Tuesday, August 21, 2007

Lancet's Boil

The American Spectator: Lancet's Boil

Michael Fumento writes about a paper from David Kane, a fellow at Harvard University's Institute for Quantitative Social Science. Kane is essentially a stats geek. What he's done is conducted a statistical analysis of the Lancet medical journal's 2004 study, which concluded that U.S. military operations caused at least 100,000 deaths in Iraq.

Kane's conclusion: Upon inclusion of data the Lancet ignored out of political convenience, the confidence limits associated with the estimate are so wide that the lower limit is negative (which means there is a small chance that the invasion saved lives or... resurrected dead Iraqis).

Lies, damned lies, and statistics.

And geeks who get it right.