Do Voters Discriminate Against Obese Candidates?

by John Sides on September 7, 2011 · 15 comments

in Campaigns and elections

Jon Corzine seemed to hope so.  A Wall Street Journal reporter thought the opposite, trolling message boards in 2008 for people who resented Obama for not being fat.  Here is some evidence, courtesy of a new paper presented at the 2011 APSA meeting.  The authors are Beth Miller, Jennifer Lundgren, Diane Filion, and Lauren Thompson.

Participants in the study were randomized to evaluate candidates who were male or female and obese or not obese.  Obesity was conveyed through digitally altering photographs, as in the example below:

Miller and colleagues find that obese male candidates were actually evaluated more positively than non-obese male candidates.  Obese female candidates, however, were evaluated essentially the same or less positively than non-obese female candidates, depending on the measure of evaluation.  Moreover, obese female candidates also elicited a stronger negative emotional reaction, as measured by the participants’ startle reflex.

To be sure, this is a small study whose participants were exclusively college students.  Nevertheless, these findings dovetail with previous work by Miller and Lundgren, who also find an “obesity penalty” for women candidates but not men.   In the current paper, they offer a bit of speculation about this:

This finding may stem from the pressure in western society on women to be thin and men to have greater muscle mass.

There’s lots more in the paper to chew on.

{ 15 comments }

AGH September 7, 2011 at 12:32 pm

The modified image doesn’t really look obese – the suit is in basically the same position. What they really might be evaluating here is a fat head.

Moby Hick September 7, 2011 at 1:15 pm

Unless you can add fat to the sides and top of your head, I agree that the modified images doesn’t look obese. It looks ‘shopped.

John Sides September 7, 2011 at 1:31 pm

Maybe yes, maybe no. But participants didn’t see these pictures side-by-side. If you confronted the righthand picture by itself, I doubt that you would think it was modified.

Moby Hick September 7, 2011 at 2:01 pm

Maybe. But if the eyes get any farther from the sides of the face, I’m guessing I’d think of a chromosomal problem.

Patrick September 8, 2011 at 9:19 am

Err… the picture on the right does not look like what most Americans would classify as “obese”, rather the impression it gives is of a burly ex-varsity or high school football player, and the haircut is vaguely military or law enforcement. I would respectfully submit that the ex-military/ex-law enforcement ex-football player would win in any US election match-up.

Patrick September 8, 2011 at 9:36 am

I guess this is only an example of one of the pictures used. Perhaps other combinations showed a morphed example that was more obviously “obese”, and not just burly or hefty.

Sara September 8, 2011 at 10:03 am

Small as they are, these findings, nonetheless, are revealing and could have lead to invigorating discussions about, among other things, subconscious biases against fat people, in general, and fat women, in particular. Instead, the posts have all been reviews about a morphed photo…

John Sides September 8, 2011 at 11:56 am

My thoughts exactly.

matt w September 8, 2011 at 12:26 pm

In this case I think the discussion about the photo is relevant — because if there are problems with the photos used in the study, then the findings are not in fact revealing (because they’re not really findings). We want our discussion to have a secure basis, don’t we? (In fact, I clicked over from the Washington Monthly to see if the commenters were talking about the problems with these photos.)

For what it’s worth, I had the exact same reaction as Patrick; the picture on the right looks burly rather than obese to me. In fact the picture on the left looks more skinny and goofy; the wrinkles on the neck are more prominent, and the ears stick out more. If this is typical of the photograph pairs they used for male candidates, then it’s not too surprising or revealing to me that they didn’t find a negative bias against “obese” candidates. Though perhaps my thoughts reflect societal pressure on men to have greater muscle mass.

For what it’s worth, my prior probabilities would’ve been that obesity would’ve been something of a problem for men and a huge problem for women, because of pressure on women to be thin (and generally, greater attention paid to women’s appearance). And this doesn’t do much to shift my priors, without a better look at the pictures that were used.

John Sides September 8, 2011 at 1:50 pm

Again, this is just an example of one set of photos out of 4. Let’s not get fixated on it.

And, matt w, the study finds that obesity is a *benefit* for men not “something of a problem.” So it might shift your priors just a little more than you indicate.

matt w September 8, 2011 at 2:57 pm

Well, I’m saying it doesn’t shift my priors* because I’m still skeptical as to whether the results were an artifact of the particular way they photoshopped the pictures — and if there were only four pairs, it seems like it might be especially prone to artifacts. And as I said, I think that’s a legitimate topic for discussion; if we had doubts about the wording of a survey question, we’d want to discuss that. (I did take a quick look at the paper, but that was the only pair of pictures they reproduce in the paper itself.)

I do think that it’s interesting that they find obesity to be a benefit, and posit increased muscle mass as an explanation. One other issue this raises is the utility of most definitions of obesity. Chris Christie and Jerome Bettis may have similar heights and weights (or not), but Christie is much more like the stereotype of an obese man. The “obese” picture there looks more like a Bettis type to me; his chin is much more defined, like Bettis’s.

In short, I’d like to see these results reproduced before I place full credence in them.

*My Bayesian friends would probably disown me for saying that — what I mean is, my posteriors aren’t much different than my priors.

John Sides September 8, 2011 at 3:45 pm

Well, if nothing else, I would love to see a study that refers to Jerome Bettis.

Moby Hick September 8, 2011 at 10:45 am

Reducing ill-structured problems with unknown solution sets to better structured problems is a common decisions making process.

Aaron September 8, 2011 at 5:53 pm

I think the obese pic makes him look younger,(fat, as does photoshop, can help push out wrinkles) which may be larger factor. I think a good follow up poll to this would be to ask how old they think each person is in the pics, (pretty sure there is a known inherent bias vs older candidates).

spookiewon October 15, 2011 at 2:23 am

And, as pointed out before, fat (or photoshop) turned out, in this small study, to be a BENEFIT to male candidates. You are saying age is preferred over youth, and fat (or photoshopped) looks younger than thin. The study doesn’t support this.

Comments on this entry are closed.

Previous post:

Next post: