Wednesday, December 21, 2005

Correlation?

I believe it was Mark Twain who said that there are three kinds of lies: lies, damned lies, and statistics. As much as I admire Twain, I also have a fondness for interesting statistics. Here are a couple that caught my eye today.

  • At year end 2004 there were 3,218 black male sentenced prison inmates per 100,000 black males in the United States, compared to 1,220 Hispanic male inmates per 100,000 Hispanic males and 463 white male inmates per 100,000 white males. (http://www.ojp.gov/bjs/prisons.htm)
  • In 2004, rates of AIDS cases were 56.4 per 100,000 in the black population, 18.6 per 100,000 in the Hispanic population, . . . 6.0 per 100,000 in the white population, . . .(Table 5a). (http://www.cdc.gov/hiv/stats/2004SurveillanceReport.pdf )
Now one set of numbers refers only to males and the other to both males and females so at least one of the apples here is looking a bit like an orange. Looking only at AIDS is also a misleading way to look at the epidemic because AIDS statistics are often reported as cumulative numbers and include people who have died. Looking at HIV statistics would be much more effective in helping us understand whether there is any correlation here, but HIV reporting is done in several different ways in the states, so we only have reliable national numbers for AIDS.

What I'm saying is that, yes, statistics have to be examined a bit to make sure they are not the subject of Twain's rant.

On the other hand, I'm also saying that disproportionate is disproportionate any way you cut it (but especially when you're dealing with an epidemic caused by a virus that really is no respecter of persons). How then does the virus manage such disproportionate effect? Just askin', y'know.

No comments: