Iraq death toll

[Walsingham]

http://news.bbc.co.uk/1/hi/world/middle_east/6040054.stm

 

I can't see anything much wrong with their methodology. But I do think there must be something to explain the massive discrepancy between this 650,000 number and all the others. Anyone know more about this? I haven't had time to look into it in more detail.

[Tigranes]

Considering civilian death in Afghanistan in the first year or so of the operation significantly outnumbered 9/11 deaths (3000 something), and this is a 3-year period including military, I can see how it is perfectly plausible.

 

Most definitely,

 

In the past, Mr Bush has put the civilian death toll in Iraq at 30,000,

 

Is bullcrap. At least in six figures. Their cluster sample will certainly be more accurate, if done properly.

[Walsingham]

Considering civilian death in Afghanistan in the first year or so of the operation significantly outnumbered 9/11 deaths (3000 something), and this is a 3-year period including military, I can see how it is perfectly plausible.

 

Most definitely,

 

In the past, Mr Bush has put the civilian death toll in Iraq at 30,000,

 

Is bullcrap. At least in six figures. Their cluster sample will certainly be more accurate, if done properly.

<{POST_SNAPBACK}>

 

I don't know where you're getting the six figure thing. Have a read of IraqBodycount.com They aren't nearly so sure as you sound. I believe they put it at 45,000.

 

I think it's also important to point out that all these studies look at deaths post invasion, not deaths directly caused by Coalition forces.

[Musopticon?]

I think it's important to note that without this Iraq adventure, there'd be half a million less corpses.

[Walsingham]

I think it's important to note that without this Iraq adventure, there'd be half a million less corpses.

<{POST_SNAPBACK}>

 

Really? I rather thought that was the point of questioning the figure!

[Musopticon?]

:D

[Judge Hades]

You know, both the extreme high number and the exxtreme low number seems a bit off to me. I am sure the real death toll is somewhere in the middle.

[kumquatq3]

Considering civilian death in Afghanistan in the first year or so of the operation significantly outnumbered 9/11 deaths (3000 something), and this is a 3-year period including military, I can see how it is perfectly plausible.

 

Most definitely,

 

In the past, Mr Bush has put the civilian death toll in Iraq at 30,000,

 

Is bullcrap. At least in six figures. Their cluster sample will certainly be more accurate, if done properly.

<{POST_SNAPBACK}>

 

As far as I have read, this is the only one to put it into 6 figures.

[Colrom]

The Iraq Body Count numbers are virtually useless for a number of reasons. They count bodies. They only count civilian casualties. They require two independent original source news reports of the casualties. They require that the news reports be in english.

 

They also engage in propaganda by using the word "maximum" to mischaracterize their higher results.

 

I actually spoke to them a long time ago about their misuse of the word "maximum", suggesting, perhaps unfairly, perhaps not, that they could easily be seen as a source providing preemptive damage control for the governement.

 

They were determined to keep their terminology.

 

Most folks recognize that their "maximum" numbers are wildly low by any rational standard - but just how wildly low is difficult for a person, like me, who reads just english language sources to really know.

 

As an cynical aside the university where they work has many UK government funded projects.

 

There are various stories on the latest casualty estimates based on population sampling. A completely different approach - much more likely to reveal overal casualty levels - including even evidence of casualties that have been burried in the dark in the dirt on the side of the road. I have seen various versions based on the AP story.

 

A key bit was the statement that most families provided death certificates. So the sample is not just hearsay.

 

CNN had added additional info supporting the credibility of the sampling methodology employed and also included a link to the actual report and the data appendix. I may be able to post a link if folks want. They are PDF files.

 

CBS had added to their version of teh AP story additional info from a political research group friendly to the administration which criticized the report - but provided no grounds.

 

I am inclined to think that the sampling methodology is solid and the results are reasonably accurate. I need to see what the estimate of uncertainty is (95% confidence level). That is a key number.

 

In any case this estimate appears to be generally consistent with the earlier Lancet estimate - which had a very large uncertainty - which covered the period from the start of the war till Aug 2004.

 

Edited October 12, 2006 by Colrom

[taks]

one of the problems with this study is that they are extrapolating from one study, apparenlty limited to the surveyors' choice of where to inverview. while on the surface, it may seem plausible, it cannot hold to any real scrutiny. why? sampling bias.

 

hades has a relatively lucid comment regarding this, probably true.

 

taks

[Colrom]

Here's more data:

 

They claim 95% confidence that the casualties since the invasion are more than 426,000 and less than 794,000.

 

They calculated an estimate based on the new sample results just for the period from the start of the war till August 2004. They get 112,000 casualties from teh new sample data as compared to about 100,000 reported based on teh old sample data.

 

For 92% of the casualties reported family members had death certificates.

Edited October 12, 2006 by Colrom

[Colrom]

one of the problems with this study is that they are extrapolating from one study, apparenlty limited to the surveyors' choice of where to inverview.  while on the surface, it may seem plausible, it cannot hold to any real scrutiny.  why?  sampling bias.

 

hades has a relatively lucid comment regarding this, probably true.

 

taks

<{POST_SNAPBACK}>

 

I don't understand what you are saying. I will look into it.

 

So far as I could tell their choices of coordinates to interview were randomly decided based on population density weighted geography.

 

That seems good to me.

 

Will check and get back.

[Colrom]

What the report says about sampling is this:

 

"Selection of the sites

Selection of households to be interviewed must be completely random to be sure the results are free of bias. For this survey, all households had an equal chance of being selected. A series of completely random choices were made. First the location of each of the 50 clusters was chosen according the geographic distribution of the population in Iraq. This is known as the first stage of sampling in which the governates (provinces) where the survey would be conducted were selected. This sampling process went on randomly to select the town (or section of the town), the neighborhood, and then the actual house where the survey would start. This was all done using random numbers. Once the start house was selected, an interview was conducted there and then in the next 39 nearest houses. The distribution of the sample sites or clusters is shown in Table 1, which is based on the 2004 UNDP/Iraqi Ministry of Planning population estimates."

 

Table omitted

 

The report continues:

 

"Conduct of the survey

The two survey teams consisted of two females and two males each with one male supervisor. All were medical doctors with previous survey and community medicine experience and were fluent in English and Arabic. All were Iraqis. All were trained in the use of the questionnaire. Rules were established about how to randomly choose another area if the first one chosen was unsafe on the day of the survey visit."

 

I do not see bias in this - other than possible bias due to families being completely wiped out and therefore unsurveyed - but they checked for that too.

 

What do you have in mind Taks?

Edited October 12, 2006 by Colrom

[Walsingham]

Still reading. Thanks for taking the trouble to quote for us, Colrom.

[taks]

"Once the start house was selected, an interview was conducted there and then in the next 39 nearest houses."

 

sorry, but that's a sample bias.

 

taks

[Walsingham]

Of course if I were getting shot at while doing my job I imagine I'd cut some corners too.

[taks]

in the end, their statistics for confidence interval are based (more than likely, i have not read) on the total number of families. however, by taking 40 homes immediately next to each other, they've effectively reduced their sampling to 1 point for each cluster (since it is likely that each of the 40 homes immediately next to each other are similarly effected as the first random choice).

 

are there statistics based on all of the homes randomly chosen, all of the homes divided by 40?

 

taks

[taks]

Of course if I were getting shot at while doing my job I imagine I'd cut some corners too.

<{POST_SNAPBACK}>

but that doesn't make it any less a sample bias. for the statistics to hold over all families, they all need to be randomly sampled. there is also a bias in choosing only the densest areas. geographic weighting or not, they've still limited their search to only areas that had a potential to be in the middle of conflicted areas.

 

sampling is an art, for sure, and i don't fault them for their efforts. they simply need to be reviewed for what they are. also, how much effort is there in determining whether these were people that might have died anyway? certainly they compared death rates prior to the invasion, but how do we know that death rate would not have risen on its own anyway?

 

correlation != causation.

 

taks

[Volourn]

When 99% of other studies including those done by those vehemently anti US ahve much less death counts, I laugh at this 'study'.

 

What a joke!

 

I'm sorry; but they're likely counting those poor folks who got ran over by a car or had a heart attack at age 99.

 

*yawn*

[taks]

of course, the other 99% need to be subjected to the same questions and scrutiny before being believed.

 

taks

[Colrom]

Here are some of the studies comments (in an appendix) regarding the issue of cluster sampling which Taks is commenting on. It may be heavy for some.

 

From an Appendix:

 

"A problem with cluster surveys is that households adjacent to each other are more likely to be similar than those located farther away. In the case of localized violent events, the same event is likely to affect households close together. This makes simple random sampling a stronger survey method where this is possible. But in war this is seldom possible.

 

To compensate for this "clustering effect" (sometimes called the design effect), the number of households or persons in a cluster sample is increased over that of a simple random sample in order to provide adequate precision. As one does not know the extent of "clustering" before the survey is started, it is usually estimated at two, meaning that a cluster survey would need twice the number of households as a simple random survey in order to have equal statistical power.

 

Afterwards the clustering or design effect can be calculated from the results to see if the estimate of 2.0 was indeed correct. In the 2006 Iraq mortality study, during its analysis this effect was found to be only 1.6

Edited October 12, 2006 by Colrom

[Colrom]

When 99% of other studies including those done by those vehemently anti US ahve much less death counts, I laugh at this 'study'.

 

What a joke!

 

I'm sorry; but they're likely counting those poor folks who got ran over by a car or had a heart attack at age 99.

 

*yawn*

<{POST_SNAPBACK}>

 

So far as I know this group is the only group actually trying to find out the effects of the war.

 

By the way, if the car they get run over by is a US military or insurgent vehicle or is a vehicle that ran out of control because of a bomb or a hole in the road caused by a bomb - those would be casualties of the war.

[Judge Hades]

The leading cause of death is life.

[taks]

Good catch on your part that that needed addressing. 

<{POST_SNAPBACK}>

i'm merely asking questions that should always be asked. sample extrapolations are very easily the easiest to corrupt, even without intent. similarly, probably even more true, poll results should be questioned for the same reasons.

 

note that i have not passed any value judgement on the results. without an in-depth analysis (cannot right now, studying for linear algebra exam), i would be just as culpable as someone that automatically accepted the results if i were to offer opinioned criticism.

 

taks

[kumquatq3]

Stats are all facts and no heart. You have to go with your gut. - Stephen Colbert