The Trouble With Homogeneous Teams

Many companies today understandably focus on workplace diversity — issues such as how to increase diversity, how to foster sensitivity to it, and how to manage a diverse workforce. But, according to MIT Sloan School professor Evan Apfelbaum, managers should also be cognizant of another, related topic: the problems associated with homogeneity. Recent research, including Apfelbaum’s own, has found, for example, that racially homogeneous groups are less rigorous in their decision-making — and make more mistakes — than diverse ones.

Apfelbaum, the W. Maurice Young (1961) Career Development Professor of Management and an associate professor of work and organization studies at the MIT Sloan School, spoke with MIT Sloan Management Review editorial director Martha E. Mangelsdorf. What follows is a condensed and edited version of their conversation.

MIT Sloan Management Review: You’re an expert in research on diversity and how it affects group decision-making. And one thing you and others have found is that diverse groups often do better in decision-making than more homogeneous ones. Can you tell us a bit about some of the important studies in that area and what they found?

Apfelbaum: Sure. A good way to think about it is that diverse groups have the potential to do better than homogeneous ones. In reality, there are a number of examples and reasons why that often doesn’t happen. But I do think there’s a unique advantage to diverse groups in certain areas.

I’ll start off by talking about cooperative decision-making scenarios, where people are trying to work together to come to some best solution. Early work from several decades ago provided the first evidence that diverse groups yielded more creative solutions, and that spurred much of the more recent research in that area.

One paper that was particularly important and useful took place in a legal setting, with jurors. In that study, a researcher now at Tufts University got access to a real jury pool and randomly assigned jurors to deliberate in six-person, all-white or racially diverse juries. The groups all considered the same fictitious case, and their deliberations were recorded on video.

In general, the diverse juries were far more rigorous in how they approached their decisions. The racially diverse juries spent a longer time deliberating. They considered a wider range of perspectives and angles with respect to the case — different things that could have happened or might have been important. And they made fewer factually inaccurate statements in their discussions.

It wasn’t the case that diverse juries were outperforming homogeneous ones primarily because, say, the black jurors were adding new information that wasn’t there in the all-white juries. It was actually white jurors on diverse juries whose behavior showed the most dramatic change. This suggested that it was something about being in the presence of a racially diverse environment that changed how people thought and discussed issues.

Subsequent research has looked at student groups working on projects, and this work has shown similar effects. For example, one project demonstrated that when you tell students that they’re going to be having a discussion regarding a written article, they prepare more rigorously if they know that they’re going to have to debate in a more diverse group.

In that case, it might have been because the article had some race-related component to it, and the students wanted to make sure that they had a well-rehearsed way to think about that. The study also showed that when researchers asked students to write essays after the fact to reflect on what they took away from the discussion, these essays typically were more complex and of higher quality when the discussion took place in a racially diverse group.

Those are some important effects of diversity. Now, let’s take a more traditional business setting, and one that’s competitive, not cooperative. In one paper that I coauthored, we asked: What happens if you put diverse versus homogeneous groups together in a naturally occurring competitive scenario? To examine this case, we looked at trading markets.

We randomly assigned people to either racially diverse or homogeneous groups in two different studies, one in Asia and one in North America. People were brought to the lab, and we formed mini trading markets of six people. Think of it as a six-person competitive group. The participants were given real money, and there were several rounds of trading where the groups were networked through computers. The participants were making decisions about whether they would like to buy and sell assets, and their goal was to end up with the most money at the end.

The only difference between the two groups was that at the very beginning of the study, people saw who would be in their market based on who was sitting with them in the waiting room. And we randomly assigned half of these groups to be homogeneous. The beauty of this is that it wasn’t just an all-white sort of homogeneity, because we were able to do this experiment in Asia in a way where the dominant culture identity was not white. So half of the groups see a singular ethnicity, and half see a more diverse one.

Then the group members are separated and begin trading. A couple of interesting results emerge here. The first thing is that there’s a difference in accuracy, in how closely people are pricing assets to their actual value. In the homogeneous groups, there was more inaccuracy and mispricing; there was a tendency to spend more for things than they were actually worth.

The other thing that was interesting is that these mistakes were, in a sense, more contagious in homogeneous than diverse groups. That is, not only were people in homogeneous groups more likely to make pricing errors, but other people in those groups were more likely to copy those errors. People in homogeneous groups were more likely to assume that other people in the group knew what they were doing.

In diverse groups, people were less likely to trust the wisdom of other people’s purchasing choices. And the result of these two different dynamics that played out is that pricing bubbles, which are very problematic in financial markets, and for society at large, were more likely in this experimental context to form in homogeneous groups. The reason for this is that people in homogeneous groups were more likely to copy another person’s mistake — presumably assuming that the mistake had some value that they just didn’t understand. In homogeneous groups, there was this escalating effect where people would copy poor decisions.

So we were able to see very different trajectories even within the lab, and in very similar ways in two very different cultures, suggesting that there is something fundamental about working with similar versus different others that affects individuals’ decision-making. Again, this is a competitive context: People were really motivated to try to eke out as much money as they could because, at the end of the experiment, they kept the money they had made. That suggests that people in the homogeneous groups were trying to make the right decisions — but something about the group context constrained their ability to do so.

One of the ideas we had is that maybe this is just something basic about conformity. So we essentially ran a variation of what is one of the most famous social psychological experiments ever to be run in the domain of groups: psychologist Solomon Asch’s conformity paradigm from the 1950s. In that famous study, participants would sit at a table with people they thought were other study participants, and they would simply look at a picture of three lines of three different lengths.

The study participants believed that the other people at the table were also participants, but the other people were actually working for the person running the experiment. And the experimenter said, for example, “Tell me which line is the longest.” What the participants would experience is that they would hear other people answer before them and all say that the third line was the longest, when it was clearly evident that actually it was the second line that was the longest. It was such an obvious answer, but the experimenters in this case were looking at how likely participants are to yield to majority opinion, even when they know that it’s the wrong answer. In other words, how much can social pressure affect us?

Asch’s research found that people yield to what they know to be the wrong answer around roughly 30% of the time — which is a pretty large frequency. We wondered if diversity would change people’s susceptibility to this bias.

We replicated the same paradigm with a few changes; instead of lines, we used a task involving fictitious college applications, where we could establish that one candidate was a clearly stronger applicant for admission than another. What we found is that in all-white groups, the rate of conformity to the clearly wrong applicant was about 30% — which was similar to the classic research on conformity. In diverse groups, however, the frequency with which people would yield to what they know to be the wrong answer dropped significantly, to 20% in some experiments, and even lower in others.

What’s interesting about these studies is that we didn’t allow people to talk with the other people in the room. What we were looking at is not the effect of having a discussion or being persuaded by arguments. We were asking: Does simply sitting down in a room and seeing the demographic makeup of the people at the table affect people’s propensity to conform to others’ decisions?

And the answer was that it does — and that people were less likely to conform in diverse groups. Similar to our suspicions in the stock-pricing experiment, there’s almost this benefit-of-the-doubt effect that happens in homogeneous groups that we don’t see in diverse groups; people in homogeneous groups are more likely to assume that the other people in the group must know something or have picked up on something that they didn’t. In diverse groups, people are more likely to not rely on those types of assumptions and come to an independent assessment of what they think to be the case.

I wonder if homogeneous groups make people feel more comfortable, and they then work less hard cognitively. Do researchers know what the mechanisms are that cause these differences in behavior?

Apfelbaum: What I say is, diversity is not better or worse — it’s just harder. It’s harder socially, it’s harder cognitively, and it makes us work. And I think that’s a useful framework to think about why diversity can be both advantageous and complicated in the workplace and in decision-making groups.

When we find ourselves in social events, there’s a natural inclination to gravitate toward people who are similar to us. It’s easier. It’s easier to find common ground with people who have similar backgrounds to us — whether it’s in terms of culture, organizational expertise, language, or school affiliation. That’s natural. It provides us with a sense of belonging and it’s easier — and I think that’s OK. Diversity is harder for the same kind of reasons. It doesn’t allow us to rest on our laurels, and we are less concerned, in some sense, with retaining our membership in diverse groups.

I think people just end up being more independent and objective in diverse groups. And that can go well in the scenarios I just talked about, but it is also, I believe, at the root of other research that has shown that diversity can breed conflict and mistrust. Some research, for example, has shown that even a normal level of team conflict is more quickly perceived as a really serious type of conflict in the eyes of managers when the group is racially diverse as compared to homogeneous. One very recent study presented participants with an exchange between members of a team. And the researchers just changed a very, very small component of that exchange — which was whether people believed that the people involved in this team exchange, which was designed to be medium-level debate and back and forth — were a diverse or homogeneous group.

What the researchers found is that people in the role of managers were much more likely to think that the diverse teams’ level of conflict was higher, even though it was exactly the same exchange. And in turn, the managers were less likely to suggest that greater resources be provided to the diverse teams to assist them with completing future projects. In some sense, managers were saying, “We can’t invest; this is an irreparable form of conflict.” So the level of conflict seems to be perceived to be artificially higher in diverse groups than it is in homogeneous groups.

I think that when it comes to gender, race, and ethnicity, these are issues in our society that are fraught and laced with mistrust and uncertainty, so there’s a lower threshold for people to find evidence that is consistent with that and either disengage from their groups, accuse others, or devolve into unproductive forms of conflict in groups.

Interesting. It sounds like in most of the studies you’re discussing, the diversity is racial. And are there similar findings with different types of diversity, such as gender diversity?

Apfelbaum: I would say race and culture are two of the most frequent ones that have been explored. Gender has been explored, and there’s been some similar experiments there. I think that research now is really only just beginning to look at, for example, how race and gender may play out differently. There is also some research out there that has looked at cognitive diversity — for example, diversity in the way people think about problems.

You mentioned research about the cost of diversity and the conflicts that can happen, and how they can quickly get unproductive. Tell me a little about that research.

Apfelbaum: There is a good amount of research that’s happened in the past few decades that has found in real work teams that people in diverse teams report higher degrees of conflict. They like it less. They’re less comfortable there. And there are a number of different studies that have demonstrated more interpersonal conflict in diverse teams.

Let me tell you about another finding that I think is pretty interesting. In one paper, researchers looked at a decision-making task that was cooperative. Participants in the group had to put together disparate pieces of information — clues — to make a single recommendation about the correct suspect to arrest.

What the researchers found in this task, as has been shown in previous research, is that the diverse groups tended to consider more perspectives, and ultimately were more likely than people in the homogeneous groups to narrow in on the correct suspect.

But another result that came out that was very, very interesting was that in this study the researchers also asked: How confident are you that you have identified the correct suspect? And though the diverse groups were factually more accurate than the homogeneous groups, it was actually the homogeneous groups that were more confident in their results — the exact opposite of what you would expect.

Now you could think of this as a challenge to diversity or a limitation to homogeneity. It suggests that diverse groups with the same results as homogeneous groups can come out of a meeting stating that they’re less confident that they have achieved the correct objective or have landed on a workable solution than a homogeneous group will be. And what we know about confidence in organizational settings is that it is reinforced. If you’re managing two groups and one group comes to you and says, “We are 95% sure this is going to be on time, under budget, and workable,” and the other group comes to you and says, “We’re 75% sure,” in the real world, nine times out of 10 it’s the group that says that they’re 95% sure who is going to get the opportunity to run with their project and try to deliver.

And what we’ve seen from this data is that you’re more likely to hear that 95% story from the homogeneous group — but it’s not because they are more likely to deliver better results. The homogeneous groups may just be less accurate. In homogeneous groups, there seems to be this inflated sense of confidence, in part because of the phenomena unearthed in the research that I’ve talked about earlier. Those groups may not be considering all the perspectives. And there is more of a tendency to narrowly see the issues in ways consistent with other people’s views and perhaps less comfort to disagree with others.

So the diverse groups are actually mitigating overconfidence bias in a way?

Apfelbaum: Yes.

What should managers take away from this research? What advice would you give?

Apfelbaum: The takeaway for me is that the diversity needs to be carefully managed. Managers need to mitigate the concerns about people not feeling comfortable in order to harness what can be some of these real distinct advantages of diversity.

And what about these findings about teams? Should executives be thinking, if we have a really complex decision to make, the decision-making group should be more diverse?

Apfelbaum: Certainly, when you have to make a large-level organizational change or you’re making a big decision, people are often involved from many different functional groups, so everyone can see each other’s blind spots to some degree.

I think that should sort of be the status quo in organizations. And I think that with a really inclusive culture, the lack of ease and comfort that people typically associate with diverse groups can be normalized.

If you think back to a lot of the data that we’ve just gone over, at least a good portion of it says that, well, in objective ways, homogeneity is the thing that’s producing the strange results. Think back to the overconfidence results. Think back to the amount of inaccuracy.

But how many leaders in organizations do you know who have thought, “Wow, what can we do about the problems of homogeneity? Where are the most homogeneous teams that we have in our organization, and what can we do to make sure that they are thinking really carefully and there’s some productive conflict?” I don’t know of a single program anywhere in the world that is focusing on the potential blind spots of homogeneous teams. And I think that’s just not the narrative, because, in many industries, homogeneous teams are normal in terms of their frequency.

But even if homogeneous teams are normal in the sense that they’re common, there’s reason to question how normal they are in these other ways. Instead of just looking at the management of diverse groups as a problem to be solved, it’s useful to flip that for a second and think of it from the other side: What can we do about the problematic aspects of homogeneous teams?