Friday, December 21, 2007

How to lie with statistics

When I was a kid, I read and loved Darrell Huff's How to Lie with Statistics. Now the web site Stats seems to be keeping that spirit alive. Here's a recent posting.
Recently, Time Out New York sent a wave of panic through the city’s single women by reporting in its cover story that there are 185,000 more single women than men here. The article cited National Geographic, which had analyzed census figures. To make the numbers even scarier, the package cited excesses of women in college compared to men.

But this is a false comparison as college enrollment figures are not measures of population: there may be fewer men enrolled in college, but among the college-age group, there is no shortage of men compared to women. And the census figures actually do not include people living in dorms.

Basically, the excess of women is due to the fact that men tend to die at younger ages than women do. If you look at the male/female numbers in the younger age groups, in most, there are significantly more men. For example, there are 211,590 men aged 18 and 19 in the NY Metro area – but only 201,282 women.

The disparity may also reflect shorter lifespans and excessive incarceration among men of color: in the white non-institutionalized population, men actually outnumber women even in the 35-44 age group, but in the whole non-institutionalized population, the male/female ratio is skewing female by that age.

No comments: