Commentary on number of reports

• There are far more media reports of civilian deaths than a casual observer might expect: IBC collects, analyses, and aggregates, around 90 reports per day.
• Most reported deaths attract multiple corroborating reports. Half of the deaths in the IBC data base are corroborated by 5 or more additional reports.
• Only 14% of deaths were reported by a single source, and these tended to involve incidents with 3 or less deaths.
• The implications are that unreported events are likely to involve few victims per event.

If, as this suggests, unreported deaths will tend to be mainly “small-n” incidents, involving one or two people, then even a very large number of such unreported deaths will raise the final death total by a relatively small proportion. If the death-toll of the average unreported incident was 2, it would have needed about 10,000 unreported incidents (to add to the 4,000 reported ones) to double the reported death toll for the months July 2006 to March 2007.

If, as some claim, the actual death toll is more than ten times higher than the media has reported (in aggregate, and including its reported deaths from official sources such as morgues), then there would need to have been 90,000 unreported incidents over that same 9 month period (that is 10,000 unreported incidents a month, or 330 unreported incidents a day). Even in the difficult conditions of present-day Iraq, it seems hard to countenance that all official and organisational surveillance mechanisms combined would be failing to register 96% of fatal violent incidents occurring, including in the form of after-the-fact bodies found.