Stick a Lancet In It

Michelle Malkin reprints David Kane putting a lancet in the heart of the biased and over-the-top Lancet "studies" on post-invasion Iraqi deaths. Some of the most salient points:

• The study's authors refuse to release their underlying post-invasion data, which is contrary to standard practice.



• The study's 95 percent confidence interval is so wide, and flawed, that the authors "cannot reject the null hypothesis that mortality in Iraq is unchanged." This is because, in part, a "significant amount of the mass of the probability distribution for the estimate of [post-invasion estimated] casualties is less than the lower tail for the distribution of [pre-invasion estimated] casualties."



• The authors are vague about pre-invasion mortality, a huge weakness since the study purports to estimate the difference in death-rates prompted by the invasion. Yet, their pre- and post-invasion estimates have widely different confidence intervals, making comparisons imprecise.

As a Malkin commenter noted:

The Lancet is a medical journal, not an epidemiologic journal. This may not seem like an important distinction, but it is: the original study probably wouldn’t have cut the peer-reviewed mustard if statisticians and epidemiologists were doing the reviewing rather than doctors. Doctors may be smart, but they don’t know statistics.

Not that this will change anything. Sadly, most lefties are indifferent to facts, logically challenged, and anxious to believe the worst about the Administration and America.