult, this notion of individual ability becomes problematic. The idea that the value of the team is the sum of the ability of its individual members doesn’t make sense anymore. This is because if two people are the same, they have the exact same set of abilities that they are bringing to bear on the problem. If you go to an engineering school like Caltech or MIT, what they do is teach you a whole bunch of different problem-solving tools, right? So if you know the same tools as me, the two of us together really aren’t that much better than we are individually. But if you know different tools than me, then there’s a diversity bonus. By diversity, I mean different ways of thinking, different heuristics, different perspectives.

S.C.: You’re talking about cognitive diversity. How does this relate to what people usually mean when they talk about diversity, which is identity diversity, whether that identity is gender, ethnicity, orientation, and so on?

S.P.: I’m not trying to make an ideological argument. The logic that cognitive diversity is beneficial is a theoretical argument that an engineer would write down. But then there’s the empirical question: Where does this kind of diversity come from? It comes from different training, different experiences. Most of these things kind of mingle. I think it absolutely depends. But it’s really hard to argue that cognitive diversity doesn’t come from identity in some respects.

S.C.: Part of the logic in your book leads to the counterintuitive idea that the best team might not have the best individuals on it.

S.P.: If you’re tackling a hard problem with people who have worked on the same projects, trained at the same schools, who all come from the same identity group, then your best engineers may think of the same 10 approaches, while somebody else may think of only 5 things, but they may be different. And the ability of the group is how many ideas do you get total. If ideas are being generated randomly without correlation, then it would be true that the best group would have the best individuals: those who come up with the highest number of ideas. But they’re not random. It’s that simple. I think the real strength of the book is that it’s not a convoluted argument. It’s so straightforward, and especially to engineers.

The Diversity Idea

graphic illustration of the diversity ideaHere, each cognitive tool or skill is labeled with a letter. Alice is the most skilled person, with five skills, while Bob and Charlie have only four and three skills, respectively. But because of the way their skills combine, the best two-person team is Bob and Charlie, not Alice and Bob or Alice and Charlie.Source: Scott Page

And that’s not even allowing for synergy, where a new idea sparks other team members to go in directions they hadn’t considered before. With real groups, there’s this sort of extra bonus that comes from accessing what Stuart Kauffman calls the “adjacent possible.”

S.C.: What kind of empirical evidence have you seen for a diversity bonus?

S.P.: Over 10 years ago, I started trying to make sense of how diversity made things better and worse in different settings. Because diversity doesn’t always make things better. But there wasn’t a lot of data at that time. Now there’s huge data in ways I never would have anticipated. And here’s just one example that blows me away. A study looked at 22 million research papers, holding everything else constant and looking at the effect of having authors teaching at different schools. The odds of getting a paper with 100 citations goes up 10 percent in the social sciences and 8 percent in the natural sciences just if you are at different schools.

S.C.: A lot of organizations have introduced various diversity initiatives, but how can they leverage those to deliver the maximum diversity bonus?

S.P.: The first thing to recognize is that the social justice angle’s still here. We have to be aware that we live in a society where there’s still a lot of issues going on. Once you’ve got that part, then you want to think what types of diversity might be relevant. A lot of places when they are hiring will score candidates, and then they hire the highest scoring people. That’s probably not right. Instead, what you want to do is think, “Who has tools or understanding that we don’t have?”

For example, if you’re an academic and you’re hiring postdocs, why would you hire postdocs from your advisor? Or you go to the International Monetary Fund and you see economists populating their teams with other economists who went to the same graduate schools. If you hire people from the same ethnic group, who went to the same schools, worked at the same places—you’re just kind of shooting for a B-plus there. Nothing great’s going to happen. So the short answer is to think of people as toolboxes, as opposed to numbers. And then ask yourself: “What does the full set of tools look like that I’m putting together to solve a problem?”

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions