Sunday, May 15, 2005

Standards, Statistics, and Fun With Numbers

The UN came out with a bunch of reports lately that have some interesting information. I had questioned an article that used UN figures, which claimed that the living conditions in Iraq was worse now than it was when Saddam was in charge.

Now, there are more questions about an old article in The Lancet, which claimed that the war in Iraq caused up to 100,000 deaths (with a 95% confidence interval of 8,000 to 198,000 deaths caused by the invasion). The UN study of the same issue resulted in a 95% confidence range of 18,000 and 29,000, with the toll in the first year of 24,000. The report also claimed that about 12% of those were under 18.

The Lancet article was clearly designed to take advantage of anti-American sentiment, and was touted by numerous anti-war groups as proof-positive that the US was evil. This new study suggests that the Lancet study was flawed, or at least seriously overstated the casualties.

It is possible that the casualties portion of the report was correct, while the portion I questioned was flawed. Or, the portion I question as incorrect was really correct, and the Lancet piece was on target. Or, the UN report was wrong on both sides.

At a minimum, this suggests that the UN's penchant for having fun with numbers extends well beyond calculating oil for food deals.

No comments: