David Broockman is a political science graduate student at UC Berkeley.

As partisan divisions in Congress have grown, concern has mounted that politicians are pursuing much more extreme policies than nearly all their constituents support. According to a common line of thinking, campaign donors and primary voters are pulling politics to the extremes. Most Americans, the story goes, would prefer their legislators to chart a moderate course.

In a recently published paper, I question this view. Using a survey designed to measure support for extreme policies, I find that the characterization of the public as largely centrist rests on shaky ground. On many issues, much of the public appears to support more extreme policies than legislators do. And while many argue that today’s engaged activists support more extreme policies than the broader public, my findings suggest the opposite: The disengaged and infrequent voters who allegedly constitute the moderate middle are actually more likely to endorse extreme policies than politically active voters.

Why might we have missed much extremism in the public generally and among the less engaged? The answer is subtle, but has important implications for how we should think about the the public’s attitudes and politicians’ positions. And it might be best explained by pretending you have a crazy uncle.

Suppose your uncle believes that the United States should nationalize the health-care system (a very liberal view) and that gay people should be jailed (a very conservative view). And suppose your uncle is represented in Congress by a moderate Republican who supports civil unions (but opposes gay marriage) and who supports helping the poor purchase health insurance (but opposes Obamacare), two positions just right of center.

Your uncle’s views can’t really be described in ideological terms like “center left” or “very conservative.” He has some mix of very liberal and very conservative views, many of them extreme. But if we try to compare your uncle’s views to his congressperson’s positions in abstract, ideological terms, as academics and journalists often do, some plain facts about your uncle and his legislator both become obscured. Since your uncle supports some liberal policies and some conservative policies, we’d call him a “moderate on average.” However, his congressperson’s conservative votes on both Obamacare and gay marriage mean we might call the legislator conservative. We thus might condemn your uncle’s congressperson for being a conservative extremist while celebrating your uncle’s moderation. However, it’s quite clear that your uncle’s views tend to be further outside the mainstream, just not consistently in one direction.

The lesson is this: Attempting to describe Americans’ general political views across multiple issues can yield misleading pictures of where they really stand relative to their representatives. Most Americans are simply not ideologically consistent enough that ideological labels such as “conservative,” “liberal” or “moderate” accurately describe them.

So, how common are the “crazy uncles” and how common are the real moderates in the public? In this survey, I sought to measure support for extreme policies, like establishing a maximum income or banning the sale of birth control pills. I presented Americans with questions on a dozen different issues. Each question had seven possible responses — a couple of extreme response options on either side (let’s call these points -3, -2, 2, and 3), options representing the parties’ general positions (at -1 and 1), and a moderate option (at 0). I also sent this survey to a sample of state legislators.

How often do members of the public support more extreme policies than legislators?

The most common way of looking at the data is presented in the figure below. Here, I averaged each person’s responses across multiple issues to generate a summary of each person’s overall views (or “ideology”) and then calculated how extreme their views appeared overall. For example, your uncle would score as a 3 on the gay rights question and a -3 on the health-care question, meaning he would average out to a 0, or “moderate on average.” The congressperson who scores at the Republican position of 1 on both issues appears more conservative — and more extreme — with an average position of 1. Looking at the data this way, the figure suggests that the public is much more moderate than legislators, as political scientists increasingly report.

But what if we don’t treat Americans as ideologues and instead look at their views on each issue, one by one? To calculate the next figure, I considered how far from the middle each respondent’s views typically were on each issue. For example, if your uncle answered a 3 and a -3, he would be on average 3 away from the middle. However, when his congressperson answers at 1 on both, it identifies his responses as “on average 1 away from the middle,” now correctly capturing the fact that his positions tend to be more moderate than your uncle’s, even though he is more ideologically consistent. When we compute extremity this way, the new conventional wisdom is upended: The public appears likelier to support extreme policies than legislators.

Indeed, although each of the parties is out of step with public sentiment on some issues, neither consistently outflanks the public. For example, about 40 percent of Americans seem to have more liberal positions on tax policies than most Democratic elected officials, while much of the public would also prefer more conservative policies on immigration and abortion than most Republican elected officials would endorse.

There is a similar inversion of conventional wisdom when it comes to the attitudes of the most politically engaged Americans, who are often blamed for pulling politics to the extremes. When I examine whether more engaged individuals are more likely to support extreme policies on any given issue, they actually appear slightly less likely to do so. (For those interested in the details, the paper dives into these issues at greater length.)

Many political observers have the sense that something has gone awry in American politics, and the metaphor of extreme elites spurning a moderate public is understandably appealing. They may be right to worry that American politicians are more partisan and less responsive than in decades past, that interest groups have succeeded in countermanding majority will in many instances, or that each of the parties is out of step on some issues. However, concerns about a wholesale “disconnect” between extreme elite positions and moderate public opinion may be overstated.

In fact, the Americans who we often call moderates may be less likely to adopt moderate positions on any given issue. These Americans appear more aptly described as “conflicted,” agreeing with each party on some issues and more extreme than either party on others.

Overall labels like “moderates,” “liberals” or “extremists” are often not able to describe individual Americans’ opinions at all — and, when we do analyze public opinion in ideological terms like these, we’re likely to be led astray. As the relationship between mass opinion and politicians’ actions differs a great deal by issue, better understanding about what may ail American politics may require attending to the unique politics each issue presents.

Further Reading

This is the latest post in our ongoing series on political polarization. The previous posts are listed below.  — Dan Hopkins