Monday, October 22, 2012

Today's Fun Research Popularization: Conservative Women Are More Beautiful!

Yup.   The study, about the femininity or masculinity of  politicians' faces, was publicized in September.  It's not in print yet, as far as I can tell, though a very kind person sent me the manuscript.

That's the first bad trend in the way these studies are discussed:  Do the discussion before the study is available for reading and criticism.  That way nobody can tell if it makes any sense!  It's like telling who won a baseball game without letting people actually watch the game.

The second slightly odd aspect in popularizing this particular study is that its lead author, Colleen M. Carpinella,  is a UCLA graduate student in psychology.  We don't usually popularize studies by people who haven't even gotten their PhD yet.  I must stress that this is not a criticism of the study or of the researchers.  Work done by PhD students can be valuable and worth looking at, but mostly newspapers and websites don't do that.

Except for certain titillating topics, such as the idea that Republican women might be better looking than Democratic women.  Now, note that the study DOES NOT SPEAK OF THAT at all.

But the popularizations do.  Here's a representative sample of the headlines:*

The Daily Caller:

Study: Female GOP politicians are better looking than liberal politicians [SLIDESHOW]

World Net Weekly:

Hubba, Hubba!  GOP women better looking?

The Examiner:

She's beautiful...does that mean she's a Republican?

Get the idea?  This study was about beauty.

Except it was not.  That word doesn't appear in the study at all.  Or studies, because the researchers carried out two separate studies.   I'm going to discuss the studies separately because they provoke different concerns.

The first study tries to measure the femininity vs. masculinity of the features of the politicians in the 111th US House of Representatives by feeding the photographs of all those 434 members into a program which analyzes the facial features for their sex-typicality.  Note that sex-typicality is not the same thing as beauty or handsomeness.   From the article: 
We downloaded photographs from each politician's government website and coded for sex and political party. We imported each image individually into FaceGen Modeler using the Photo Fit Tool (Blanz & Vetter, 1999), and we measured each face's sex-typicality (i.e., masculinity for men and femininity for women) using the Gender Morph tool.1 Theoretical values ranged from −40 (highly male-typed) to +40 (highly female-typed). We converted this to a common scale for men and women, reflecting the objective level of sex-typical facial cues. Thus, positive values indicated sex-typical characteristics (i.e., masculine men and feminine women); negative values indicated sex-atypical characteristics (i.e., feminine men and masculine women). 

Why did they do this?  Because the hypothesis in the article is that Republican women would be more sex-typical than Democratic women.  The Republican Party supports traditional gender roles and its supporters might require more sex-typical looks from the women who want to exert an atypical leadership role in that party.   To counteract for the latter, perhaps.

The Democratic Party is less invested in traditional gender roles and therefore can allow more sex-atypical faces on their politicians.  Because its supporters don't care about rigid definitions of masculinity and femininity and so on.

What did the first study find?  That the Republicans and Democrats overall did not differ in sex-typicality but that the Republican women were the most sex-typical of all politicians. You may already have figured out what that combination must mean about the Republican men.  Yup, they were less sex-typical than the Democratic men.

If we translated THAT into those weird popularization headlines, how would they look?  "Democratic men are Hotties!, Republicans As Ugly As Elephant's Anuses?"

I think you are getting my point here, which is that first certain studies are picked for closer examination and then they are closely examined in one direction only.

I'm not familiar with how the FaceGen Modeler works.  Does it allow for the fact that "sex-typical" facial features vary by racial and ethnic group? ** If the Democratic Party has more racial and ethnic variety, not controlling for ethnicity and race could distort the results.

The second study in the article tries to find out whether sex-typicality vs. sex-atypicality of the politicians' faces could be used to predict the political party that politician belongs to.  This study is considerably weaker in my view than the first study, because it uses a group of UCLA undergraduates (120 total, out of which 35 were men) for the prediction part.

For instance, I don't believe in the argument that American undergraduates would be so removed from day-to-day American politics that they wouldn't already know the party affiliation of quite a few people in the pictures.  And this would affect the female politicians more because they are fewer and therefore somehow more memorable.  The most memorable of all the sex-party groups would be Republican women politicians as there are not very many of them.  Who doesn't recognize the features of Michelle Bachmann, for instance?

Likewise, the way one is dressed, made-up and coiffed can affect the findings in the second study because none of these were controlled for and to my untutored eye there are pretty big differences between the Republican and Democratic women politician on those issues. The researchers point out that the first study didn't rely on those indicators which is true.  But the second party could have relied on them.

So let's recap:  A study found that Republican female politicians have the most sex-typical faces of all US members of the 111th House of Representatives.  Republican men are less sex-typical than Democratic men.  The researchers believe that the greater sex-typicality of Republican women has to do with what kinds of politicians are allowed to wield power in each party.  From the article:

We predicted that judgments of political party affiliation would rely on the sex-typicality of facial cues. Our prediction was guided by the gendered nature of the liberal-conservative continuum, in both policy advocacy and gender attitudes.
Across democratic political systems, women's historic realignment with more liberal politics (Inglehart & Norris, 2000) reflects shifts in political parties' values. In the U.S., for example, the Democratic Party is associated with socially liberal policies that aim to diminish gender disparities (e.g., women's rights, abortion rights); the Republican Party is associated with socially conservative policy issues that tend to bolster traditional sex roles (e.g., military spending, national defense; Winter, 2010). These policy platforms are manifest in each party's image. Consequently, politicians may exhibit characteristics that reflect these values.
Gender attitudes also differ reliably by political ideology. Conservatives, in particular, encourage adherence to traditional gender roles (Lye & Waldron, 1997). Thus, communal and feminine women are highly regarded. Consequently, Republican women may be uniquely prone to exhibit sex-typical characteristics.

And how was all this popularized in several places:  Republican women are hotter!

*Science Daily's headline:  The GOP Has a Feminine Face, Study Finds, is better than the others because it's closer to truth.  The Republican women are more sex-typical, the Republican men are less sex-typical, so the overall pulls the party towards what the study calls the "feminine" end of the spectrum.  But it's tricky to use the term "feminine" without pointing out that all this is about sex-typical features, not feminine in the sense of pink-fluffy-rabbits-dancing-among-roses.

**For instance, does the study compare a Latina's facial features to the average facial features of all Latinas?  Or what?  This matters if the parties have different ethnic/racial percentages.

The study:

Colleen M. Carpinella, Kerri L. Johnson "Appearance-based politics: Sex-typed facial cues communicate political affiliation," forthcoming in the Journal of Experimental Social Psychology.