Although a seemingly simple concept, the issue of polarization has long frustrated political scientists. A superficial examination of the American political scene suggests an intensely polarized electorate, divided along partisan and ideological lines. Watching cable news, we see competing camps that have few points of agreement, with anger the dominant emotion. Yet a dive into public opinion on questions of policy tells a different story.

In 2004, Stanford University political science professor Morris Fiorina and his colleagues persuasively argued that Americans are not bitterly divided on the most contentious policy questions, that in fact Americans lack true ideological convictions. Their argument today remains as sound as ever.

The claim that most of us have a coherent bundle of ideological constraints that inform our policy preferences and voting choices has little empirical support. The number of consistent liberals and conservatives in the electorate remains very small. The Bible-thumping, pro-war, free-market purist is a rare creature. So is the gun-grabbing, abortion-loving, socialist atheist. Perfect conservative and liberal stereotypes are hard to find in the real world.

Especially on economic issues, Americans exhibit a remarkable consensus, for better or for worse. Across the partisan divide, most people endorse a form of welfare capitalism—we just disagree on the minutia of tax policy, regulation, and the strength of the social safety net.

This claim, that polarization is not occurring, seems at odds with our everyday experiences. People are angry about politics, and strongly dislike their political opponents even when they substantively agree with them on many policy questions. After countless empirical studies and debates, scholars are inching their way towards an explanation for these contradictory trends.

Part of the apparent paradox may be explained by the nature of partisanship. Rather than the result of a rational analysis of various policy positions, it may be better to think of party allegiance as an element of personal identity. This is a point that many conservatives who decry “identity politics” often miss. Party politics itself can be a form of identity politics, even if our party identifications are downstream from other elements of identity, such as race, religion, and class.

Yet this still leaves an unresolved puzzle. We know that Republicans and Democrats strongly dislike each other. But what pundits don’t always like to talk about is how much partisans themselves increasingly disapprove of their own parties. Republicans are particularly unlikely to report positive emotions towards the GOP. The partisan media would like to interpret these findings as evidence that people are frustrated with their parties’ lack of ideological convictions—that the GOP, for example, has become a bunch of unprincipled “RINOS” (Republicans in Name Only), and thus its conservative voters respond with frustration. Such an explanation, however, is at odds with the finding that few Republican voters are interested in principled conservatism at all.

Eric Groenendyk of the University of Memphis may have found a solution to this puzzle. In a recent article, Groenendyk offers a new explanation for how partisans’ antipathy toward the opposing party can coexist with growing frustration towards their own. In his earlier book Competing Motives in the Partisan Mind, Groenendyk developed the “dual motivations” theory of party identification. In short, partisans have different motives for identifying with their parties, and these sometimes conflict. Because party identification is an important part of personal identity, we want to be good and loyal partisans and we feel good when our team wins. On the other hand, we like to imagine ourselves as rational beings, forming political opinions and loyalties according to our analysis of what is happening in the world.

Ideally, there should never be dissonance between the two. When our party wins elections and enjoys real power, we hope that it delivers on its promises, providing peace, prosperity, and stability. When this fails to occur, however, our two motivations are in conflict, and we can suffer psychological turmoil. From a purely rational perspective, when our party disappoints us, we should reevaluate our support for that party, becoming independent or even joining the other side. If our party identification is a crucial part of our identity, however, this is easier said than done. Party allegiance is not fixed, but it’s also not something most of us abandon easily.

According to Groenendyk, we can resolve the tension between our party identifications and our frustration with our parties by increasing our antipathy toward our parties’ opponents. In other words, we can justify our vote choice if we believe the opposing party is worse. This allows us to acknowledge our disgust with our parties without jumping ship.

This process results in a curious variety of polarization. Few people love their party and think it represents their interests well. Nevertheless, we increasingly hate the other option, and this is enough to keep us in our respective camps. And the more exasperated we become with our parties, the more we demonize the other side.

Using survey data over multiple election cycles, Groenendyk showed that decreasing fondness among partisans for their own parties was associated with growing hostility toward the opposition party. This did not definitively prove that his proposed psychological mechanism explains all the dynamics of polarization. However, it did make a persuasive case that his hypothesis better explained these findings than alternative theories.

Groenendyk suggested that his hypothesis could explain other curious political developments. If out-group hostility is more important to party identification than support for particular policies or ideologies, we may not actually place very many ideological demands on our parties. Defeating our enemies may be more important than advancing specific liberal or conservative agendas. According to Groenendyk: “If partisans’ identities are increasingly anchored to hatred of the outparty than affection for their inparty, electoral dynamics are likely much more fluid than many accounts suggest. Thus, insurgent candidates with questionable ideological credentials (e.g., Donald Trump) may be more appealing than one might expect in the age of ideologically sorted parties.”

It is not obvious how troubling we should find this possibility. These trends long predate the Trump presidency. We may pine for the days when Americans viewed electoral politics as a competition between the greater of two goods, rather than the lesser of two evils, but those days are long gone, and American democracy continues to function reasonably well. Whether this is sustainable in the long term is an open question.

George Hawley (@georgehawleyUA) is an assistant professor of political science at the University of Alabama. His books include Right-Wing Critics of American Conservatism, Making Sense of the Alt-Right, and Demography, Culture, and the Decline of America’s Christian Denominations.