Quote:
Meaning that they included no Anbar data in their estimate of the death rate in Anbar. Which can only mean they guessed. Look at Figure 2 in the Lancet 1 report. It states that "Governorate rates are on a scale of 15 deaths per 1000 person-years, except for Anbar governorate, where deaths were more than ten times higher." I ask you, if they threw out Falluja from the final figure, how did they come up with that ten times higher rate for Anbar?
Figure 2 is not their final figure. Their final figure was 100,000 deaths for the period in question. They didn't throw out the Fallujah data in the sence of not telling us about it. They tell us about it and illustrate that it is of an entirely different scale.
You continue to obstinately hide from the obvious. According to multiple sources the Lancet 1 researchers made a public estimate that there were 57,600 violent deaths for all of Iraq excluding Anbar. Since Lancet 1 said that about 98,000 died in all of Iraq, that means they must have concluded that about 40,400 died in Anbar (i.e., 98,000 - 57,600). Now the population of Anbar is about 1.2 million compared to a population for the rest of Iraq of about 25.8 million. That means the death rate in Anbar was about 1 per 30 persons while the death rate in the rest of the country was 1 per 448 persons. That means they concluded in their final results that the death rate in Anbar was 448/30 = 15 times higher. So you are wrong again. Figure 2 does indeed appear to reflect the final results. So I ask again, since the Lancet 1 researchers threw out their Falluja data point, how did they come to the conclusion that the death rate in Anbar was 10 (to 15) times higher than in the rest of the country. I'll tell you. They guessed based on their own personal bias against the war, Bush and the military.
Quote:
Falluja was the only cluster in Anbar. They can only have guessed
They didn't guess. That was the number they got. Then they saw that the number is of a totally different scale to the other numbers and treated as an outlier.
If they threw out the data point as an outlier, they had no other data in Anbar. So the ONLY way they could have come up with an death rate estimate in Anbar was to use someone elses data in some manner (like NEJM did for those portions of Anbar and elsewhere where they weren't able to sample) or guess. They guessed.
Clusters were assigned at random. And then they left the Fallujah data out of their final figure -- because Fallujah was so much worse.
They state very clearly in their report that there was only 1 cluster in Anbar ... and that was the Falluja cluster. They guessed. And guessed very high.
The NEJM study indicates the violent death rate remained relatively constant over 3 different time periods after the invasion.
And you don't think that's strange, given the way the number of reported deaths varies?
Over the time period (March 2003 - June 2006) in question? No. Because there are other sources that indicate much the same thing. Iraq Body Count for one. They certainly don't indicate that the violent death rate went up by a factor of 4 between the two time frames as Lancet 2 study does.
Quote:
Well, let's see. NEJM says that the rate of excess violence related death was 1.67/1000/year (after bumping the survey results up 50% to account for possible undercounting). With a population estimated at about 27 million, over a 40 month period that works out to about 150,000, just as they said. But then again, that is entirely dependent on that ridiculous pre-war mortality rate of 3/1000/year.
Those are the excess violent deaths.
But we want to compare apples and apples. The Lancet 1 study said that 59 percent of the deaths were violent. The Lancet 2 study concluded that an even higher percentage were due to violence. So naturally, we want to compare Lancet results to the violent portion of the NEJM report. Because after all, you said that direct polling is more accurate than using other means and those people they polled supposedly told the JH researchers the deaths were due to violence ... not disease or old age. And surely they wouldn't misremember or lie about a little detail like that ... right?
What is your number for the excess total deaths?
Let's look at what NEJM said the total excess deaths were from March 2003 to June 2006. And by gosh, their report doesn't actually give a number. What it said is that between 104,000 and 223,000 Iraqis died violent deaths between the start of the U.S.-led invasion of Iraq in March 2003 and June 2006 with a median number of 151,000. And it gives the raw data from which I concluded that about 14 percent of total deaths were due to violence (and by that I meant murder and armed conflict).
Remember: you are saying that 150,000 is just 14% of the total excess deaths.
Actually, *I'm* not saying that. I just did a quick calculation that suggested NEJM was saying that. But let me do the calculation more carefully using the data in Table 2. It says that for the (about) 39 months after the invasion, the survey recorded 1121 deaths in the surveyed population. Of that total 157 were due to intentional injuries or armed conflict. That's where I got the 14% of the total: 157/1121. But I admit that 1121 and 157 are not excess deaths.
So the question becomes how to calculate the pre-war deaths that should be subtracted from these two quantities. According to Table 2, the survey found that 204 died in the (about) 15 months before the invasion. So in 39 months, the same pre-war rate would have yielded 2.6 times more deaths ... or 530 deaths. And there would have been about 16 violent deaths in that same period based only on the pre-war survey numbers multiplied by 2.6. So now we can recompute the percentage.
(157-16)/(1121-530) = 24%. Which splits the difference between what you and I were claiming the percentage was for violent deaths. Fair enough? That's still far different from the 59% claimed in Lancet 1 and the over 90% claimed in Lancet 2. Right?
But now I'm curious. How did the NEJM researchers come up with their violent rate figures? They surveyed a total of 9345 households with a total population of 61,636. The population of Iraq was about 27 million so they surveyed 0.228 percent of the population. That means we should be able to multiply the number of people who died in their survey by 1/.00228 to roughly get the number of deaths in the time period. Let's check. If we multiply the excess number who died violently: 141 * 1/.00228 = 61,800. They said 100,000. So that seems a little low.
If we include the people who died in road accidents and unintentional injuries in the excess violent death total we get (141 + 42-4 + 61-11) = 229 deaths. 229229/.00228 = 100,000. Ahhhhh! So perhaps they included those two categories in what they labeled violent deaths? In the abstract of the NEJM report, they state that "violence is a leading cause of death for Iraqi adults and was the main cause of death in men between the ages of 15 and 59 years". Nowhere in the report do they specifically explain what comprises violence. But in the section describing Table 2 they state "Overall, the proportion of deaths from INJURIES
(BAC - in other words, all four injury categories) increased from 10.5% before the invasion to 23.2% after the invasion. The increase was most dramatic among men between the ages of 15 and 59 years, among whom deaths from injuries increased from 31.2% before the invasion to 63.5% after the invasion and became the leading cause of death in this age group". Putting 2 and 2 together, I think it is quite clear that NEJM included road accidents and unintentional injuries in their "violent" category.
We can argue about whether that is proper or not later, but first lets see if the Lancet researchers did the same thing. The first JH report (Lancet 1) states that "Violent deaths were defined as those brought about by the INTENTIONAL acts of others". The Lancet 2 report states "We define non-violent deaths as not due to intentional violence—that is, our non-violent deaths include deaths in “accidents,” such a traffic fatalities." So clearly, they did not do the same thing as NEJM.
In which case the discrepancy between the NEJM results and the John Hopkins studies is even larger than it seems. If road accidents and unintentional injury deaths had been included in the violent death category, then the percentage of violent deaths claimed by the Lancet studies would have been even higher than the 59% and 91.8% claimed in the two Lancet studies. Or if you eliminate the non-intentional deaths from the NEJM violent results, you get a result that is even closer to what Iraq Body Count and others have been saying. I rest my case.
If you take the numbers they give for death rate, before they account for under-reporting, then the violent deaths are 1/3 of the excess deaths.
False, as anyone can see, given the numbers and calculations I've noted above.
Do they account for the under-reporting of non-violent deaths?
As noted, they don't make a final estimate of total non-violent deaths so the answer is obviously no. But I suspect underreporting of non-violent deaths would be even higher than violent deaths. Just ask any reporter.
NEJM says that "Only 0.4% of households declined to complete the questionnaire."
15 (0.8%) households refused to participate in Lancet-2
So NEJM had a better response rate.
As far as houses where someone was home refusing to participate? Yes, that appears to be the case but then I would expect that, given that tensions in Iraq at the time the NEJM study was done were probably less than during the Lancet 2 visits.
NEJM says 0.7% were absent for an extended period of time. Lancet-2 says that in 16 (0.9%) dwellings, residents were absent. What is suspicious?
Sigh. As already pointed out to you, the NEJM percentage is based on multiple visits to the same location. If the occupants weren't home the first time, they tried again and again. The likelihood that 99.1% of residences would have adults home during a single randomly timed visit with no warning is in itself quite suspicious. No other survey in any country at any time on any subject has had this success. The claim by John Hopkins that 92 percent of those who claimed deaths were able to find and show death certificates at the very end of their interview (in a matter of a few minutes, in other words) is also suspicious.
But the news of the death has to travel to the recorder.
Yes, but the LATimes in a few instances said they compared the recorder numbers to those doing the death certificates and found they matched. They certainly weren't off by 90 percent.
Quote:
And so far you haven't demonstrated a case where with such a system in place only 10 percent of the deaths notices reach higher authorities.
What was the sytem of recording deaths in Congo before the surveys were made?
You tell us.
How many death certificates were printed? Who has the authority to print a death certificate? Does the LA Times answer those questions?
The LA Times didn't mention any other source for them other than morgues, hospitals and the health ministry. The LA Times article said "If a victim of violence dies at a hospital or arrives dead, medical officials issue a death certificate. Relatives claim the body directly from the hospital and arrange for a speedy burial in keeping with Muslim beliefs. If the morgue receives a body — usually those deemed suspicious deaths — officials there issue the death certificate." If you have proof that folks got death certificates without going to morgues, hospitals or perhaps the health ministry, provide it.
And by the way, the LA Times said "The morgue records show a predominantly civilian toll; the hospital records gathered by the Health Ministry do not distinguish between civilians, combatants and security forces. But Health Ministry records do differentiate causes of death. Almost 75% of those who died violently were killed in "terrorist acts," typically bombings, the records show. The other 25% were killed in what were classified as military clashes. A health official described the victims as "innocent bystanders," many shot by Iraqi or American troops, in crossfire or accidentally at checkpoints." But both Lancet studies concluded that most violent deaths were due to the coalition. Again, another indication of deliberate bias on the part of JH researchers.
