Time Thinks It Can Predict Your Politics Using Tired Profiling and Stereotypes

January 11th, 2014 10:44 AM

Jonathan Haidt and Chris Wilson at Time.com claim that "your preferences in dogs, Internet browsers, and 10 other items predict your partisan leanings." So a left-leaning mag which is philosophically united with the crowd that insists that we must be equal opportunity friskers of 4 year-old children and 80 year-old grandmothers at airports because "we shouldn't profile" has no trouble profiling people as conservative or liberal based on the answers to 12 inane questions.

Conservative Rush Limbaugh — cat lover, rebellious teen, and Mac user — will certainly be amused at the questions in the survey, the authors' breezy contentions about what their answers supposedly mean, and the other assertions they make.


There are two factors which skew the survey into virtual irrelevance.

First, each agree/disagree question forces you to at least slightly agree or disagree, i.e., there is no "I don't care" equivalent). The fact is that many people, to name just one example, don't have a preference between cats and dogs, but are forced to indicate one if they answer the question.

Additionally, based on the dialog box warning, the survey will run as long as you answer only five of the 12 questions:

TimeDialogBoxWarning0114

("science.time.com"? How stereotypically pretentious.)

So Haidt and Wilson are profiling participants based on coerced preference expressions and potentially very limited information.

Here is the pair's introduction, followed by the survey questions (HT Twitchy; bolds are mine throughout this post):

Can TIME Predict Your Politics?
See how your preferences in dogs, Internet browsers, and 10 other items predict your partisan leanings.

Social scientists find many questions about values and lifestyle that have no obvious connection to politics can be used to predict a person’s ideology. Even a decision as trivial as which browser you’re using to read this article is imbued with clues about your personality. Are you on a Mac or PC? Did you use the default program that came with the computer or install a new one?

In the following interactive, we put together 12 questions that have a statistical correlation to a person’s political leanings, even if the questions themselves are seemingly apolitical.

TimePreferCatsToDogs

Survey questions:
1. I prefer cats to dogs.
2. I prefer watching documentaries to action/adventure movies.
3. Respect for authority is something all children need to learn.
4. I keep my desk and other workspaces very neat and organized.
5. I believe that self-expression is more important than self-control.
6. If I heard that a new restaurant in my neighborhood blended the cuisines of two very different cultures, that would make me want to try it.
7. My government should treat lives of its citizens as being much more valuable than lives in other countries.
8. If I were married or in a serious dating relationship, I would think it is perfectly OK for my partner to look at erotic or pornographic pictures or videos, by himself/herself.
9. The Internet browser I most often use is: Chrome; Firefox; Internet Explorer; Safari; Other/Don't know.
10. I wish the world did not have nations or borders and we were all part of one big group.
11.If I were to visit New York City, I would rather go to Times Square than the Metropolitan Museum of Art.
12. I am proud of my country’s history.

Now get a load of the stereotyping imbued in the authors' elaboration:

We created this survey by drawing on several sources. Research by Sam Gosling, at the University of Texas, has found that liberals generally score higher than conservatives on the trait of “openness to experience.” They are more likely to seek out new experiences (such as fusion cuisine), choose to watch documentaries, or enjoy art museums. They have less conventional notions of what is proper in a romantic relationship, so solo pornography consumption is OK. Conservatives are more likely to stick with what is familiar, what is tried and true. Hence, they are more likely to use a PC than a Mac and are more likely to stick with that PC’s default browser, Internet Explorer. Conservatives score higher than liberals on the trait of conscientiousness. They are more organized (neat desks), punctual, and self-controlled (rather than emphasizing self-expression).

We also drew on several surveys from YourMorals.org for data about how values correspond to politics. Conservatives, for example, tend to value respect for authority and group loyalty more than liberals do. Conservatives, therefore, typically show less ambivalence about American history and have a stronger preference for dogs, who are more loyal and obedient than cats.

Then they predictably (sorry for the typecasting, but it works) claim victory, after changing the playing field:

Update: Jan. 10, 2014, 11:00 a.m.: It works! Our analysis of 17,000 responses from readers who chose to report their actual ideology found a strong correlation (r=0.604, for those of you keeping score) between a person’s self-reported ideology and the output of the quiz. This is a particularly strong correlation given the wide degree of personal variation in taste that is intrinsic to this sort of research.

The biggest weakness we discovered is that the results from our survey were less distributed across the spectrum than the figures for people’s self-reported ideologies. A person who reported themselves as “very liberal” or “very conservative” tended to receive scores that were artificially close to the center. As of this update, the quiz now employs a basic statistical correction to more accurately reflect the extremity of one’s politics.

Translation: We moved the goalposts to make ourselves look good.

Exit point: As seen at the beginning, the authors apparently don't understand or don't care about the difference between "partisan leanings" (Dem, GOP, Independent) and ideology (liberal, conservative, libertarian). How predictably ignorant.

Cross-posted at BizzyBlog.com.