Jousting with the Lancet: Pajamas Media Interviews Professor Gilbert Burnham
The second paragraph of my last post came out garbled (seemingly because of “greater than” and “less than” symbols confusing the computer), so I’m posting it again:
I can assure you that I read the Burnham study very carefully in its entirety. I found Figure 2 disconcerting. It is a colored map of Iraq and its governorates, with the colors specifying the rates of violent death per 1000 population per year, in three categories: low (less than 2), medium (2-10), and high (more than 10). It was impossible to determine from this what the actual breakdown was. To add up to 600,000 violent deaths, it could have been (for [ low: medium: high], based on the respective populations in Table 1) [50,000: 370,000: 180,000], or it could have been [0: 80,000: 520,000]. The latter extreme would say that nearly all the violence was in Anbar, Ninewa, Salah-al-Din, and Diyala provinces – a very different picture from the former extreme. I suspect that the real data may be closer to the latter than the former, and there might be outliers like the Fallujah cluster of 2004 fame. That’s why it would be useful to see the details.
