The Lancet: Iraq casualty report...

Baketech · 10-13-2006, 12:44 PM

Not much chatter about this, although Bushy addressed it directly in this week's speech from the Rose Garden.

Info is readily available online, but in short the poll taken by the Lancet asserts that there have been ~650,000 extra deaths in Iraq since 2003 due to the ongoing conflict.

Most mainstream estimates I have seen have been in the 50-60,000 range, so I am curious about what you all think about this report...especially you mathematical types...

Apparently they used the same methods used for epidemiological studies that have proven to be very accurate. Their polling confirmed approximately 600 deaths, and that number was used to extrapolate into the 650K figure...

Are their methods sound, or politically motivated...what say ye?

steingar · 10-13-2006, 12:50 PM

Its not like an esteemed scientific journal like the Lancet to play politics inthis way. Scientific data is not Democrat or Republican, it just is.

luvtolean · 10-13-2006, 12:51 PM

I just skimmed over my stack of unread WSJs and didn't see an article; I saw that # mentioned earlier in the week on a Yahoo news headline and laughed at it.

Do you have a decent link?

Baketech · 10-13-2006, 12:56 PM

There's some background on Wikipedia, and some great reference links at the bottom, including a PDF of the report...

Lancet surveys of mortality before and after the 2003 invasion of Iraq - Wikipedia, the free encyclopedia

bwhip · 10-13-2006, 1:08 PM

Door to door interviews to make the determination?
http://today.reuters.co.uk/news/arti...-C1-Headline-2

steingar · 10-13-2006, 1:20 PM

I'll trust a Scientific Journal more than a politician any day of the week.

sheepofblue · 10-13-2006, 1:33 PM

If you call 2003-6 1000 days then divide you get 650 deaths a day. Seems real high to me.

luvtolean · 10-13-2006, 1:47 PM

650k is an exceptionally large number of dead people. Where are all of them? Even modern weapons normally leave remains.

The other part of that report that has a bad smell is from bwhip's link, they were counting deaths from heart attacks and such.

Phone surveys are the gold standard, but I see the trouble of that in a war zone, so the door-door stuff probably is about as good as you can do. Problem is, in face-face interviews, it is so friggin' easy to bias the interviewee without saying anything.

Baketech · 10-13-2006, 2:17 PM

I know this was survey conducted and verified by very smart people, but I'm having a hard time wrapping my head around the validity of extrapolating ~640 reported deaths into a number like ~650,000....

steingar · 10-13-2006, 2:59 PM

I agree the numbers seem high, but the senior author on this study knows his stuff. I looked him up on Medline, he's done things like this before, and has published in all the top Journals. This does make me suspect that the war has been far worse for the Iraqis than we have been lead to believe.

Baketech · 10-13-2006, 3:12 PM

Just to be clear, I'm not attempting to refute the report or it's methods...but I'm trying to understand the disconnect between the many agencies that have been calculating and reporting on this...how could they ALL be off by roughly an order of magnitude...

phobiaphobe · 10-13-2006, 4:24 PM

And in what locale(s) was their sample taken?

I bet if they'd polled households in Nagasaki in 1945 they could have calculated Japan's losses in WW2 to be about 900 million.

Sunbert · 10-13-2006, 5:00 PM

Politics don't stop at scientific journals.

Heavy-Dee · 10-13-2006, 5:05 PM

Quote:

Originally Posted by Sunbert

Politics don't stop at scientific journals.

what!?!?! Are you implying that it might be in someone's best interests to "find" specific results?!?! OMG! I'm appaled.

HD

steingar · 10-13-2006, 5:57 PM

A scientist who fabricated data for political purposes would be ruined upon discovery, and he or she would be found out. Data is neither conservative nor liberal, Democrat or Republican. Data is data.

luvtolean · 10-13-2006, 6:12 PM

Quote:

Originally Posted by Sunbert

Politics don't stop at scientific journals.

Uh...yeah.

Baketech · 10-13-2006, 7:10 PM

Quote:

Originally Posted by steingar

A scientist who fabricated data for political purposes would be ruined upon discovery, and he or she would be found out. Data is neither conservative nor liberal, Democrat or Republican. Data is data.

Exactly...that's why the global warming issue is so clearly defined...

phobiaphobe · 10-13-2006, 8:03 PM

Quote:

Originally Posted by steingar

Data is neither conservative nor liberal, Democrat or Republican. Data is data.

But how to interpret the data?

jaim · 10-14-2006, 1:39 AM

I haven't read the report and probably won't. Got better things to do. If they are counting all the deaths in Iraq of Iraqi's since the war started I could see it being 650K. The conservative estimates say 100K just from direct warring activities. If you throw in everybody who had died from stuff like malnutrition, improper medical care, blowing themselves and others up in the name of Alah, then sure I could see the 650K. Is Bush to blame? Maybe not for all of them but 10% of that number sure. But remember folks. It is all in the name of keeping America safe.

Mojave954 · 10-14-2006, 8:43 AM

How many of those polled were relatives reporting the same death twice. People over there are pretty whacked as well and think they are all related therefore could even report a "friend" of theirs as a relative and therefore report their death many times over.

EITHER WAY, the Iraqi's want the bombings to stop they need to rise up against the insurgents themselves. Only way it is going to end.

As with any study it should be base off the demographic pool from which it was drawn. Such as the last one they did where they said, "Iraqi's want the Americans out". What they didn't tell you is that was that the poll was taken in a terrorist rich area and only 1000 out of the 100,000 were polled.

Becareful what you read in polls. Saw once that toasters cause cancer. Hmmmmm????? Yeah right.

oldfogey · 10-16-2006, 5:01 AM

It seems to be a respectable survey and extrapolation of the data from the survey. I've seen several comments about the potential weaknesses of the survey and the assumptions used in the extrapolations, that would mean the variance around the estimate would be very large. The headlines on this kind of thing always focus on the central point of the result, where the reality is really the range of probable answers at a given probability.

Baketech · 10-16-2006, 9:04 AM

Interesting post...

I don't think I have a problem with the methods of the study or even the extarpolation. What I do question is the basis of how they state the delta. In other words, unless the previous death rate (prewar) was calculated using similar means, how valid can the stated delta value be?