Iraq Poll Indicates 1.2 Million War Deaths

A household survey, conducted in Iraq by market research agency ORB, has indicated that about 1.2 million deaths have occurred from war-related causes since 2003. Like the previous Lancet surveys, these figures represent an estimate of total deaths, not just civilians.

The convergence between this survey result and a projection estimate calculated by Just Foreign Policy is striking. The Just Foreign Policy estimate is derived by extrapolation from the Lancet 2006 mortality survey (pdf) combined with trends in media reported fatalities, and today stands at 1,044,607. Taking into account the probable size of the survey confidence interval, the results are very similar. The poll also found that 48% of war-related deaths had been the result of gunfire, a figure very consistent with the 2006 Lancet survey.

However, despite the convergence of the two results there are, in our opinion, a few issues with the survey as it is reported:

• The methods employed in the survey are not described in the detail that would be expected for a scientific publication. It is therefore not easy to understand, for example, how they constructed their sampling plan and subsequently applied sample weighting during analysis.
• Confidence intervals (a measure of statistical certainty) for the reported percentages are not given.
  
• The number of injuries recorded is actually less than the number of deaths (1.1 vs. 1.2 million . This result is not what would be normally expected unless the respondents are only reporting very serious injuries. The definitions used in the survey are not made clear.
• Karbala, Al Anbar and Irbil were not included in the survey for security and administrative reasons.

Therefore, while the results are of the expected order of magnitude and very consistent with previous surveys, a degree of caution is required in interpreting their significance. However, in any event, the existence of a humanitarian catastrophe in Iraq remains irrefutable.