Politics and Personality: Most of What You Read Is Malarkey

Many political psychologists believe that personality traits and politics go together. Researchers are not so sure.
Many political psychologists believe that personality traits and politics go together. Researchers are not so sure.Illustration by Wren McDonald

You’ve seen the headlines. You’ve maybe even quoted them, especially now that it’s election season. “Conservatives More Susceptible to Bullshit Than Liberals.” “Study: Are Liberals Smarter Than Conservatives?” “Studies: Conservatives Are from Mars, Liberals Are from Venus.” All of these tantalizing connections between personalities and politics sound academic if not downright scientific—they cite studies, after all—and, depending on your own political leanings, you’ve likely had one of two reactions: Of course, I knew it! or This is bogus! As it turns out, if you’re in the second camp you’re probably right—but maybe for a reason that’s different from what you thought.

For many political psychologists, it seems abundantly clear that traits and politics go together. There’s evidence that many aspects of personality develop quite early in life and have a genetic component, but we don’t become actively political until we are older. So it’s sensible to assume that the one might have some bearing on the other. But most of the work on the subject in the past decades has consisted merely of scientists conducting surveys and observing correlations. Few researchers have ever asked whether what they’re seeing actually implies causality or if the correlations are even meaningful. (The fact that correlation does not equal causation has been amply illustrated by a self-styled correlation debunker, Tyler Vigen; a recent visit to his eponymous Web site shows a 0.998 regional correlation between U.S. spending on science and technology and suicides by hanging, strangulation, and suffocation.) And so, almost a decade ago, Brad Verhulst, a behavioral geneticist now at Virginia Commonwealth University, asked himself just that: Is the personality-politics link truly causal? A relationship between personality and political leanings is “a completely reasonable thing to expect,” he told me when we spoke recently. He wanted to use his knowledge as a geneticist to explore the causal linkage that he was certain would be there.

He found something quite different. “Unfortunately, the empirical evidence doesn’t seem to support that strong causal hypothesis,” he said. In an analysis of 28,877 people from the Mid-Atlantic Twin Registry (known as the Virginia 30,000), and then in a second, separate longitudinal study that followed a group of more than eight thousand twins and siblings for ten years, he found no evidence of any form of causation. Instead, he found a relationship that was more complicated. It is far more likely that politics and personality traits are both influenced by some earlier genetic and environmental factors. In other words, they may indeed be related, but the fact that someone is liberal does not make him more tolerant, for instance, just as being tolerant does not make someone liberal.

In the first sample (the Virginia 30,000) Verhulst and his collaborators found that, while some modest correlations did exist between certain traits and attitudes—for example, between conservative economic views and measures of neuroticism—there was no evidence that there was anything causal to the correlations. But genetic analyses can be tricky. So Verhulst and Peter Hatemi tried a longitudinal study: follow people over time, map their personality and political leanings, and see if changes in the one cause changes in the other. They looked at two samples, one of adults (7,610 twins and siblings who were between the ages of nineteen and seventy-eight in 1980) and one of adolescents (1,061 twins and siblings who were between sixteen and nineteen in 1998). Each group had been tested in two waves, ten years apart. At both points, researchers looked at political attitudes (views on topics such as abortion and gay marriage, as well as responses to statements such as “I believe we should look to our religious authorities for decisions on moral issues”) and personality measures. They found that personality did shift over time—not by huge amounts, but perceptibly. People could become more or less extroverted, more agreeable or conscientious, or any number of things. Political attitudes were slightly more stable, among both the adolescents and the adults: people who were conservative tended to stay conservative. And, most important, changes in personality did not predict changes in politics. “We conclude that both personality traits and political attitudes are independently part of one’s psychological architecture,” the authors write.

None of this work denies the possible existence of correlations between some traits and some beliefs, but it does raise the question of what those correlations signify—especially since, setting aside the issue of causality, some early research on politics and personality may have overstated the connections because of a built-in tautology. As it turns out, in many of the early conceptions of personality traits political leanings were purposefully built into the survey questions used to assess personality. Early theorists explicitly wanted to capture political attitudes with their scales. For instance, in order to measure “openness,” Robert McCrae and Paul Costa, the creators of one of the most widely used personality measures, the NEO-PI-R, use “favors conservative values” as one of the items to assess the degree to which someone possesses a trait. In measures of agreeableness, there are items like “we can never do too much for the poor and elderly” and “human need should always take priority over economic considerations”—both statements that are embedded in a specific political ideology. In other words, our political beliefs are actually used to assess our level of the trait. By definition, we are higher on openness, say, if we are politically liberal.

That circularity, however, is often lost on modern minds. So what happens when we look for correlations between, say, openness and liberalism? We find one because it’s largely tautological. It doesn’t help that most of the scales are proprietary and, hence, not publicly available: it’s hard to find a list of items that contribute to each facet and, therefore, to see for yourself just how embedded politics already are in traits that people mainly think of as abstract concepts.

When I asked Verhulst if it was possible to circumvent this, he pointed out two things: first, not all traits are tautological (openness is the most egregious example). If you wanted to study risk-taking, for instance, you could conceivably avoid any sort of circularity. And, second, some people try to remove the political items—but with mixed results. “If you are concerned with political attitudes, you might remove items that are explicitly political,” he told me. “But just because you’ve taken those out doesn’t mean you’ve fundamentally changed what openness measures. Even if you’re removing some of the completely tautological items, you aren’t quite hitting that problem.” So in his studies Verhulst also used an entirely different scale, one that is “relatively untainted by explicitly political items,” as he and his co-authors write. The non-causal correlations he does find are in the weak 0.2–0.4 range. (A perfect correlation is 1; 0.2 is considered “negligible,” while 0.4 is “low.”)

But the desire for causality, or at least some basic truths—Of course those Republicans are closed-minded people! Of course those damn Democrats are neurotic!—persists. And despite studies like Verhulst’s, we can’t seem to let it go. Headlines keep appearing; researchers keep pointing it out. Verhulst’s 2012 paper—the analysis of the Virginia 30,000—was recently in the news because of an authorial correction. It seems that one of the minor correlational directions had been reversed: it was liberals, not conservatives, who scored slightly higher on a measure of psychoticism, which takes into account aggressiveness, antisocial tendencies, and egocentricity, among other traits. (“Science Says Liberals, Not Conservatives, Are Psychotic,” the New York Post wrote.) In the correction note, Verhulst, Hatemi, and a co-author, Lindon Eaves, stressed that the error, while regrettable and sloppy, didn’t actually affect any of the paper’s main conclusions: that there was no causality between personality and politics, and that correlations were small. Really, neither liberals nor conservatives are particularly likely to display traits of psychoticism, and, to the extent that liberals are, it’s no more related to their politics than to their shirt size. “We found no evidence that personality traits play a causal role in the formation of political attitudes,” the authors write. “Our focus and novel results showed that whatever the directions of the correlations are between personality traits and attitudes, the relationships are spurious.” And so, while it of course needed to be fixed, at the end of the day it didn’t matter all that much.

Others in the media and in politics thought otherwise. At least in the U.S., the party you believe in plays a big role in how you conceive of yourself. It feels good to think that your party is smarter, and that the smarts are what drive people to your party. It also feels good to say that the other guys are psychos. “ ‘It’s spurious, there’s no causal relationship,’ ” Verhulst says. “That could be pretty depressing for people who’ve invested a lot of time in this.” Here’s what won’t make a good headline: “Small and Spurious Correlation Shown to Have Been Backward, but It Doesn’t Matter That Much, Because the Point of the Paper Was That There Is No Underlying Causation After All.”

Social psychology touches on our most cherished beliefs about how the world should operate. We are more than happy to accept the counterintuitive—as long as it doesn’t conflict with our central notions of what ought to be. And, when it comes to politics, it can be awfully difficult to put your desires aside and to acknowledge that the world is a much messier place, where open-minded people might be conservative and liberals may well be conscientious. “People get passionate when they’re talking about these things,” Verhulst points out. “It would be nice if it was less passionate, but political values really do have a profound impact on our daily lives, as do the personality traits we have. They are fundamentally striking at what it means to be human.” And so levelheadedness remains difficult. Here’s a headline you’re not so likely to see: “Conservatives Are from Earth, Liberals Are from Earth, and Scientists Aren’t Really Sure About How They End Up That Way.”