This is a guest post by Harvard political scientist Ryan Enos.
Here’s a thought experiment: try to think of one Democrat you know who voted for Mitt Romney because he was so good looking. Have you thought of anyone yet? I ask this sort of question to my undergraduate students every time I teach my political psychology course, and I am yet to find anyone who will claim to know anybody who cast a ballot based on the attractiveness of a candidate.
Despite that nobody seems to know anybody who will admit to voting based on looks, there is a popular impression that good looks matter in politics — we’ve all heard about the famous Kennedy-Nixon debate – and a constant flow of research from psychology explaining why and how good looking politicians are more likely to win elections. Just recently, two articles in peer-reviewed psychology journals explored this very topic. One article, claiming that the tendency to vote for physically attractive candidates is related to an evolutionary need for disease avoidance, was covered in The New York Times. The premise of this article is that humans have evolved to avoid disease and the attractiveness of another person is a clue about the possibility that they carry disease. Voters in places where they are more likely to catch a disease show a stronger tendency to vote for physically attractive politicians than do voters in places where they are less likely to catch a disease. Perhaps you could also ask yourself if you know of any Democrats who voted for Romney because they had the flu on Election Day?
Here’s the problem with this type of research: several years ago, my colleagues and I extensively studied this phenomenon and found that the best available evidence says that politicians don’t get elected because of their good looks. In fact, after examining every contest for the U.S. Senate between 1990 and 2006, we couldn’t find a single election where the candidates’ appearance made the difference in the election. This isn’t surprising: Politics and voting are greatly affected by factors such as partisanship, the economy, campaigning, and even policy – all of which leave little room for voters to cast votes based on politicians’ looks.
What are psychologists finding then? Why do they see appearance as a deciding factor in elections, when our research showed that it didn’t? The basic problem is in how this phenomenon is often studied. If we really wanted to know whether a politician’s appearance affected an election, we would have to run an experiment – maybe by flipping a coin and randomly assigning some politicians to be good looking and some not so much and then seeing how voters reacted to these politicians. Of course, such an experiment really isn’t possible. Because we cannot randomly assign politicians to be good looking, we can’t know whether it is actually the looks of the politician that are causing them to win an election, or whether it is other characteristics that a good-looking politician might also have. For example, they might be more intelligent, wealthier, be better public speakers, etc.
Good-looking candidates are also not randomly assigned to run in a particular election – so features of where and when they run for office might also make them more likely to win. In fact, this is what my colleagues and I found: Physically attractive candidates run for office in elections that they are likely to win. This is probably because good-looking people also tend to be rich and successful and don’t want to waste their time on races they are likely to lose. When we statistically control for this factor, the effect of politicians’ looks on winning an election is close to zero.
So, why do some studies still claim that appearance causes politicians to win elections? In most cases, it is probably because the institutional boundaries of academia mean that psychologists don’t read the literature of political scientists (even “political psychologists” like me). But there is also a fundamental problem with the approach to research: These studies are acting like there is an experiment where none exists. For example, take the claim that avoidance of disease causes voters to favor good-looking candidates. The researchers established this by looking at the life expectancy and infant mortality rates in congressional districts and demonstrating that good-looking politicians are more likely to win in districts with high infant mortality and low life expectancy. The problem is that both these characteristics are not randomly assigned to districts, and they are associated with lots of other things – for example the demographics of the district. Factors like race, education, income, the competitiveness of the district, and other factors are all correlated with infant mortality – that is districts that have high infant mortality are more non-white, lower education, lower income, and less politically competitive. All these factors are also correlated with how much voters are engaged in and have information about politics, which could also affect whether or not a voter makes a decision based on attractiveness. In fact, we know that voters that do not care about politics — so called “low information voters” — are more likely to be influenced by the appearance of a candidate than those that do care about politics. If more low-information voters are found in districts with higher infant mortality, then this might be the explanation, not disease.
In fairness to this study, the authors did demonstrate the plausibility of their claim in a laboratory raising subjects’ awareness of disease and demonstrating that this made them more attracted to good-looking leaders than those whose awareness hadn’t been raised. From an evolutionary perspective, this makes sense. But how does it translate into real-world politics? In the real world, voters’ choices about politicians are filtered through strong, long-standing attitudes, which research tells us affect voting decisions, such as party identification, racial prejudices, and the state of the economy. With all of those and other things to consider for voters in the real world, it is hard to see a candidates’ appearance as having much influence in a real election.
So, what about Kennedy-Nixon? Didn’t Kennedy win in 1960 because of his good looks and mastery of the new-found art of television make-up in the first Presidential debate? We are famously told that television viewers thought Kennedy had won, while radio listeners thought Nixon had won, thereby demonstrating the power of appearance in politics. In fact, there is surprisingly little evidence of this claim and, when you look a little closer, it seems like this favorite gem of high-school history classes might actually demonstrate the power of these long-standing, predictable attitudes in politics. As Mark Blumenthal summarized, academics that have examined the evidence have found little to support the Nixon/radio vs. Kennedy/TV claim: the only non-anecdotal evidence is a single survey by a marketing firm, conducted long after the debate, and lacking the basic standards of scientific integrity that would be found in modern political surveys. In fact, a very plausible explanation for Nixon’s superior radio performance is that most radio listeners were rural Protestants who were inclined to be skeptical of Kennedy as a Roman Catholic. Like with studies of disease and voting for good looking candidates, we find that long-standing factors, such as religion, rather than appearance, are most likely what is influencing voting decisions.
Do appearance and other short-term psychological factors influence politics and decision making? Of course, but these factors must compete with the forces that make politics largely predictable and stable. Does this distinction matter? I contend that it does. Claiming that voters are making decisions based on good-looks or fear of disease or other such factors makes many of us uncomfortable because we think that it is not how a democracy is supposed to work. Rather we believe that voters are supposed to carefully weigh issues and make informed decisions. Unfortunately, there is pretty strong evidence that voters do not do this either. But in either case, the stakes of the question are quite high: Researchers are asking whether voters behave in a way that makes democracy seem like a good form of government. The high stakes of this question should also place a high burden on researchers to get the answer right.