Many experts and expert groups from a range of fields are attempting to combine their knowledge to understand the lethality to Iraqis of the invasion and post-invasion violence in Iraq.
This is a slightly abridged and amended version of an invited "meta-analysis" of IBC's potential contribution to that understanding, presented in a closed meeting of the Ad Hoc Expert Group on mortality estimates for Iraq, convened by WHO in Geneva, May 2007.
Completeness revisited (b): Number of reports per incident
- If far more incidents involving mutliple deaths have gone unreported than were reported by the media, then we would also expect to find many such incidents which were just barely reported (at most in one or two sources).
- However we find that, based on 15,392 reports covering 3,988 incidents since July 2006, the more civilians are killed in any given incident, the more reports it attracts.
1 See Reality Checks: some response to the latest Lancet estimates for an earlier discussion of this issue. (16 Oct 2006)
Some of the claims that IBC could be a gross undercount require that the media miss many incidents in which there are multiple deaths (since they posit that their estimates are substantially comprised of these types of incidents, eg., car bombs).1 We can use the IBC press and media archive to test this claim. Are there subtantial numbers of incidents involving multiple victims which were “almost unreported,” in the sense that they appeared in only one, or perhaps two, sources?
To answer this question we analysed some 4000 incidents occurring between July 2006 and March 2007, and found a clear relationship between numbers killed and number of reports: the more civilians that were killed in an incident, the more reports it received. This relationship is shown in the following graphs.