Pretty random?  Three threads on the exact same subject in 3 or 4 days when I've never seen threads disappear before?  Seems to be a pretty clear pattern emerging.  Anyways, given that I don't like retyping, I saved my post and am now ready to react to hacking on a moment's notice :)

 

Eu4ea,

 

You can easily find the CIA figures by Googling “CIA Factbook death rate 200x”. It is interesting how you demand links and yet when you quoted a CIA figure of 5.0 (one which I couldn’t find or verify) there was nary a link. In any event, I’ll provide the 2006 link – https://www.cia.gov/cia/publications/factbook/geos/iz.html - you’ll find the other years I quoted in one of the first five hits from Google when you use the search string I provide above (substituting the specific year for “x”). However, haggling over links really isn’t important to my argument. 

 

What is important is the fact that you endorsed a CIA figure and now you want to discredit a CIA figure. You cannot have it both ways, unless you can demonstrate that the pre-war methodology was satisfactory and that there was some change that caused the methodology to be incorrect. Given that you are asking for the methodology, it sounds like you had no basis to endorse the pre-war figure.

I do not know the methodology that the CIA used, and given that we didn’t have agents in Iraq pre-war, I really can’t imagine how a black box figure coupled with a lack of agent infrastructure in Iraq is that defendable. So, I find the CIA figures pre-war to be dubious (and while we certainly have the agent infrastructure now in Iraq, the post-invasion figures are still black box, and so I don’t put weight on them as well).

 

So, this leaves us to question the fundamental underpinnings of the Roberts et al numbers. They are based on comparing the mortality rate of pre-war Iraq with post-invasion Iraq to derive the “excess” death figure. However, when you benchmark their pre-war numbers, they underestimate these numbers. Due to this, the published results overestimate the “excess” deaths figure. For example, if we use the UNPD numbers from below, you could estimate that they overestimate the “excess” death figure by nearly 281K alone by underestimating the pre-war death rate. What is even more important about this is that the 95% CI now moves the lower limit much closer to 0 (it would be approximately 112K if the variance were to remain the same). Thus, the finding of significance at the 5% level becomes much more sensitive to rejection if consistent biases exist in the study (i.e. the dropping of 3 clusters where it is reasonable to assume to that the figures would be biased towards overestimating the “excess” death figure based on the geographic location). 

 

I will quote from some comments made on the econ blog Crooked Timber that were from the original survey.

 

”link

 

Comment 99. “But there are other numbers for the before the war scenario. In the 1980s, the last time international agencies could check any real numbers, Iraq’s mortality rate was always in the region of 7 to 8 (per 1,000 people). According to the World Bank’s “World Development Indicators” link report, the 2002 figure for Iraq was 8 as well. And if you look at some other reports from various reports, you will find that most reports put the figure somewhere around 8.

 

Comment 106. Some more data on the mortality or death rate per 1,000 in Iraq before the war: Source: UN Population Division & Unicef Downloadable for example at: pdf.wri.org/wr98_hh2.pdf Mortality rate in Iraq: 1975-1980: 8.8 1995-2000: 8.5 1980-1991 doesn’t produce any good data because of the Gulf wars; “smart sanctions” were introduced in 1996, so sanctions don’t have a great influence on the data (beyond what was the government’s responsibility). Average figures are especially valuable as they eliminate statistical outliers. I think it can be safely said that Iraq’s “natural” mortality rate is somewhere around 8 per 1,000. Of course, Les Roberts et al. think that the mortality rate in 2002 was 5 per 1,000. Here’s the challenge: Can anyone find anywhere in the academic literature and scientific databases a source which would correspond to Roberts et al finding of 5 (+/- 10%) for a year before the Iraq invasion? All the statistics for various years before the invasion lie somewhere around 8, which is the best scientific data we’ve got so far. Up to now, Roberts et al are the only one who claim it was 5. Now the question is, which claim looks more likely, is based on more solid methodological grounds, is validated by reiterated tests, and provides greater overall consistency with other comparative historical and cross-country datasets? I think it’s pretty obvious that Roberts’ sample is the outlier here, not all the other studies before.

 

To overcome this weakness in the study, you’d have to find some academic literature that finds a crude mortality rate similar to the Roberts et al finding of 5.5 per 1000. Good luck, as I have yet seen anybody provide these. 

 

The countdown is on......

 

 

"Random"?  Perhaps.  This has happened not two months ago with YOUR topic about civil war.  In fact, I asked you about it, as it occurred while you were on vacation, if you'll recall.

 

Don't take this as a suggestion that I think you're up to something nefarious.  Not by a long shot.  Still, something's up that's inexplicable.  Sure hope SYSOPS comes clean.

Eu4ea,

 

A bug that specifically targets Lancet threads while leaving others on the same board intact?  Not so random in my book.  Maybe the CIA analyst that runs the crude mortality numbers didn't take too kindly to my words ;)

 

Shek

EW3,

 

In one of the MIA threads, you had asked about cluster sampling.  I couldn't find any good internet sites that explain it in detail well.  If you'd like, I could email my notes which include some of the basic derivations that prove that under cluster sampling you'll get much larger variances than you do with simple random sample.  If so, I'll have to scan them in at work next week.

Shek

catch me at geronimo_skipper@hotmail.com  

Eu4ea,

 

Two questions for you (in addition to the original benchmarking question I posed):

 

1.  Roberts et al spent nearly an entire paragraph on page 7 of the study benchmarking their "excess" deaths figure to the Congo, Vietnam, Timor, and the Darfur (although they don't really discuss at all motivating why the gross figures from these conflicts provide a proper benchmark).  From this paragraph, it is clear that they understand that benchmarking is a standard in statistical analysis.  Given this, why do you think they make no attempt to benchmark their pre-invasion figure given that there are published figures out there to benchmark against?

 

2.  Why does Roberts et al only use approximately 14 1/2 months of pre-invasion data points given that they argue that their findings, to include their 2004 findings, don't suffer from recall bias:

 

“October 16, 2006: The British medical journal, The Lancet, has again turned over its pages to political propaganda pretending to be science. The latest report claims that a very flawed survey of the Iraqi population proves that military and terrorist operations have killed over 600,000 Iraqis in the past three years. Several things should be noted. First, the normal death rate of the Iraqi population would leave about 550,000 dead since early 2003. Second, the terrorist, and counter-terrorist, violence in Iraq is largely restricted to four of the 18 provinces. About a third of the population is involved, mainly because Baghdad is a principal battleground. But the Lancet study implies that a third of the population has suffered these losses, which means over seven percent of the people living in that area would have died since 2003. That's a lot of bodies. Where are they? Where are the standards required for statistics and data in a study like this? No matter, the Lancet did a similar study in 2004, just before the U.S. presidential elections. That study was eventually discredited, just as the recent one will be.”

 

In my post in the Lancet threat - that is apparently being held captive in Area 51 – I posited that half the population was suffering these “655,000” killed. That was simply a ballpark wiff but Strategy page has provided better totals with the same insight. A region with 7.5% killed – violently – would make for some significant migration figures. In my city of 200,000, 7.5% would mean 15,000 headless bodies turning in trailer parks, hundreds of potholes with adjacent shrapnel damaged buildings and a thriving u-haul franchise. Not that Baghdad is not suffering many deaths, but when does a level of killing impact society to a major degree. 

 

Dunnigan has written that military units will break after a predictable number of casualties. When does a society break? I am most interested in an analysis of the effect that the casualties and anticipated duration will have on Iraq. Maybe four million more exit visas for Sunnis will quell the violence.  

David Frum - writer of the "axis of Evil " speech, specululates on the actual interviewing techniques by the locals that the survey team hired to collect the data:

 

  link

Do we have reliable data for Anbar province?  Does the Sadrist Health Ministry keep stats on the Sunni death toll?  Name for me the government's liaison to Anbar, or the morgue directors for Samara, and Ramadi?  I doubt we have any hard data out of the most violent province in Iraq.  I would also dispute the notion that only 4 provinces are violent; Basra has as many civilian deaths as Diyala (one of the four violent provinces), and only slightly less violent than Anbar, according to Brookings.  Does that seem accurate to you?  Wouldn't Anbar be twice or three times or more violent than "stable" Basra?

As for the bodies argument, we still haven't dug up all of Saddam's bodies, nor is there a full accounting for Rwanda.  Speaking of Rwanda, the genocide there killed 937,000 people according to a census using population samples (not an actual counting of the bodies), and resulted in ten percent of the total population being killed in just four month.  Most were Tutsi, which had according to some demographics 77 percent of their population wiped out.  Yet the Tutsis survived and so did Rwanda.  By comparison 7.5 percent of the Sunni population is not that significant. 

 

link

 

 Third World countries have high birth rates which lead to positive population growth, Rwanda quickly recovered form its genocide and has a 2.5 percent growth rate.  The DRC has a growth rate of over 3% even with a violent war that has killed over four million people, its death rate is 13.27 per 1,000 or 822,740 per year (according to CIA).  That includes 100,000 AIDS deaths per year.  But CIA's death rate is underreported; for Iraq it lists 5.37 deaths per 1,000 population, which translates to 14,000 deaths per year (from all causes), we know that at least double that many Iraqis have already died this year through violence alone/  By contrast we can look at the death rates for two other Muslim Third World countries.  Egypt, which is relatively stable, (little or no violence) has a death rate of 5.23 per 1000 pop.  Afghanistan by contrast is listed with a death rate of 20.34 per 1000.  It is difficult for me to imagine that Iraq would be considered as stable as Egypt is , and that the violence in Iraq does not lead to a spike in the death rate that would be more than the average Third World Muslim country.  Also it is hard to believe that Afghanistan would be a more violent place to live than Iraq. Of course there are many more diseases in Afghanistan than Iraq, but there is a lot more violence in Iraq than Afghanistan.