Thursday, October 12
Bush opened a new front in his ongoing War on Science in a press conference Wednesday, when he claimed that a standard social science survey methodology has been "pretty well discredited".

Bush was responding to a question about a study appearing in Lancet estimating the number of Iraqis who have died as a result of the war. The study used cluster sampling, where survey respondents are chosen from specific communities rather than at random from the population as a whole.

Now, I'm no expert, but every book on sampling I've ever seen talks about cluster sampling as a valid approach to large-scale surveys. And it's my understanding that most national social science and education surveys right here in America use some form of cluster sampling. This is because the SRS (simple random sample) you learn about in Stats 101 isn't feasible to generate for a national population.

Bush went on to insult the scientists behind the Lancet study by saying of the number the study arrived at, "600,000, or whatever they guessed at, is just -- it's not credible." (emphasis added)

A considerable amount of work from a large team of professionals went into that "guess". I find that "guess" more credible, in fact, than the Bush administration's guesses about whether there were WMD in Iraq before the war, whether their diplomacy with North Korea has been effective, and whether the mission was really accomplished when Bush stood on the carrier in a flight suit.

But demographers and public health researchers be warned, you are now evidently part of the Axis of Academic Evil, along with global warming researchers and believers in evolution.

1 comments:

fraud, in denim said...

I have no words (at least no words suitable for this forum).