Correlation is Not Causation

Time Alive Correlates with Foot Problems

2008-04-21T20:47:00.002-04:00

at least among people in the United States, a region that enjoys the most advanced footwear technology around. So are foot injuries an indication that we need (even) better shoes, or of something else entirely? To find out, read You Walk Wrong. It's another good read that will have you pondering which came first, the bunion or the Pradas?

The joy of prose

2008-04-18T06:35:00.003-04:00

I just read an absolutely delectable yet fulfilling article on nothing less than elevators. Nick Paumgarten writes in the New Yorker about the history and influence of elevators. The piece is wondrously well-researched, thoughtfully written, and pleasantly playful. It really was a joy to read, so get to it (it's kind of long at 8 pages, but oh-so-worth every word).

And my excuse for blogging it here?

Two things make tall buildings possible: the steel frame and the safety elevator. The elevator, underrated and overlooked, is to the city what paper is to reading and gunpowder is to war. Without the elevator, there would be no verticality, no density, and, without these, none of the urban advantages of energy efficiency, economic productivity, and cultural ferment.

That's correlation and causation, by golly.

While we like to focus on questionable conclusions based on unproven correlations around here,

2006-04-03T08:54:00.000-04:00

any statistical monkeybusiness is fair game.

For the cave dwellers, let me summarize the situation: numerous studies have indicated that moderate drinking is beneficial to one's health. For years now, those of us who are teetotallers have had to endure jibes and derision, but in our hearts we knew that one day we would be vindicated. That day has come.

Now a new study (a meta study - a study of studies, no less!) has contradicted prior claims, but it has done so by specifically pointing out which error the prior studies had committed that led them to make such crazy claims:

The common error was to lump into the group of "abstainers" people who were once drinkers but had quit.

You know, I actually kind of like saying "I told you so."

What about when correlation really is causation?

2005-12-19T09:24:00.000-05:00

Check out this article in Reason about a recent World Bank Study (pdf).

Basically, the wealth of nations correlates very highly to just two factors: education and rule of law. This is, of course, common sense (or it should be), but it's nice to have a really cogent data set to chew on.

Now, as the article's author, Ronald Bailey, asks:

How can the people of the developing world rid themselves of the kleptocrats who loot their countries and keep them poor?

I sure wish I knew.

Paul Krugman: less controversial, or just less influential?

2005-11-11T17:35:00.000-05:00

This post on Paul Krugman and TimesSelect is more about influence than statistics, but this paragraph made it a winner in our book:

Alternative explanations? They exist. One never-forgotten lesson from stats class twenty years ago is that correlation ain’t causation. Perhaps Krugman just hasn’t written a controversial column since September (but … we can’t tell!). Also, absent statistical analysis, we could be seeing patterns where there really aren’t any (but we don’t think so).

True, true.

Not to get off topic, but having a renowned economist who seems to have lost touch with what it means to be a good economist [too many examples to link to] (and btw - how can there even be lefty economists? Of course, in a similar vein, not all lefties have lost touch with reality) lose a little influence isn't such a bad thing in my book. {Sorry for that nasty lump of parentheticals.}

I can only hope that this corporate 'social consciousness' is a marketing ploy

2005-10-25T14:36:00.000-04:00

In an article in today's Wall Street Journal titled Wal-Mart Urges
Congress to Raise Minimum Wage (subscription required), Ann Zimmerman reports that Wal-Mart Chief Executive Lee Scott, in addition to pushing for a higher minimum wage (forget for a moment that less than 3% of all hourly workers in the U.S. get minimum wage or less), told executives and directors that

...he has spent the better part of last year exploring ways to use the company's heft and resources to have a more positive impact on society.

One supposes that someone who has reached such an eminent position in one of the world's most prominent companies is smart, intelligent, and capable on many levels. But I wouldn't expect such a person to say something so preposterous. Unless. Unless it were just for marketing purposes, a feint to increase shareholder value. The rate of return might just justify it, and someone that smart ought to have done that analysis.

Wal-Mart makes the most positive impact possible by providing consumers with quality low-priced goods. That leaves cash in consumers' pockets for them to dispose of however they want - whether that be consuming additional goods, contributing to any "socially conscious" effort they feel like supporting, working a little less to increase their quality of life by spending more time with family or in other pursuits, or even, heaven forbid, saving.

Bottom line: Wal-Mart's impact on society will be substantially less positive if it takes on the role of central planner, deciding what pursuits are best for the masses and consequently being forced to proportionally increase prices (or cease to decrease them further - same difference) rather that sticking to what it's really good at - giving people good stuff for low dough.

Marketing ploy? We can only hope.

Remember Charles Murray and The Bell Curve?

2005-10-12T12:08:00.000-04:00

He's back with an article in the Wall Street Journal. He describes the reaction to the aformentioned tome:

...the furor over its discussion of ethnic differences in IQ was so intense that most people who have not read the book still think it was about race.

I've never read it, either, but I was subjected to lengthy professorial speculation and opinion on the topic and the book.

I bring it up because this article (which is very enjoyable - dispassionate, yet intense; subdued almost to the point of weariness, yet thorough) has a whole lot of correlation, and the central point of the argument's implication is all about causation; in a nutshell, does being of a certain race or gender cause you to be smarter or dumber than those of another gender or race?

The issue is discouragingly complex, and Mr. Murray, while concise, treats it quite well in this essay (read the whole thing), so I'll refrain from extensive comment. The point I want to make is that in these truly interesting real-world cases of statistics in action, the attempt to bring in and properly analyze as many pertinent variables as possible exposes the simple reality of correlation and causation: it's complex, ambiguous, and usually a function of numerous interactions of variables and combinations of those variables.

The takeaway for this blog comes from Steven Pinkers The Blank Slate, which Mr. Murray cites:

...equality is not the empirical claim that all groups of humans are interchangeable; it is the moral principle that individuals should not be judged or constrained by the average properties of their group.

It looks to me a lot like most racing video games, where you choose a vehicle that has a certain score on a variety of characteristics, for example, traction, acceleration, speed, etc. Who's to say which vehicle is the best? Obviously, the point is that each vehicle is the best under certain conditions. Same goes for people.

For practical application, I'll close with one last quote from Mr. Murray:

The differences I discuss involve means and distributions. In all cases, the variation within groups is greater than the variation between groups. On psychological and cognitive dimensions, some members of both sexes and all races fall everywhere along the range. One implication of this is that genius does not come in one color or sex, and neither does any other human ability. Another is that a few minutes of conversation with individuals you meet will tell you much more about them than their group membership does.

Does race determine potential criminality?

2005-10-05T10:46:00.000-04:00

Well, blacks do have a higher crime rate (see 2003 from FBI stats, for example, where the number of offenders is almost evenly broken between whites and blacks in number [almost all male, by the way] but recall that blacks make up only around 12% of the population. You get the idea.) But is it just as simple as that? Of course not! I bring it up because of the following recent event.

Bill Bennett, speaking on a call-in radio program, made the now-famous statement that aborting all the black babies in America would lower the crime rate. Since what we're really interested in here is the correlation involved (race to crime rate), we'll avoid delving into the finer points of rhetoric, the advisability of using reductio ad absurdum arguments on talk radio, etc., and just focus on the statistical element of what he said.

Freakonomics co-author Steven D. Levitt already did the legwork for this stat while writing the aforementioned book, and explains in his post on the Bennett kerfuffle:

It is true that, on average, crime involvement in the U.S. is higher among blacks than whites. Importantly, however, once you control for income, the likelihood of growing up in a female-headed household, having a teenage mother, and how urban the environment is, the importance of race disappears for all crimes except homicide. (The homicide gap is partly explained by crack markets). In other words, for most crimes a white person and a black person who grow up next door to each other with similar incomes and the same family structure would be predicted to have the same crime involvement.

This does, of course, inspire us to talk about the correlation between race and poverty (among the other indicators listed above), but we'll leave that for another day. Till then, we continue to chant our mantra: correlation is not causation.

I'd like to give the world a Coke

2005-10-05T02:21:00.000-04:00

Of course, we also enjoy highlighting good examples around here, so a hardy congrats and thank you very much to National Geographic News writer Brian Handwerk for keeping it straight in his story on history's 6 most influential beverages:

While Coke may not always produce a smile, a survey by the Economist magazine (Standage's employer), suggests that the soft drink's presence is a great indicator of happy citizens. When countries were polled for happiness, as defined by a United Nations index, high scores correlated with sales of Coca-Cola.

"It's not because [Coke] makes people happy, but because [its] sales happen in the dynamic free-market economies that tend to produce happy people," Standage said.

Ice cream sales have been found to track murder rates,

2005-10-02T14:40:00.000-04:00

increasing going into summer and decreasing again heading into fall. Does ice cream contain some malevolent psychotropic ingredient that drives people to turn on their fellow 'sicle suckers? Do murderes, between disposing of bodies and bleaching down the crime scene, run out to grab some Ben and Jerry's to cool off a bit, clear the mind, and avoid some of those 25 mistakes they might otherwise make?

Clearly, the answer is none of the above. This example is an oldie, but a goodie, and while it's ridiculous enough that most people wouldn't accept either of the proposed explanations, most instances of this type of comparison involve items with a little more plausible connection. The result? An almost inevitable conclusion of causation before the story's through.

Of course, correlation does not rule out causation; it simply doesn't establish it. We won't be giving many answers here, but we'll be looking at alternate explanations, and reiterating the point, again and again, that correlation is not causation.

If you see any examples out there, send them in and we'll take a look.

Oh, and in the example above? It's a third factor - the heat of summer brings people out of doors and into more frequent contact, increasing the desire for ice cream, and the opportunity for savagery.