Researchers weren’t looking for this result when they first encountered it. They were just trying to see if groups have an intelligence of their own that’s analogous to human intelligence. It was a great research idea—this was one of those obvious, important questions that no one had answered.

Human intelligence as measured by an IQ test is more remarkable than most people realize. An IQ test is actually 10 subtests that measure widely divergent abilities, and there’s no particular reason to expect that someone who’s good at, say, arithmetic would also have a large vocabulary, for example. You might reasonably expect the reverse—that people who spend all their time reading would build big vocabularies but would have no time for math. Yet the amazing revelation of the IQ test, first observed in 1904, is that scores on all the subtests tend to move together; people who do well on one tend to do well on all. The researchers figured there must be some factor that accounts for this result. They could have called this factor any number of things, but they decided to call it general intelligence.

Now, back to the big question: Do groups have that same kind of intelligence, which accounts for performance on a range of different tasks? The answer is yes, as researchers from MIT, Carnegie Mellon, and Union College discovered in experiments involving some 200 groups. That’s an important finding on its own. Then, as the researchers looked deeper, they found a surprise. A group’s collective intelligence was affected only slightly by the average IQ of its members, and it had nothing at all to do with other factors that many people assume are important—the group’s motivation, cohesion, or satisfaction. But it was affected considerably by the members’ social sensitivity, their ability to sense the thoughts and feelings of others. And because women on average score much higher than men on social sensitivity, groups with more women showed higher collective intelligence—the greater the proportion of women, the better.

That striking result got the researchers wondering about something else. They had measured group members’ social sensitivity by giving them a widely used assessment called the Reading the Mind in the Eyes (RME) test. You’re shown photos of people’s eye regions, and you guess what they’re thinking or feeling. But suppose a group’s members couldn’t see each other at all and instead interacted only through typed online messages. Would social sensitivity, as measured solely through the visually based RME, still increase an online group’s collective intelligence?

The answer again is yes. Online groups with the highest average RME scores, which meant groups with the highest proportion of women, had the highest collective intelligence—and the effect was just as strong as when the groups worked face-to-face. The RME test thus captures something larger than just eye-reading; it measures a general ability to sense what’s happening in other people’s minds. And again, the more women in the group, the smarter it was.

Together, these findings are extraordinarily powerful and pose deep questions for how groups are composed in an era when ever more work is done by teams. But now a warning: Researchers have also found a way to destroy the whole effect. Just introduce competition for status. In an experiment at Carnegie Mellon’s Tepper School of Business, some groups elected a leader who remained the same through all the group’s testing, so the hierarchy was stable, with no status competition. Other groups elected a leader but were told he or she could be changed later, creating an unstable hierarchy with status competition.

Groups in the first category recreated earlier results: More women equaled smarter groups, the more women the better. But in the second category, with the possibility of status competition, gender had zero effect. All-women, all-men, a mix, it didn’t matter—group intelligence was the same. And it was way, way below the intelligence of an all-female group with stable hierarchy. Competing for status poisoned a group’s effectiveness regardless of gender.

These results are not widely known, but they should be and will be. As team success increasingly drives business success, they’re too important to remain obscure.