Dating a Stereotype (Getting to Know You – Not)

January 26, 2014
Posted by Jay Livingston

The useful thing about cultural stereotypes is that to some degree, they’re often accurate – a convenient shorthand.

(Sorry about the lousy sound, but this was the best clip I could find, and it does have subtitles.)

The Woody Allen character learns her name, her thesis topic . . . and all the rest follows. Note also that Allison (my neighbor Carol Kane) doesn’t say that Woody is incorrect.  

What reminded me of “Annie Hall” (the Annie character too is a cultural stereotype) was this:

(Click on an image for a larger view.)

 The three axes are percentages: 

• Width - seeking a one-night stand 
• Depth - had same-gender sex
• Height - say God is important to them)

The graph is a typology of women – women on OK Cupid, the dating site founded by four Harvard math majors.*  The graph appears in this Wired article about Chris McKinlay, a 35-year old guy who took nerditude to the n+1^th  degree, creating bots to Hoover up data on responses to the hundreds of questions OK Cupidians can answer.** Eventually, he had six million answers from 20,000 women.  But how to analyze the big data?

A modified Bell Labs algorithm called K-Modes. First used in 1998 to analyze diseased soybean crops, it takes categorical data and clumps it like the colored wax swimming in a Lava Lamp. With some fine-tuning he could adjust the viscosity of the results, thinning it into a slick or coagulating it into a single, solid glob.

He played with the dial and found a natural resting point where the 20,000 women clumped into seven statistically distinct clusters based on their questions and answers.

The names of the clusters –Tattoo, Dog, etc. – are basically cultural stereotypes. 

In the younger cluster, the women invariably had two or more tattoos and lived on the east side of Los Angeles. In the other, a disproportionate number owned midsize dogs that they adored.

The article also has graphics on how the seven stereotypes differed from one another in four areas. (The “Green” tag is not political; Greens are merely recent arrivals at OK Cupid. They are also the most sexually adventurous. As the placement of the green ball on the graph shows, 50-60% would be comfortable with a one-night hook-up, and 40-50% have had same gender sex. Not surprisingly, they do not find God to be an important part of their lives.)

The stereotypes, based on clusters, were very useful for finding, well, clusters. McKinlay tailored his two OK Cupid profiles to maximize his chances of getting a response  so he would do better than the six OK Cupid dates he’d managed to get in the previous nine months. He did.  His scientifically customized profile was getting 400 hits a day.

Cultural stereotypes may get you into the right room (and save you a lot of time wandering into wrong rooms), but they’re no guarantee of compatibility with an actual person. McKinlay went on more than 50 first dates – a big improvement over six in nine months – but only a handful of these led to a second date, and none went further. 

Given this data, most of us would figure that it was time to start thinking about our interpersonal skills or perhaps our grooming and hygeine. Wired says merely that McKinlay “had to question his calculations.” 

But finally, something clicked, and the story seems to be heading towards a happy ending – a year-long relationship, some of it long-distance since the woman is on a one-year fellowship in Qatar.

on one of their daily Skype calls . . . McKinlay pulls out a diamond ring and holds it up to the webcam. She says yes.

----------------------------
* Previous posts on OK Cupid are here and here. 

** The technical details:

he set up 12 fake OkCupid accounts and wrote a Python script to manage them. The script would search his target demographic (heterosexual and bisexual women between the ages of 25 and 45), visit their pages, and scrape their profiles for every scrap of available information: ethnicity, height, smoker or nonsmoker, astrological sign—“all that crap,” he says.

The phrase “Python script” of course poses a tremendous challenge for me to avoid the obvious joke – surely one made so often that it has long been an ex-joke. 

Another Opening, Another Show

January 22, 2014
Posted by Jay Livingston

My first class of the semester is tomorrow.  I’ll begin, as usual, with Durkheim and suicide and rates of behavior as social facts.  Rates, I’ll remind the students, are made up of individual cases.  That’s basic skills math.  But those rates, unlike the individual cases, have a strange constancy. If 42,000 people in the US killed themselves last year, the number for this year will be close to 42,000.  Why?   It can’t be the same people.

I’ve been teaching this for years, yet I still find it eerie.

More important, I will tell the class, the ideas that explain individual cases don’t work so well in explaining the rate that those cases add up to. It’s very likely that people who commit suicide are less happy than those who don’t. But does happiness explain suicide rates?

International suicide rates are not hard to find.  Now, fortunately, we have international data on happiness – The World Happiness Report.

So I put together a simple scatterplot of European countries (I added the US since I thought the locals here might be curious to see where we stand.)

The overall correlation is about –0.24. More happiness, less suicide, but only slightly less. Sad and suicidal Hungary is the hero for the happiness hypothesis.  Remove Hungary and the correlation drops to –0.15. As for the rest, those northern, social-democratic countries (Iceland, Finland, Sweden, Netherlands, in addition to the ones identified in the chart) may be the happiest, but their rates of suicide are not noticeably lower than those of less happy countries like Bulgaria.  (What’s up with Bulgaria anyway?)

OK. Now that we’ve put those individual-level ideas in their place, let’s spend the next couple of months doing sociology.

Correlation and Causation: Marriage, Poverty, and Teeth

January 15, 2014
Posted by Jay Livingston

Is dentistry destiny?

A couple of days ago, I commented (here) on the idea that marriage was the solution to poverty.  Or as the Heritage Foundation (and Sen. Rubio) put it,

Being raised in a married family reduced  child’s probability of living in poverty by about 82 percent.

The evidence for this assertion was a comparison of two-parent and single-parent families. Sure enough, a higher percent of single-parent families were poor. That was enough for PolitiFact to give Rubio a “Mostly True.” And now The Wall Street Journal has given Bush’s press secretary Ari Fleischer space to repeat this idea (“How to Fight Income Inequality: Get Married” - here).

Rubio, Heritage, Fleischer, and PolitiFact are committing a fallacy most undergraduates would see through in two seconds – mistaking correlation for causation. 

Here’s an analogy that comes via the Christian Science Monitor (here – it first appeared in October, but I discovered only yesterday):

“What is your social class? Take our quiz to find out!”

The quiz has thirty items, including this one

16. Have you visited a dentist in the past year?

A “Yes” answer counted towards the upper end of the social class scale. 

people with more than a high school education were twice as likely to have visited the dentist in the past year. Those living below the poverty line or without a high school education were also twice as likely to be edentulous, or toothless.

Or as Heritage-Rubio would put it, “Visiting the dentist once a year reduced a person’s probability of being poor by 50%.” 

Using the same conservative perspective, we can easily see the logic of the dentistry-poverty connection and its implication for policy.  People with bad teeth or no teeth wind up with bad jobs or no jobs. They are not attractive as potential employees. Because of their poor personal decisions regarding dental care, they suffer economically. If only they would visit the dentist annually, they would almost certainly rise from poverty. Needless to say, the government should not do anything directly to alleviate their poverty or dental care.  These are matters of personal virtue, and the government’s role should be only to exhort them to visit the dentist regularly.

That almost sounds reasonable. But the reverse causation is so much more likely. It’s not that having bad teeth causes poverty. It’s that taking yourself and your kids to the dentist regularly costs money – something poor people don’t have a lot of.

The dentistry-marriage analogy isn’t perfect, but it does illustrate the fallacy of assuming causation. It also points to something in the real world. The correlation between single parenthood and poverty is not automatic. It depends on government policies. In countries that provide low-cost childcare, medical care, and other benefits and services, single parents and their children will not suffer economically as they might under more punitive policies. Many countries have seen large increases in unwed parenthood – much greater than in the US – and their rates of single parenthood are greater.  (The graph is part of this CDC report.)

(Click on the chart for a larger view.)

But how are children in these countries faring?  The comparison of poverty rates shows a negative correlation – the higher the rate of unwed parenthood, the lower the rate of child poverty.*

(Child poverty data are from this OECD report.)

The question is not whether marriage and regular dental check-ups and other matters of personal virtue enhance economic prospects for parents and their children.**  The question is what governments should do for children who made the mistake of choosing parents who were not financially well-off, educated, healthy, and virtuous. 
---------------------

* The OECD measure of poverty is relative rather than absolute – the percentage of children in families with incomes less than 50% of the national median.  The measure could more accurately be termed a measure of inequality. However, in surveys in the US, when people are asked what they think the poverty line should be – i.e., what’s the minimum amount a family around here needs to just get by – the answer is usually a number that is about 50-55% of the median income.  So the OECD number does reflect widely-held ideas about poverty.

** The Oklahoma Marriage Initiative consisted of “federally-funded relationship skills training programs to promote marriage among unmarried parents.”   Philip Cohen at the Family Inequality blog outsourced the analysis of the Intiative to Kristi Williams, who writes (here) of these programs: “The conclusion: They have failed spectacularly.”

Location, Location, Location

January 14, 2014
Posted by Jay Livingston

The chemical spill in West Virginia should be a much more important story than Chris Christie and the George Washington Bridge. But it’s the Bridge that’s getting far more attention in the media. 

Anne Marie Cox has a good piece in the Guardian (here) about “how it came to be that Bridgegate continues to attract punditry while West Virginia only generates the kind of sympathetic-if-distant coverage we usually grant far-off and not too devastating natural disasters.”

In West Virginia, there are 300,000 people without useable water, and an unknown number who may fall ill because the warning to avoid the tainted supply came seven hours after the leak was discovered – and perhaps weeks after it happened. (Neighbors of the plant have told reporters they detected the chemical’s odor in December.)

Surely, that’s more important than four days of traffic jams, which, truth be told, are hardly a strange and new horror for New York and New Jersey drivers.

Cox has several explanations for the disproportionate weight given to the Christie story. Not only might Christie be president in a couple of years, but he’s known. He’s a political celebrity.* And for some reason, stories about the personal deeds and misdeeds of celebrities are newsworthy. Apparently we prefer a story about personalities rather than about policy (especially policy that involves science, especially environmental science). 

Cox lists other reasons, but the one I think is most telling is geography.**

It is taking place in the literal backyard of most national political reporters. It has very little to do with policy, or numbers, or science.

In the old days – with no satellite transmission, with no Internet –  stories from New York, Washington, and perhaps a few other places dominated the news because that’s where the news business was located. Stories from other places were more expensive to produce and transmit.  Film would have to be flown from the hinterlands to production studios in New York. 

Today, remote stories do not run up costs. And in many ways the chemical spill should make for better news – the visuals are potential more striking, the potential interviews with the plain folks who are affected, the corporate baddies (it doesn’t get much better than “Freedom Industries”), the political influence, etc.

But it’s not just the cost. The sophisticated, cosmopolitan people who bring us the news turn out to be just as provincial in their own way as are the rubes they tacitly disdain.  If the 4-methylcyclohexane-methanol had been polluting the Hudson or the Potomac, it would have been a national story.  As it is, the unstated message in the media coverage is, “Forget it; it’s only West Virginia.”

-----------------
* Christie’s celebrity status is not an accident. One of the nuggets that the investigation has unearthed is that in choosing an ad campaign for the state to show its miraculous post-Sandy recovery, Christie chose a $4.7 million ad campaign over one that cost about half that much.  The pricier PR job Christie chose gave much greater prominence to Christie himself.

** Social scientists and media experts who know more than I do about how news is made must have written about this, but I have not come across any posts on these two stories.