DATA ABUSE

Posted in Quick Hits on March 25th, 2013 by Ed

I know that most people suck at logic, statistics, and a bunch of other things without which they are easy marks for bad arguments and misleading numbers. It would be nice, however, if major media outlets and publications had at least a basic grasp of how to interpret data like non-morons.

In "Ladies Last", National Geographic presents a map of the gender gap in life expectancy in the United States. Interesting and interactive map. Well done. And here's the analysis:

How long do you have? It depends on gender and geography. In the U.S., women live longer—81 years on average, 76 for men—but a recent study by the Institute for Health Metrics and Evaluation reveals a troubling trend. Though men's life spans have increased by 4.6 years since 1989, women have gained only 2.7 years, perhaps because a larger percentage of women have lacked adequate treatment for high blood pressure and cholesterol. "This is a wake-up call," says study co-author Ali Mokdad.

Wait, what? That's the dumbest thing I've ever seen, and I've seen Rick Perry.

To recap, female life expectancy has increased to 81 while men lag behind at 76, but we should be alarmed that women only gained 2.7 years to 4.6 for men since 1989. So the only thing that has decreased is the female advantage in life expectancy. While they used to outlive men by eight years, now it's only five. Absolute life expectancy for both genders has grown.

Using the magic of inductive reasoning, there appear to be some fairly obvious explanations.

1. In 1989, most of the WWII age cohort was alive and made up a substantial portion of the total population. Now many of them have died and subsequent generations – ones in which 500,000 men did not die young due to war – have outnumbered them. In 1940, women outlived men by 1 year. By 1989 that had grown to nearly 8 years. HMMM.

2. There has been a staggering decline since 1989 in highly dangerous, male-dominated, and historically common types of work. Coal mining and logging come to mind immediately, given how the gap has shrunken precipitously in the Pacific Northwest and Appalachia on the interactive map. Not only are those industries in decline, but productivity and mechanization allow fewer people to do more work.

3. There are mountains of evidence that American men are far more reluctant to get medical attention than women, making them easy victims for otherwise preventable or treatable problems.

4. Men are far more likely to die from "unnatural" causes like violence, road accidents, and suicide than women.

5. Math. Life expectancy is roughly bounded at the high end. In other words, it can't just increase infinitely and the marginal cost of increasing it once we reach the 80s is quite high. Imagine that I run the 100m in 15 seconds but you run it in 10. If we both work our asses off for a year, I'll probably improve to 12.5 seconds, while you'll be lucky to trim down to 9.9 seconds. By starting out closer to the theoretical upper limit of how fast a human can run, of course you're going to show less of a "gain" compared to someone who is lagging far behind.

But faced with all of these really, really obvious potential explanations and data that shows women still outlive men by a sizable margin, the writer (and the author of the damn study) conclude that the best explanation is women getting too few prescriptions for Lipitor.

Cool.