Sunday, January 13, 2008
Oil is to water as reporters are to knowledge of statistics
The New York Times is running a front-page story today about a supposed spate of homicides committed by veterans returning from Iraq and Afghanistan. Unfortunately, moments after its publication Marc Danziger at Winds of Change rather ably demonstrated that the homicide rate for returning veterans is almost certainly lower than for the same age cohort in the American civilian population.
Struggling to be fair, the Times does make the point that the homicide rate among active duty and new veterans since 2001 is higher than it was for the same groups (recognizing that they are different people) in the six years prior to 2001, which corresponds to a period when the homicide rate in the United States was declining. In effect, if you take a couple of million young people, mostly men, and train them to kill and then subject them to extraordinary psychological and physical pressure on the other side of the world, make them confront one of the most vicious enemies the United States has ever fought, and force them to act with great restraint in the face of terrorism targeting women and children, their propensity to commit violence in other contexts will go up, but still be less than for American civilians of the same age.
If anything, that is evidence of the extraordinary professionalism of our military.
The question, therefore, is whether the editors of the New York Times were disingenously slamming the military, or just did not understand the math.
My father, a medieval historian, ordered me to take only one specific class in college -- statistics. His point was that a basic knowledge of statistics was now essential to being an informed citizen. Apparently, however, it is not essential to writing about statistics in the newspaper that claims to be the best in the country, if not the world.
CWCID: Glenn Reynolds.
8 Comments:
By 
, at
Sun Jan 13, 10:15:00 AM:
The editors of the NYT are knowingly doing evil, vile things for political gain. The long term damage to the paper is going to be incalculable.
By 
Unknown, at
Sun Jan 13, 01:00:00 PM:
You're father, the historian, and my Mother, the economics professional, both with the same advice to their college-bound offspring. Hmmmm.
By Gary Rosen, at
Sun Jan 13, 02:00:00 PM:
"The question, therefore, is whether the editors of the New York Times were disingenously slamming the military"
I'm sure TH is just being rhetorical. If anyone actually questions this, they have not been paying attention.
"or just did not understand the math."
Remember the old ads for Certs candy/breath mints? "Stop, you're BOTH right!" This is not an either-or question. I went to high school with Pinch Sulzberger, but it was only for a year because he flunked out. Believe me, if he hadn't inherited his job he wouldn't have gotten as far in journalism as Jimmy Olsen.
By Escort81, at
Sun Jan 13, 05:41:00 PM:
I agree that Stats is a vital course no matter what you want to study -- it's the language of data interpretation and a key to the scientific method. It is important if you are studying in the social sciences or the "hard" sciences. Try getting something published in a meaningful peer-reviewed journal without understanding statistics. Try attending a meeting with FDA if you work for "evil" big pharma or a medical device company and don't know the numbers of your IDE, 510(k) or other pending application inside and out.
That said, there seem to be many reasonably intelligent people, even in the business world, who don't understand even rudimentary aspects of statistics. I wonder if the NYT would have understood whether to use a difference between means test (for independent populations) or a paired difference test (which you would use, for example, to compare tire wear among different brands of tires on mounted on the same car, even if you had multiple cars, because the tire wear is not independent). Even harder would have been figuring out whether the data lent itself to a normal or non-parametric analysis.
The idea that we are creating a generation of human time bomb veterans has been out there since a few months after Saddam's statue fell. Seymour Hersh was among the first to write and speak about it. Vets should have access to as much quality care as is needed, regardless of the degree if relative incidence of mental health issues across similarly situated population samples -- maybe we can all agree on that point.
By , at
Sun Jan 13, 06:07:00 PM:
The NYTimes is written by the left for the left. Why should anyone be surprised at this latest farrago of lies?
By jj mollo, at
Sun Jan 13, 08:53:00 PM:
I believe it's really just innumeracy and the simple lack of any scientific training. These people talk only to each other. There is no skeptical analysis of their articles. A simple question might have forced the writer to form the insinuation into a hypothesis. The writer needed to come out and explicitly state he believed that returning soldiers were subject to violent behavior. The natural skeptical response is, "Compared to what?" After that the article might have taken on a dramatically different tone.
Look at how badly the pundits screwed up in New Hampshire. So much of what the journalism types do is predicated on anecdote and emotional barometry. "You could just feel the excitement building." Very few people thought to ask themselves if they could get a random sample of such "excitement events" in order to determine the distribution of subsequent outcomes. Perhaps, if journalism involved some training in skepticism instead of cynicism we would start seeing articles that recognized their own biases and asked the appropriate questions.
By Fritz, at
Mon Jan 14, 08:30:00 AM:
What JJ Mollo said. The degree of innumeracy among the general population is truly astonishing, and it's no better among reporters, unfortunately. For a more trivial example, read any particular column arguing about baseball HoF votes. It's really saddening.
By Miss Ladybug, at
Mon Jan 14, 01:10:00 PM:
I've always been good at math (although at a certain point, I don't like it too much). Until I get a "real" teaching job, I work retail part-time to supplement my subbing income. Too many customers can't calculate, even with rounding, the sale discounts (25, 40, 50, 70 percent off the ticketed price) to figure out sale price. It's not great leap to see why too many of the American public can't think critically about these statistics issues... Sad, but true.
