Archive for April, 2009

Lies, damned lies, and statistics

Wednesday, April 22nd, 2009

It continues to amaze me how loosely most people deal with statistics, making up or inflating numbers to prove a point or make something seem more significant than it really is.

Exhibit A is an article on CNN today, about a really sad case where a boy shot himself with a gun that his parents had stored in a closet. Why it wasn’t locked up, and why it was loaded, I can’t say. But here’s the statistic in the article:

The CDC says three children per day, on average, died in accidental incidents in the United States from 2000 to 2005, the last year data are available.

The CDC has some really great data resources available, so I went to WISQARS and ran the report. Here’s what I selected, based on how the CNN article described the set:

  • Intent: Unintentional (that is, accidental) deaths
  • Mode: Firearm
  • Years: 2000 to 2005
  • Ages: Custom range from <1 to 17 (this should be what “children” means, yes?)
  • No age adjusting

The result returned was a total of 724 deaths for the six year period. Too many, yes, but three children per day? Hardly. More like one child every three days. The CNN number is inflated 9 times higher than the actual. That’s past lies and damned lies, that’s statistcal malpractice. (I sent a comment to CNN to see if they’ll correct it, but I don’t have high hopes.)

Long live Mark Twain

P.S. Here’s a great article from The Week entitled “When numbers deceive.” It looks at cancer rates and survivability and quizzes of doctors, and it’s some surprising stuff.

P.P.S. They actually corrected it, but the wording is horrendous:

The CDC says one child, on average, every three days died in accidental incidents in the United States from 2000 to 2005, the last year data are available.

And it turns out that I wasn’t the only one to catch it and write about it. The “Stormin Mormon” used this case to ponder again the limits of journalistic stupidity.