What does a strong liberal in Clay County, Fla., have in common with a strong conservative in Denver County, Colo.? Not much, you might guess, considering how politically polarized our nation is.

But in fact, these two sometimes hold similar positions on political issues. For example, the average “strong liberal” in Clay County is likely to be slightly against using enhanced interrogation techniques — as is the average “strong conservative” in Denver County.

Our research has uncovered several examples of this discontinuity, in which individuals who define themselves on opposite ends of the political spectrum — but live in different parts of the United States — hold the same policy stances.

The converse is also true. Individuals who identify with the same political label — but live in different parts of the country — commonly have different political stances. For instance, self-declared moderates of Santa Clara County, Calif., are strongly in favor of Obamacare, while moderates in Perry County, Ky. strongly oppose it.

Political identity does not mean the same thing from place to place. We report this finding, based on the results of two studies, in a recent paper.

In the first study, we used data from the 2012 American National Election Study (ANES). For each survey respondent, we looked at self-reported ideology on a scale ranging from “strong liberal” to “strong conservative” and positions on nine issues, such as affirmative action, abortion, and military spending. This allowed us to see how respondents’ positions on a wide range of issues map onto their ideology.

Not surprisingly, people who call themselves “conservatives” tend to have more conservative issue positions. Similarly, self-described liberals tend to have left-leaning views. And moderates tend to be somewhere in the middle. As is well documented, ideology and issue positions are highly correlated.

But it turns out that the strength of that relationship depends on where people live. On average, self-described conservatives who live in very “blue” states — that is, states that tend to vote Democratic — have more liberal issue positions than conservatives in states that are more “red.” Similarly, based on ANES data going back to 1972, we found that liberals in red states often have more conservative issue positions than liberals in more Democratic states. In some years, this pattern is more evident for liberals; in others, it’s more pronounced for conservatives.

We replicated this finding in a second study using a convenience sample of Mechanical Turk respondents across the United States. We again asked people to place themselves on an ideological scale and to tell us their policy positions, this time on 10 issues. We also asked respondents which county they lived in, which allowed us to explore political contexts closer to home than the state level.

Once again, we found that the “bluer” a county was, the more likely its residents (of all political labels) were to hold liberal policy positions and vote Democratic than were residents of “redder” counties who professed the same political identity. For instance, a self-described moderate in a very Democratic county was more likely to hold liberal policy positions than a moderate in a very Republican one. And vice versa — a “liberal” in a very Republican county held more conservative views than a “liberal” in a very Democratic one.

What explains this pattern? We believe it stems from what we call “the political reference point.” The basic idea is that where we live, and the people around us dictate what’s seen as politically “normal.” In other words, if you live in a very red state, what seems middle-of-the-road and therefore moderate will likely be different than if you lived in a deeply blue state.

Consider how this might affect your answer if someone asked you where you land on the political spectrum. If your views fall to the right of most people around you (your political reference point), then you’ll probably see yourself as more conservative. If they fall to the left of most people around you, then you’ll probably see yourself as more liberal.

Take San Francisco and Tuscaloosa, Ala. San Francisco is heavily liberal (Hillary Clinton won 85.5 percent of the vote in 2016), while Tuscaloosa is very conservative (38.9 percent voted for Clinton). What may seem like a very liberal position in Tuscaloosa might seem pretty moderate in San Francisco. And what may seem like a very conservative position in San Francisco might be pretty moderate in Tuscaloosa.

Of course, political identity is still very meaningful. Individuals’ self-categorizations tell us a lot about their political views. But social context — and geography — influence the content of these labels.

Where you live, and the people around you, affect what you think these labels mean. And the complex social geography of the U.S. makes it difficult to accurately reduce Americans’ political views down to positions on a scale, or binary labels of “liberal” vs. “conservative.”

Alexa Tullett is an assistant professor of psychology at the University of Alabama, where her research looks into the social psychology of belief systems.

Matthew Feinberg is an assistant professor of organizational behavior at the University of Toronto’s Rotman School of Management, where his research looks into the psychology behind cohesive and productive groups.