Edwards points out the generally recognized fact that IBC's methodology - only listing deaths reported by two or more Western sources - likely results in their tally being a conservative estimate of civilian deaths. However, Edwards goes further by showing that there is a systematic source of bias in that Western news agencies are more likely to report deaths caused by "insurgents" than those caused by "Coalition" [aka, American] forces. Edwards reports on an examination of the IBC database for the six-month period from January through June, 2005. They found that, of 58 incidents involving at least 10 deaths, only one was attributed to US/Coalition action. Further, during this period, only 15 civilian deaths total were "attributed to 'coalition' airstrikes, helicopter gunfire and tank fire," a result that is completely implausible to anyone who has followed news of the repeated massive attacks by US and allied forces on alleged "insurgent strongholds."
Very disturbing was the tone of IBC's founder John Sloboda's response to being emailed a question about this potential bias. He implied that IBC had no bias in that it recorded all such events reported in the Western media, while ignoring Edwards' point that the Western media may itself have a bias in what it reports. He stated correctly that "We have always publicly acknowledged that our numbers must underrepresent the true figure." He then goes on to state "the question of by how much is one that exercises us, as it does many others." However, he gives no evidence of wrestling with this issue or of recognizing its overriding importance in evaluating what the IBC numbers tell us about the extent of Iraqi deaths.
IBC does, indeed, "acknowledge" the limits to their work. But, like many academic researchers throwing in a pro forma "limitations of the study," this acknowledgment is done in such a way as to give it little emphasis. For example, in their July, 2005 report A Dossier of Civilian Casualties 2003-2005, on page 24 of 28, in a discussion of why they use their "maximum" estimates [they cite a range of deaths from a "minimum" to a "maximum," but the "maximum" is of reported deaths, not of likely deaths], they state "even our max figure is likely to under-represent the full toll, given that not every death is officially recorded or reported." Certainly this language gives no indication that their maximum may, in fact, radically underestimate the true toll.
"We acknowledge that many parties to this conflict will have an interest in manipulating casualty figures for political ends. There is no such thing (and will probably never be such a thing) as an 'wholly accurate' figure, which could [be] accepted as historical truth by all parties. This is why we will always publish a minimum and a maximum for each reported incident. Some sources may wish to over-report casualties. Others may wish to under-report them. Our methodology is not biased towards 'propaganda' from any particular protagonist in the conflict. We will faithfully reflect the full range of reported deaths in our sources. These sources, which are predominantly Western (including long established press agencies such as Reuters and Associated Press) are unlikely to suppress conservative estimates which can act as a corrective to inflated claims. We rely on the combined, and self-correcting, professionalism of the world's press to deliver meaningful maxima and minima for our count."
This statement clearly suggests that the true figure is between their "minimum" and their "maximum," as these words would imply. Nowhere in this Limitations section do they acknowledge the problem of systematic bias due to reporting bias.
Despite the strength and originality of Edwards' critique, I believe that it does not give enough attention to another source of bias that is minimized by IBC. This is the difficulty in reporting from Iraq and the absence of Western reporters from most of the country. Iraq reporting has been dangerous from the beginning of the war, with several reporters killed by American forces during the invasion. By the fall of 2003, as Reuters reported, reporters were also at risk from rebel forces and, sometimes, the dissatisfied Iraqi population.
The situation became much worse after the April, 2004 rebellions across much of the country. Most of Iraq became off-limits to Western reporters. Since then, the situation has only deteriorated, with a number of reporters being killed and several kidnapped. To date, 79 reporters and media assistants have been killed and another two still missing; 37 media workers have been kidnapped, five of whom were subsequently killed. [Jill Carroll, currently kidnapped reporter for the Christian Science Monitor, is a current reminder.] Of course, in conditions of active rebellion, the safer areas accessible to Western reporters are likely to be those under US/Coalition control, where deaths are, in turn, likely to be due to insurgent attacks. Areas of insurgent control, which are likely to be subject to US and Iraqi government attack, for example most of Anbar province, are simply off-limits to these reporters. Thus, the realities of reporting imply that reporters will be witness to a larger fraction of deaths due to insurgents and a lesser proportion of deaths due to US and Iraqi government forces.
Of course, some Western sources are able to report stories generated by Iraqi reporters they hire, as IBC pointed out in their July, 2005 report:
"Current reporting is increasingly undertaken by Iraqi staff working for western media outlets, with Iraqi names now appearing more regularly as authors or coauthors. Western journalists have always relied on Iraqi assistants (drivers, interpreters, etc.). In a very real sense, therefore, the IBC database increasingly depends on the bravery and dedication of Iraqi media workers continuing to risk life and limb to inform the world about the situation in their country."
Given, as indicated in that report, that ten media outlets provided over half the IBC reports and three agencies [Associated Press, Agence France Presse, and Reuters] provided over a third of the reports, there is simply no reason to believe that even a large fraction of Iraqi civilian combat-related deaths are ever reported in the Western media, much less, have the two independent reports necessary to be recorded in the IBC database. Do these few agencies really have enough Iraqi reporters on retainer to cover the country? Are these reporters really able comprehensively to cover deaths in insurgent-held parts of Iraq? How likely is it that two reporters from distinct media outlets are going to be present at a given site where deaths occur? How many of the thousands of US bombings have been investigated by any reporter, Western or Iraqi? Simply to state these questions is to emphasize the fragmentary nature of the reporting that occurs and thus of the of the IBC database.
If IBC believes that the vast majority of Iraqi deaths are reported by the Western media and, thus, recorded in their database, IBC should provided an argument to that effect. IBC does not provide such an argument. Neither do they remind readers of these potential limitations in any way that would attract attention to them and decrease the ability of others to deliberately misuse the IBC numbers.
Producing "conservative," bottom-line estimates that are known to be inaccurate can be a useful research technique in some cases. Such figures help to remind readers that there is a phenomenon - in this case, Iraqi civilian deaths caused by the conflict - and to thus focus attention upon it. In the early days of the war and occupation, IBC's figures played such a role. They existed as a floodlight focusing attention upon the fact that the war and occupation, despite whatever merits they may or may not possess, had real costs in terms of lives lost that needed to be included in any moral reckoning of the rights and wrongs of the war.
Conservative estimates lose their value, however, when they serve to obscure, detracting attention from the true magnitude of the phenomenon. Thus, as the fighting has intensified and as other estimates of Iraqi civilian deaths have become available, IBC's low-ball estimates have increasingly been used to mask the true magnitude of the suffering, rather than as a call for better, more precise estimates. Such misuses of the IBC figures could only be avoided, or at least reduced, if IBC took every opportunity to prominently call attention to the fact that their estimates are nothing but rock bottom figures, almost certainly far below the true mortality figures. Indeed, a September, 2005 report by the Humanitarian Practice Network, Interpreting and using mortality data in humanitarian emergencies: A primer for non-epidemiologists, lists seven studies from which estimates of violent civilian deaths in Iraq can be derived. Since each study covers a different period and length of time, the results are standardized as "violent deaths per day." Of the seven studies, IBC has the lowest estimate, at 17 deaths per day, followed by 22 deaths per day estimated by the Iraqi Ministry of Heath. Two studies produce estimates of 50 and 56 violent deaths per day. The Lancet study leads to an estimate of 101 violent deaths per day, while two other studies generate even higher estimates of 133 and 152 deaths per day. Thus, as suggested by our analysis, the IBC estimates are far below those from most other sources and cannot credibly be taken as being anything but rock-bottom minimums. [Thanks to Les Roberts for calling my attention to this report.]
Unfortunately, rather than emphasize the extent to which their estimates are severe underestimates, IBC appears to deliberately undercut attention to the weaknesses in their tally. Thus, on June 6, 2005, the BBC quoted IBC's John Sloboda as saying "Everyone can agree that there are good reasons why our count can never be complete, but there is not as much confusion as you think. Since the end of hostilities was declared, we are confident in the figures."
Rather than call deserved attention to its limitations, IBC's reporting style exudes an illusory precision. Thus, page four of their report states "24,865 civilians were killed in the first two years, almost all by violence. 82% of those killed were adult males and 9% were adult women." No use here of those phrases like "were reported killed by Western media outlets" that would remind the reader that these numbers are not precise, but are rough estimates, potentially severe underestimates, of the real numbers. And how can one interpret a phrase like "24,865 civilians were killed" - rather than "around 25,000" - except as a precise number, again suggesting a precision, and a false accuracy, that the data surely don't warrant.
This type of false precision is especially disturbing since the IBC figures have been used by many in a campaign of disinformation, discussed at the beginning of Edwards' IBC critique, to counter attention being devoted to the epidemiologist Les Roberts and colleagues' October 2004 Lancet study estimating around 100,000 "excess deaths" from all causes [See my commentary: 100,000 Iraqis Dead: Should We Believe It? as well as the excellent Wikipedia discussion]. The IBC estimate also appears to be the basis, as Edwards suggests, for President Bush's recent claim that "30,000 more or less have died as a result of the initial incursion and the ongoing violence against Iraqis.
Robert Fisk has rightly taken the Western press to task for its "hotel room journalism." Fisk does not blame the reporters for doing what it takes to stay alive. [The recent news that ABC News' Bob Woodruff and cameraman Doug Vogt were critically injured provides a potent reminder that even famous network anchors are at risk if they actually try to do any reporting.] What Fisk does blame Western reporters for is pretending that their dispatches are based on actual reporting, rather than, as is largely the case, Green Zone press conferences and related propaganda efforts by the US and Iraqi governments.