Quote:
Originally Posted by Yirmeyahu
The first argument is fallacious. Most other estimates are far lower than the one published in the Lancet, and therefore the Lancet study must be "bogus". This as fallacious on its face.
Why the difference? Methodology, plain and simple. Take Iraq Body Count, presently estimated around a minimum of 48,755 civilian deaths. First of all IBC counts only civilian deaths. The "Lancet study" (the study was by the Jons Hopkins Bloomberg School of Public Health, published in the Lancet) estimated total Iraqi deaths, civilian and combatant. IBC counts only deaths reported on in the news media, by at least three sources. Naturally, this means the IBC estimate is extremely conservative. The "Lancet study", on the other hand, used the standard methodology for estimating based on cluster samples, a far more scientific approach, and without question far more accurate.
The low estimates use "bad methodology", not the "Lancet study".
|
Benchmarking is important. To argue otherwise is to ignore survey methodology. There is a degree of apples and oranges in the IBC/Roberts et al comparison; however, you'd have to argue that 9 out of every 10 deaths is not captured by IBC. Can you tell me where these nearly 550K dead bodies are?
In terms of comparing the results to other available studies, their original study stacks up very poorly against the UNDP study which had a much, much greater degree of precision. Also, do you care to provide a benchmark that stacks up to scrutiny that will validate their pre-invasion #s?
Quote:
|
Originally Posted by Yirmeyahu
It is bad practice to use a cluster sample for a distribution known to be highly asymmetrical."[/i] It is not "bad practice" to use cluster samples. It's the standard, well-accepted, and widely used methodology for arriving at such estimates.
|
I wouldn't state that it's bad practice, but their precision in their estimates is horrible, and some of their methodology doesn't hold up to the required assumptions.
Quote:
|
Originally Posted by Yirmeyahu
"18 of the 33 clusters reported zero deaths". That is false.
|
Read page one of the findings and the snippet that you quoted from. Read the way that you improperly edited, it would be false. Use the entire quote, and you are totally wrong. Was this deliberate editing or just a faulty reading?
Quote:
|
Originally Posted by Yirmeyahu
"Two-thirds of all violent deaths were reported in one cluster in the city of Falluja" This quote does not come from the report. The only mention of Falluja in the report notes that "Over half of excess deaths recorded in the 2004 study were from violent causes, and about half of the violent deaths occurred in Falluja." Falluja was excluded from the study on account of the high rate of mortality there.
|
Read table 2 from Roberts et al's first study - 52/71 violent deaths occured in Fallujah. By my calculations, that's 73.2% of all violent deaths occured in Fallujah. If anything, you could make an argument that the authors of the one report could have claimed 3/4 of all violent deaths occured in the Fallujah cluster, but they instead stuck with the fully defensible 2/3 mark.
Quote:
|
Originally Posted by Yirmeyahu
The rest of the "debunking" is follows along the lines if this fabrication, warranting no further response.
|
No it doesn't. I've provided many other things to look at. If you want to discuss my challenges to the study that invalidate their findings, then let's discuss. However, please read what I wrote and provide contextually correct quotations so we don't have to fight over misquotes or attempts at parsing to mislead.