Palin a 9.5, scientifically speaking
September 11, 2008 | 11:34 AMRyan Enos
How good looking is Sarah Palin? I can tell you...exactly.
She is 0.75 on a scale of -1.2 to 1.06.
It seems like nearly every delegate at the Republican Convention last week, following Sarah Palin's introduction to the world, described her leading qualification as being "a beautiful lady" or some derivative thereof. Putting aside the matter of whether there should be concern that the politically sophisticated attendees at the convention offer this as a qualification for office, it is interesting to note how smitten they seemed to be by her. Is she really all that?
I have a feeling that I will end up mentioning often in this blog that, as political scientists, we don't have that much scientific to say regarding this election. That is not to say we don't try, but just that elections are extremely complex phenomenon, and they are hard to measure scientifically.However, I do have something scientific to say about Sarah Palin's appearance.
There is a buzz around about whether Palin's looks will help her move from Juneau to Washington. An honest political scientist will tell you that they do not know the answer to that question. However, I can begin to have some informed speculation because I have actually quantified Palin's looks - as I have those of Joe Biden, Mitt Romney, and hundreds of other politicians. The Grand Old Delegates might be disappointed to learn that their candidate's appearance is only 0.75. Based on all the chatter about her on websites, such as
http://vpilf.com, and the schoolyard talk of some of my Right-minded friends, you might think her supporters are ready to declare her a 10, the fact that my scale only goes to 1.06, not withstanding.
However, despite the seemingly unimpressive score of 0.75, her new throngs of admirers would probably be happy to know (and have their opinion confirmed) that she is far above the average politician, in fact she is firmly in the top 5% of politicians when it comes to looks.
Let me explain how I can claim to know all of this.
My colleagues, Matthew Atkinson, Seth Hill, and I undertook a survey that quantified the "facial competence" of hundreds of candidates for Congress and governor's mansions, between 1990 and 2006. This was following on the work of the psychologist Alexander Todorov and his colleagues at Princeton. They showed undergraduate students photographs of the faces of competing candidates and asked them to pick the more "competent." In the journal Science, they demonstrated that the difference between the ratings of which candidate was more competent, correlated, that is matched up, with the difference in actual vote share between the two candidates. The more competent looking candidate won the actual election almost 70% of the time. What's more, they only flashed these photographs for a very short time, about 1 second. So, the judgements that correlated with election outcomes were actually snap-judgements, not reasoned opinions.
Todorov and his colleagues did not take a position on whether their research means that voters are making decisions about for whom to vote based on who is better looking. It is an intriguing question though, that raises some others. Do people actually even know what candidates for relatively obscure offices, like the United States House of Representatives, look like? If they do, what type of person would actually have their vote changed by this? Would a Republican really vote for a Democrat because the Republican candidate isn't pretty enough?
In many ways, their finding was compelling and curious: there is well-supported political science evidence that the more attention a person pays to an election, the more likely they are to be a firm believer in one side or the other. So, would a person that pays enough attention that they know what a candidate looks like, and is a firm believer in, say, the GOP, really vote for a Democrat because that Democrat was just so much better looking?
Questions like these led us to undertake a similar study. However, we created a scale that allows a comparison of all candidates, Palin to Biden for example, and not just those candidates that actually ran against each other. To do this, we sat hundreds of UCLA students in front of computer screens and flashed photos of the candidates at them. Instead of just flashing the pairs of candidates that actually ran against each other, we randomly matched candidates, and asked the students to choose which looked more competent. This allowed us to create a scale with a score for the face of every candidate. Seth has an example of the survey on his webpage at
http://sjhill.bol.ucla.edu/faces/ The numbers don't mean anything in an absolute sense. The scale is arbitrary, but the numbers have a relative meaning. So, if you like Palin, don't get too upset about her seemingly low score. I'll explain how she stacks up against her competitors shortly.
For those of you that like sports, this method of scaling is somewhat similar to how those computer rankings are made for college football. Those rankings allow for two teams that have not played each other to still be compared because the computer keeps track of how they did against common opponents or a team that one team played that played a team that the other team played. Or a team that one played that played a team that played a team that the other played, etc. If there are enough games, these rankings can give a fairly accurate idea of how two teams compare, even though they never actually played each other. Now, for those of you saying that those darn rankings keep your team out of their bowl game every year!, you might be right because in college football, there are not many games, and the accuracy is dependent on the number of games that occur. In our study, each face engaged in hundreds of "games" against other faces -- so we are fairly confident about the accuracy of our ratings. Can we separate Palin from Elizabeth Dole who has a score of 0.755 (Palin is actually 0.752)? Not very well (the typical score has a standard error of about 0.11). But we can be almost certain that Palin is far better looking than her opponent Biden. He scored a relatively measly 0.38 (although Biden is still doing better than the mean score of 0.22).
Now, before you express your disgust at having read this far just to have somebody tell you the shocking news that Sarah Palin has a better face than Joe Biden, there is some less obvious information out there. How does Palin compare to other names that were floated for the GOP Vice Presidential nomination? As far as faces go, Palin was a far better choice than the often mentioned Minnesota Governor Tim Pawlenty (-0.14). However, she can't quite stack up with the even more oft mentioned, and acclaimed as handsome, Mitt Romney (0.93). Of course, there was some quiet chatter about John McCain choosing Texas Senator Kay Baily Hutchison, presumably to court the female vote, as some claim is the reason he chose Palin. When it comes to faces, Palin was the better choice, as Hutchison only came in at 0.56.
How does Palin do against the rejected Democratic VP possibilities? Quite well against Governors Tim Kaine of Virginia (0.11) and Kathleen Sebelius of Indiana (0.14). Even better against Senator Chris Dodd of Connecticut (0.05). However, had Barack Obama chosen Evan Bayh of Indiana, he might have stood a chance, facially, against Palin, with a score of 0.63.
In fact, we found across all candidates, for all offices, Republicans, on average, have more competent faces than Democrats. Take that for what you well, but incumbents tend to have more competent faces and Republicans were in the majority for most of the years covered in our survey.
Now you say, what about the politicians we really want to hear about? McCain? Obama? Hillary Clinton? I have not avoided mentioning them for the purposes of suspense. I have no scores for them. In doing the ratings, we removed the photos of candidates that the students were likely to recognize. The idea is to have the students make decisions based on appearance alone, not other opinions they might have about the person. Romney stayed in the sample because in spring 2007, when we conducted the survey, Romney was a nobody - at least to college students - the overwhelming majority of whom could not even identify a photo of their own, long serving Member of Congress, Henry Waxman. Romney, however, was not as much of a nobody as Sarah Palin, even though one of my colleagues now recalls being struck by her beauty when we were preparing the photos for the survey. It seems then, that in getting her face rated, the half-term Governor from Alaska, benefited from her obscurity. The same obscurity that some have argued helped her in becoming a finalist for the position of "heartbeat away" from the most powerful post in the world.
Of course, I did start by giving the "hottest VP candidate ever" only a 0.75. To avoid any accusations of bias on my blog, I should repeat that she is in the top 5% of politicians we rated. The 95th percentile ain't bad. I don't know if it will satisfy her most ardent supporters, but this means we can give her a 9.5 out of 10 (scientifically speaking).
On-topic comments that add to the discussion are welcome. Please respect each other and the forum by using your real name and a civil tone. Spam and comments judged by UCLA to be libelous, offensive or abusive may be deleted without notice
Comments
Ryan Enos says:
Concerned - you might take a look at my blog posting about Presidential IQ. Unfortunately, I have not expertise in assessing IQ, so I will not attempt to do so for Palin. The study mentioned here had not focus on Palin - we just happened to have a rating for her - as part of a larger concern for how candidate traits affect elections. With all the talk about Palin's looks - it became an interesting thing to write about.
Concerned says:
I am kind of wondering why Ms. Palin is being analyzed in this fashion. It would appear that voter's reaction to looks is important, but I would rather like to hear from you guys how Palin is doing IQ wise. How could that be assessed (maybe from her interviews where she did not have scripted answers?) and what value would we come up with? I have a feeling it will not be a 9.5...
Ryan Enos says:
Jason,
"better looking" is me playing fast and loose with terminology. In actuality, what I mean is that a face has scored higher when the judgments of naive respondents in one to one comparisons of 'facial competence' are scaled on one-dimension. Respondents were not overtly informed that they were looking at politicians faces, but most probably guessed. The exact phrasing of the question that they were asked was "which face looks more competent"?
Whether Dole is better looking than Palin - I suppose is all in the eye of the beholder, no?
Jason says:
When you say "better looking" throughout the article, do you mean more "facially competent" every time? Does "facially competent" mean that they have a face that would lead you to believe they are competent [as a politician]? Or just competent generally?
I ask that because I would hold any study suspect that found Elizabeth Dole to be "better looking" than Palin. Dole does look like a more competent politician though, imho.
Ryan Enos says:
Tom - that is an interesting question - I can't exactly answer whether we value womens' looks more than those of men - although I will say that I did some quick analysis it did not seem to reveal an interactive relationship between facial competence and gender and their effect on vote share - which means that facial competence appears to have the same effect on vote share for both women and men. In regards to whether women win 70% of the time too. I looked at the numbers for the Senate only, this is between 1992 and 2006, and, in this subset facial competence correctly predicts the winner for men 74% of the time and for women 77%. I made a graphic of this, in case you're interested, you can see it here: http://renos.bol.ucla.edu/gender_predict.pdf
Ryan Enos says:
Trevor -
right, this method does assume transitivity. If you want to read about the method, you might want to look at the appendix of the paper that inspired this blog at http://renos.bol.ucla.edu/AtkinsonEnosHill.pdf .
I wish there were transitivity in college football, it is the only way I can get Cal to win a national championship.
Anyway, the question of transitivity in the data is interesting. We have extracted eigenvalues of the matrix of scale distances between candidates and there is probably more than one dimension on which these candidates are being compared. This is inevitable given the different way humans will respond to the prompt of 'which is more competent'. In this sense, it is possible that a particular candidate will out-perform another candidate on one dimension, although not on another. However, our scaling restricts the estimation to a single dimension, so it seems that the collapsing of the dimensions into a single scale will provide for transitivity on that single scale.
Tom says:
You mention 70% of the time winners had the higher score. Is it the same broken up between men and women?
Do we value women's looks more than men's?
Trevor Stone says:
The method also assumes transitivity. College football victories often are not transitive; teams often beat everyone in the conference except a rival who only won a few games. It's possible that candidate X's features in some way "counter" candidate Y's features, even though Y beats most people who beat X.
Do the data support a transitive assumption?
Ryan Enos says:
Trevon - I am so glad you liked the article and I hope your interest in politics continues to grow and that you can put it towards a productive endeavor. I should be gone by the time you get here - but make sure you look up some of the professors on this blog when you get to UCLA. Good luck.
Trevon Fambro says:
Dr. (soon) Ryan Enos,
I am a fourteen year old high schooler who loves politics, and I must say, this article--while trivial nonetheless--intrigues me and incites my politcal interest even more. I plan on atteneding UCLA as an undergraduate, and now after reading this, my decision is made.
Thank you, and keep the articles coming!
Ryan Enos says:
LW:
That is interesting:
a couple of thoughts: one is that I think it is commonly accepted that people like Roosevelt (disabled) and Lincoln (ugly) just couldn't be elected today. I don't actually know if that true, but it might be - in that sense, politics is certainly more visual. However, that was the case before the 24 hour news cycle. I think Kennedy had appearance as part of his toolkit in 1960 and tv was a very different animal then.
It is also certainly true that race and gender play a role in how people vote. But this was also true long before the 24 hour news cycle. Race, or the spectre of race has had a role in elections long before visual images were prominent. Rumors of miscegination go back at least as far as Jeffersonians (ironically) passing rumors about Hamilton.
Now, we don't know how voters react to a Black candidate for president, but we do know that racial attitudes affect voter's political opinions, so I think it is clear that race will affect how people vote. How much is the big question. My guess, in a close election, enough to make a difference - so Democrats better make sure it is not close. Gender might be similar.
LW says:
Can't we make the arguement that since the introduction of television that politics has become more and more visual? As we look at the 24 hour news cycle and now so much politics discussed online should we even be suprised that a candidates eye glasses are selling out? Even a month Palin before Palin was in the picture everyone was all a gossip with the photos of Obama on the beach.
This all leads me to believe that race and gender has to be playing a role in how people are going to vote. Thoughts?
Brian Law says:
Great work!
Ryan Enos says:
tlucian - that actually isn't true, although, yes,that would indeed be weird. The Speaker of the House does not necessarily ascend to the VP, like she does to the Presidency. The 25th Amendment provides that if the VP becomes vacant, like it would if the VP ascended to the Presidency, then the President will appoint a VP, subject to approval by both Houses of Congress.
That has only happened once since that rule was put in place, when Ford nominated Rockefeller as VP.
tlucian says:
Why isn't anyone pointing out that if McCain kicks the bucket Palin's vice president will be Nancy Pelosi. Yes, this evolving drama can get even more weird!
