"Steven E. Moore's Oct. 18 editorial-page commentary "655,000 War Dead?"

I simply posted links to two sources saying approx 40,000.

If you prefer the 655,000 for whatever your reasons, so be it. I have no number.

http://www.alertnet.org/thenews/newsdesk/L24878128.htm

http://www.iraqbodycount.org/database/

Steven E. Moore's Oct. 18 editorial-page commentary "655,000 War Dead?" calls into question the estimate by Johns Hopkins scientists that there have been more than 600,000 "excess" deaths since the U.S. invasion of Iraq relative to what would have been expected from levels of mortality in the 15-month period before the invasion.

We agree with Mr. Moore that a cluster survey is the preferred approach to quantifying post-invasion violent deaths in contrast to counts of deaths from newspaper articles and morgues or not counting at all. Cluster surveys have two components, the number of people included in the survey and the number of points from which these persons were sampled. Mr. Moore's objection is with the number of clusters in the survey, not with the methodology of the survey or the number of people included. Because there were only 47 clusters, he claims one would have "crazy results." Survey experts know that estimates of mortality with 47 or 470 or 4,700 clusters are all unbiased. That is, unlike news reports, surveys accurately estimate death rates. What is different between these examples is the degree of precision in their estimates; the more clusters, the more precise is the estimate if sample size also increases.

What is remarkable about survey methodology is that in addition to a best estimate, it also produces a range of plausible values, that is, an interval that reflects the margin of error in the estimate. Based upon our 47 clusters, the Hopkins team reported a best estimate of 655,000 but we also gave a range of plausible values from 393,000 to 943,000. Had the study used 470 clusters, this range would have been about three times narrower.

While there were 47 clusters in the survey, there were also 12,800 people surveyed, a fact that Mr. Moore omits from his piece. Surveyor researchers use the "effective sample size" defined to be the number of independent people who would give a similar degree of precision as their survey to indicate the "size" of the study. In the Hopkins study, the effective sample size is roughly 3,700 independent people, nearly three times larger than the typical political survey that reports +/-3% that Mr. Moore offers as the gold standard.

We agree that more clusters increase the precision of the result. However, more clusters also mean more risk to the survey team. Our final decision on the number of clusters represented a balance between obtaining a meaningful sample and having survey team members return alive.

We agree with Mr. Moore on the importance of estimating the number of civilian deaths in the conflict, and hope he will join us in our recommendations for an independent body with adequate resources to monitor deaths among civilians in conflict using scientific methods as was done in our survey.

Gilbert Burnham

Shannon Doocy

Les Roberts

Johns Hopkins Bloomberg

School of Public Health

Baltimore

You left out the "a" before WHOLE ( that sounds bad ), misspelled "direct" and "shown". I know typos are not the same as misspellings but in the end, it doesn't make any difference. I mean, it doesn't disqualify anyone from being The Preznent.

and it's spelled hypocrite, btw…

larry, you have some talent for being funny. if i were you, i'd stick to that…