Social Science Confirms that Mitt Romney Is Really, Really Good-Looking

by John Sides on August 31, 2012 · 28 comments

in Campaigns and elections

Harvard University political science Ryan Enos reports some important findings:

Mitt Romney is better looking than almost everyone reading this blog.  Back in 2008, I wrote about how Sarah Palin’s looks put her in the 95th percentile of politicians.  Romney has even Palin beat—he scores above the 99th percentile.

These results come from a study with my colleagues Matthew Atkinson and Seth Hill, in which we developed a method for obtaining the ratings of the facial competence of governor and Senate candidates from 1994 to 2006 by showing the images of these candidates to undergraduate students for 1 second, as pioneered by Alex Todorov.  In 2007, when we collected this data, we removed highly-recognizable candidates so that opinions about the candidates, other than their appearance, would not affect the ratings.  However, as with Palin, we are fortunate that Romney was a relative unknown at the time (at least to the undergraduates in California that we used), so we obtained a rating of his face.

And what a face it is!  We gathered the ratings of 728 candidates for Senate and Governors’ seats and Romney outscored all but four of them.  The only persons to win election that beat him are Russ Feingold (the best looking Democrat) and John Thune (the best looking overall).  Romney also appears to far outdo Paul Ryan, who came in in the 67th percentile of the 2004 House candidates (although the photos did not include abs).  (Also, that study only included white male candidates and the House was not measured on a common scale with Senators and Governors, but I’d feel pretty confident saying the 67th percentile of the House puts you well behind Romney).

We don’t have a rating of Obama because we deemed him too well-known, even in 2007, because his Senate race had attracted a lot of attention and there was already an excitement building around a possible White House bid.  However, we do have a score for Biden—and Romney has him beat badly.  Biden only comes in the 62nd percentile of Senate and governor candidates.

So, if the election were decided on looks, it would be no contest.  Fortunately for Obama and Biden, the election is not decided by looks.  As we point out in our paper associated with the study, most of the correlation between candidate appearance and election outcomes is probably spurious.  Very few voters are willing to cast their ballot for a candidate based on looks – we estimate that if a candidate moves from the 25th to the 75th percentile in attractiveness, this is likely to gain that candidate about 3.5 percentage points in vote among independent voters, which was not enough to decide the winner of even a single Senate race out of 99 that we examined.  Rather than good looks directly affecting voters’ decisions, it is likely that good looking people like Romney have a lot of success in life, obtain significant human capital—education, career success, education—and because of all they have to lose, they are strategic about which races they enter.

In a certain respect, Romney’s career both fits and is counter to this explanation, because he ran for Senate in 1994, against Ted Kennedy when he did not have a good chance of winning, but he did not run for Governor until 2002 where he used his good looks and the considerable capital he had earned from the Salt Lake City Olympics to run in a seat with no incumbent.  Of course, Romney may not have been able to be as strategic about when to run for President – and unfortunately for him, most voters seem to have made up their mind long ago—nevertheless, if a candidate’s appearance every can make a difference, it should make a difference for Mitt Romney and his face in the 99th percentile.


If Romney wins, perhaps this will be another piece of evidence for Lee Sigelman’s classic work, “Toward a Stupidity-Ugliness Theory of Democratic Electoral Debacles.”

{ 28 comments }

RobC August 31, 2012 at 2:32 pm

And we know these results are reliable because the perceptions of undergraduate students at one university (or at most a few) are a great stand-in for perceptions of Americans of all ages, regions and social classes. Or so we can infer from the number of times studies based on undergraduate respondents are fecklessly treated as revealing truths about the general population. But they’re conveniently at hand, so quit yer belly-aching.

John Sides August 31, 2012 at 2:58 pm

RobC: It’s easy to complain when social science does not rely on a representative sample of Americans. But you’ve got to go further if you want to make this critique. Why would students’ perceptions of attractiveness differ from those of others? If they do differ, would this bias the results reported here? If so, how? See also:

http://www.ipr.northwestern.edu/publications/workingpapers/2009/wp0905.pdf

RobC August 31, 2012 at 3:24 pm

Offhand, I can think of several ways in which undergraduates’ perceptions of attractiveness may differ from those of others. If we assume that students seek to make their own appearances attractive to themselves and to those whom they would like to impress, the prevalence of scraggly beards among undergraduates and the prevalence of body art may well be considered more attractive to students than they are to the general population. Coming from a significantly middle class background, might students find the country club good looks of a Mitt Romney more appealing than others from a poorer or more ethnic background would?

But beyond these particular issues, should the burden be on those who say a sample is representative of the general population to prove it or should it be as you would have it, that those who complain about an unrepresentative sample have the burden of demonstrating that it is unrepresentative? That seems as if we are turning the nature of scientific proof on its head.

If People Magazine were to rank popular perceptions of the attractiveness of candidates, my guess is that they’d use a more broad-based sample than a few dozen undergraduates. Shouldn’t we aspire to the high standards of People Magazine?

Ryan Enos August 31, 2012 at 4:06 pm

RobC: in other similar studies, we’ve surveyed a general population using some of the faces in our same results and recovered similar scores. Moreover though, there is considerable evidence that there is some common currency of attractiveness because researchers have surveyed adults in Mexico about U.S. politicians and found opinions that have similar properties to U.S. undergraduates and the same thing has been done with Swiss school children. There is a long list of countries where this has now been demonstrated, so I’d say we have pretty good evidence now that undergraduate opinions can be generalized to to larger population.

Andrew Gelman August 31, 2012 at 4:50 pm

Also, not too many serious candidates for higher office have scraggly beards. At least, not yet.

JBeerbower September 1, 2012 at 8:52 am

Hmm I flipped through the article and it had no data — it only referred to other studies. The idea expressed above “why should students be different” is not very persuasive to me. The burden of proof lies with the study to show that their sample is representative.

From market research experience, I would say that the “Social Science” above is an entertainment statistic. Good for amusement.

Bob September 2, 2012 at 3:10 pm

Agreed. And quoting from one study which shows that students and the general population share similar tastes is misleading: a literature review would be more convincing.

Kevin Hill August 31, 2012 at 5:13 pm

Mitt is purty…..

Marie Burns August 31, 2012 at 8:21 pm

I dunno. The students were looking at still photos of faces. When you see the Full Romney in motion, he walks like a robot, & as I watched him waving to the crowd during the convention, the way he gestured reminded me of Richard Nixon. The resemblance was so strong, it was kinda creepy.

So if voters never see anything but a still photo of Romney’s face, maybe they’ll be positively impressed. But if they see Mitt on the teevee, doing his excellent Nixon impression, they’ll probably be more likely to pull the lever for the skinny guy with the big ears & a winning smile.

paul gronke August 31, 2012 at 9:14 pm

Is there a parallel study for good looking-ness of political scientists? I want to see a Monkey Cage smackdown!

Patrick Stewart August 31, 2012 at 11:27 pm

Interesting study, although I think Obama might be considered equally attractive _and_ due to his neotonous/baby-faced features a better candidate for the current social/economic environment than Romney with his strong jawline and more mature features (along the lines of the Little et al. 2007 paper on facial appearance suggesting candidates like Romney would be a better external crisis/wartime candidate). I would be interested in seeing whether measures of symmetry have been carried out with Romney and Obama as this is a key factor in attractiveness (I believe Jim Schubert found that Bill Clinton had model-level facial symmetry).

Regardless, Mitt Romney’s smiles – even the Felt/Duchenne smiles – seem to have an element of either contempt or disgust to them when coded on a muscle-movement basis (unlike Obama’s rather impressive felt/Duchenne smile). That, combined with his less-than-fluid body movements, make him less than accessible to egalitarian-minded individuals…

joe arrigo September 1, 2012 at 10:49 am

From the reports I’ve seen on TV, and from what I’ve read, looks do make a significant difference in success, including how tall an individual is. I believe that elections would probably turn out differently if radio was the only means of mass public communication. With TV, it’s a whole different ball game. Would Lincoln have won election if TV were available?

Patrick Stewart September 1, 2012 at 5:12 pm

Joe Arrigo: They sure do – especially in low information elections when short-hand decision rules are relied upon – height plays an especially important role (Kerry notwithstanding) – the question is how to disentangle all the factors, especially when body language does a good job of conveying personality, the face emotional intent, and physical characteristics such as facial features, height, and body morphology providing information concerning genetic quality. So, to answer the question about Lincoln – he might have won the Presidential election (it was a highly important election), but likely wouldn’t have made it to the level of contender in the first place.

You do raise a good point concerning vocal quality as well – I truly believe a major reason behind Al Gore’s loss was his accent…

Patrick September 2, 2012 at 2:47 pm

My confusion here is that 1 second long glances at still faces with nothing else included doesn’t on it’s face resemble any sort of social interaction or media messaging that is encountered in non-experimental setting. So is this “1-second-still-face-attractiveness” a good proxy for any sort of actual impression of people we have? This measurement may be very precise and very accurate, obtained under strong controls, but in finding a controlled measurement have you controlled out what you are actually looking for?

My uninformed guess is that our actual exposures to people in life and in media produce more heterogenous evaluations. The piece aslo notes that this very particular version of “attractiveness” doesn’t correlate well with most elections. OK. But does that mean that political success isn’t well correlated with some other standard of measuring attractiveness?

Putting aside the de-natiralizing of the data, the other obvious question is if Mitt is very handsome and Paul Ryan less so among politicians, how do they compare against the population as a whole? Hypothetically it could be that “attractivenss” is very important, but it is subject to diminishing returns. Not everyone who aspires to political office ends up a candidate for Senate or Governor. Is the set of 728 much more attractive than a race, age, and sex weighted sample of the population as a whole?

Patrick Stewart September 2, 2012 at 5:09 pm

Patrick: good points – this is a nascent science, and plenty needs to be done to understand how well this relates to the “real world”. Jim Schubert had a poster presentation in the early aughts (2001 or 2002, I believe) in which he considered facial attractiveness and presidential primary campaign survival rates (how long a candidate lasted in a campaign) – if memory serves (the poster has been taken down off the Northern Illinois University website), facial attractiveness didn’t directly impact survival or media coverage, but did influence how much funding a candidate received, which in turn affected media and electoral outcomes. So, there’s that… (although I think I remember they used a longer time frame as their stimulus – 5 seconds I believe – and showed a silent clip of them speaking – which may have conflated attractiveness and Markus Koppensteiner’s research on body language and personality). One other factor to consider is that we make “gut” decisions (less of an analogy now that research suggests this is actually the case!) that are based upon fleeting, even pre-conscious stimuli – that then inform our conscious/cognitive evaluations.

As to your second point – well taken – although I would suggest that we might consider attractiveness is both universal (through symmetry and indicators of higher levels of testosterone with our male leaders – with women, it is a bit more complex as to the testosterone) and affected by social/cultural factors. More to the point – Bailenson and Iyengar had a Fall 2008 Public Opinion piece that suggested morphing politicians faces with those resembling oneself had a positive effect on willingness to support that candidate.

In other words – plenty of fun research to be done!

Patrick September 2, 2012 at 11:42 pm

I’m aware that our decisions are often more rapid than they seem from the perspective of conscious experience, and therefore, by necessity, based on quick stimuli, but that doesn’t mean that the relavent stimuli are necessarily captured in a still image with no sound, does it? I may have emphasized the “1-second” in my initial comment, but the “silent, still” is actually a bigger question mark for me. I can see that if your question is “facial attractiveness” why it is done this way, but I am skeptical that a purely “facial attractiveness” is what matters, I suppose.

For an extreme example, although one that is marred by being anecdotal and subjective: The movie Tron:Legacy contained performances by Jeff Bridges where digital effects were used to make him look more like the age he was many years previously when the first movie was filmed. In a straight on still shot this effect was very convincing, but as he moved about the limits of the process used to make the altered face move like a real human face became very apparent. For me at least the image became somewhat disconcerting.

Certainly no politician will behave like a weird CGI artifact, but I would be very surprised to learn that how people move and sound aren’t part of how we end up judging their attractiveness.

Casey Klofstad September 3, 2012 at 7:28 am

On the question of vocal signals and thin judgement of candidates (albeit in hypothetical elections, conducted in a lab, and mostly with undergraduate subjects), see these two studies:

http://www.voiceresearch.org/pdf/tigue%20et%20al%202012.pdf
http://www.as.miami.edu/personal/cklofstad/15_voice_pitch.pdf

The references in these papers point to a broader literature on vocal perception and judgement of the speaker on factors such as attractiveness and strength.

Casey Klofstad September 3, 2012 at 9:34 am

With regard to the influence of voices on votes (albeit with hypothetical elections, in a lab, with mostly student subjects), see these two recent studies:

http://www.voiceresearch.org/pdf/tigue%20et%20al%202012.pdf
http://www.as.miami.edu/personal/cklofstad/15_voice_pitch.pdf

In short, lower pitched voices are more electable.

Patrick Stewart September 3, 2012 at 10:57 am

Patrick: You’ve latched onto some very interesting research questions – more information is carried with more channels of information (image, sound, movement, smell, etc.) – but a still of 33 milliseconds is enough to alter perceptions. So, with a nod to your discussion of “uncanny valley” of Tron:Legacy, I’d agree that attractiveness is a multi-dimensional construct – but I’d think that a 1-second still would capture a major proportion of the variance in judging physical attractiveness. However, I think you’ve captured a very important second generation scientific question of trying to disentangle what elements contribute what to the definition – which in turn needs to be redefined to reflect the measures defining it…. More specifically, how much of attractiveness is an automatic reaction to symmetry and other indicators of health, how much of it reflects valued in-group attributes, and how much of it reflects personality cues that we value individually?

Patrick September 3, 2012 at 11:19 am

In addition it would seem that in disentangling how “attractiveness-as-it-matters-in-social-reality” is built up, this is an area where cross cultural sampling is needed as well. As I understand it the symmetry and such does matter everywhere for everyone, but as we identify other factors, are they similarly universal? Are the “weights” the same everywhere?

I admit my not particularly informed guesses will always be biased towards particularity and contingency, rather than universality, which is probably an artifact of being trained as a historian in the early 21st century.

Patrick Stewart September 3, 2012 at 1:08 pm

Casey: Thanks! This is awesome stuff (and in very, very nice journals too!). If you don’t mind, I’d love to keep in touch concerning your research.

Patrick: Leslie Zebrovitz (sp?) has done some rather nice work on cross-cultural studies of attractiveness – she has an edited book out with Gillian Rhodes from 2002 that answers a lot of your more general questions concerning cross-cultural attractiveness (although probably the best place to start, in my opinion, would be to consider a developmental perspective – how babies respond to different faces). However, research concerning politicians is probably a bit more sparse – it has been only relatively recently that political scientists have taken a more biological bent with their research (and usually after hedging bets with other more traditional research) – and given that political choice such as explored here and by Casey’s research agenda are contingent on more-or-less democratic systems (i.e., more Westernized societies), cross-cultural research is even more difficult to carry out effectively.

In any case – we need all perspectives, driven by rigor, to move forward with understanding voter decision making!

Patrick September 3, 2012 at 5:59 pm

@Patrick Stewart. Will look up the Zebrovitz. As to cross cultural attractiveness and politicians, democracies gets us cultures as different from the US as India and Taiwan. So while there are no democracies without “Western influence” there are certainly democracies that have very large cultural differences from the USA and Western Europe.

mike September 4, 2012 at 1:44 pm

In case people are still reading this, another thing to consider is that attractiveness is not constant–I’ve found it varies by the content of political statements/actions. So, for example, Angelina Jolie is a “8.8″ on a 10-pt scale, on average, but if you find out she supports/endorses a candidate from the opposing political party, it drops to a “7.8″. I think this gets at the comment that a picture of Romeny doesn’t capture his “ugly” behavior that would affect his attractiveness ratings.

Jack Bender September 5, 2012 at 1:02 pm

A good question is… How does the physical appearence effect the way we as Americans see the presidential candidates? Granted Mitt Romeny is no Barney Frank or Dennis Kucinich in terms of looks, what if he did though? Could he still win the election? What is also interesting is that Obama wasn’t even included in the study, even though he was “well known”. It is intriguing as well to see that even in Paul Ryan’s younger years he is didn’t make it in the ninetieth percentile range. Ryan seems to be a good looking guy now, imagine him back in the day.

Emily September 5, 2012 at 9:06 pm

There’s an article I just read (http://www.guardian.co.uk/science/2011/aug/14/beautiful-people-selfish-nature) that questions if “beautiful” people have a more selfish attitude. It would be great if there were a follow-up study looking at the policies of the elected officials involved in the study mentioned here to see if the results line up.

Billy M. September 6, 2012 at 11:57 am

“Social Science Confirms that Mitt Romney Is Really, Really Good-Looking”

This will be a startling revelation and very useful information for the blind community. For the rest of the world, meh, not so much…

Richard S September 7, 2012 at 1:02 am

More evidence that John Adams was correct when he called beauty a “talent.” And once again social science does little more than confirm what keen observers of human nature already knew.

Matt September 8, 2012 at 1:01 am

It is somewhat confusing (and misleading?) that the blog post refers to “attractiveness” and “good looks.” It appears as though the study only examines the association between facial *competence* and candidate selection. The relationship between competence and attractiveness is complex, and far from perfect, as research by Linda Jackson and others has demonstrated. Am I missing a footnote or sentence somewhere? Or is there decent evidence that competent faces are always perceived as attractive (and not vice versa)? It seems easier to show that attractive faces are perceived as competent. In general, I don’t think social psychologists would want to be caught using a measure of facial competence as a proxy for attractiveness.

Comments on this entry are closed.

Previous post:

Next post: