All around us, we see people battling for the benefit of those like themselves—women clamoring for abortion rights in Poland; Black Lives Matter protesters fighting to end police brutality against people of color in the U.S.; LGBT people calling for marriage equality worldwide. But what makes these people care so much about these issues, and not about others connected to their group?

One possible answer is painfully obvious: we speak up when our fellow group members suffer. People tend to think and behave in ways that benefit their group; it’s called solidarity, or in-group favoritism, depending how you feel about it. This is true even when group divisions are almost meaningless. For example, we give preferential treatment to people who like the same paintings we do—or, according to this little in-class experiment, agree with us on whether or not a hot dog is a sandwich. Social categories like gender and race are a lot more meaningful than agreement on sandwich questions. It would make sense if people favored societal changes that benefit their social groups the most.

Certainly, when a social movement picks up an issue, it is almost always the case that the issue disproportionately affects the members of that social movement. The suffragists crusaded for women’s votes, not for veteran disability benefits. But there are many missing combinations of group and issue. Black and Hispanic Americans are much more exposed to air pollution than are whites, but that hasn’t resulted in a Black Lungs Matter movement yet. And there has there not been an LGBT #MeToo wave, even though gay and bisexual Americans are three times more likely to have been sexually assaulted than heterosexuals. Conversely groups don’t only fight for “their own.” Lesbians and Gays Support the Miners was just one colorful example.

So, do groups mainly take on issues when it is in the group’s interest to do so? Or do they do it for other reasons? To answer this question, I invited nearly 500 men and women to participate in an online experiment. Some randomly chosen participants were given a piece of information connecting an issue to their social group. For example, some of the male participants learned this fact: American men are twice as likely as women to die in a car crash. Some female respondents learned that women are 30 percent more likely to be poor. These particular questions were the result of a long quest on my part to find problems that have a real bias toward one gender. But that information had to be news to participants. So, combinations like women and sexual harassment were out.

The results of the experiment were baffling. Learning this piece of information changed nothing about participants’ opinions. They did not feel more concerned about the issue. They did not move it up a single place in their rankings of important social problems. They would not support any more government spending on the issue than before. I repeated the experiment with 700 Black, Latino and white participants. Nothing. I recruited LGBT participants. Nothing. Perhaps the effect was only there for people who truly identified with the group, saw it as a very central part of themselves? It wasn’t. Simply learning that a problem affects your group does not make you care.

Of course, the world is full of social groups caring about issues. How can we square this with the results of the experiment? It could be that the issue must somehow connect to the core identity of the group. This is true for topics like gay marriage or bans on Muslim head coverings. It is also possible that leading figures need to rally the group behind the issue, as Martin Luther King did for civil rights. Perhaps it helps to have a common enemy. Or perhaps people need to feel that caring about the issue is already a norm within the group. We already know that group norms impact political behavior: for example, Black Americans vote Democrat in part because they feel socially pressured by their communities to do so.

Finally, maybe statistics aren’t enough to make people care. We know that stories, because of their emotional impact, are often more powerful than numbers in changing opinions. Perhaps you need to know a fellow group member who was harmed: a woman who was harassed, a Black person who has hurt by police. Political scientists are discovering more and more how political opinions are rooted in emotions and social relationships; they are not rational responses to the world.

I care about the #MeToo movement, and that feels logical, because I am female. But far from all American women agree with me. And in a decade, perhaps my niece will wonder why I didn’t instead focus my energy on the epidemic of depression—a lethal disorder that affects women twice as much as men. In the end, the list of issues that I care about “as a woman” is arbitrary, and a product of many experiences, examples and exchanges in my life.

What about yours?