1. The calculation is based on cumulative figures reported in the media rather than incompletely reported individual events on specific days and times.
Because reporters were barred from entering the city during the April 2004 siege of Falluja, IBC's estimate of 572 - 616 civilians killed during the siege is based on reported cumulative totals rather than a series of individual reports: these are necessarily incomplete even under "normal" conditions in the occupation, but under the restrictions just noted could only provide glimpses of the totality of events that took place inside Falluja. (Nonetheless IBC has gathered and made available to the public the most comprehensive collection of information related to the human impact of the siege in its IBC Falluja Archive, which catalogues many recorded incidents - from attacks on ambulances to sniper fire on children - as reported in nearly 300 news stories.)
2. The calculation uses figures reported by local hospital and NGO sources rather than those produced towards the end of April by the Iraqi Health Ministry
There were two competing "final totals" produced for the human toll of the siege. One set of cumulative numbers was derived from growing hospital and NGO figures which had reached 600 by the 12th and ultimately passed 800, swelled by deaths during a series of nominal "ceasefires" as well as by the gradual recovery of bodies buried in the rubble of destroyed buildings or in makeshift graves in private gardens. These numbers were reported widely in mainstream sources, and the first 600 deaths included a breakdown showing that 160 women and 141 children under the age of 12 were among the dead.
The other "final total" was produced by the Baghdad-based Iraqi Health Ministry (IHM), apparently at the behest of US authorities who had dismissed the locally-reported numbers as "somehow filtered through some of the local propaganda machines that are operating inside Fallujah" and said they would ask the IHM to "get a fair, honest and credible figure". (Gen. Mark Kimmitt, 12th April 2004).
The IHM announced their tally little more than a week later, on the 21st of April: 264 dead, including 28 children and 24 women. This number was incrementally increased soon afterwards to 271 and finally 280, but never reached much beyond about a third of the locally-produced numbers. The official who first released the IHM figure asserted that his widely-quoted colleague in Falluja had been under "political pressure" to inflate the numbers of dead, and that the new count had been obtained from the very same doctor; the claim of inflated numbers [1] was repeated the following day by US-appointed Health Minister Khudayer Abbas, who told the Associated Press (AP) that the death figures in Falluja
"were exaggerated for political factors. There was some parties and elements who were pressing on the people working in Fallujah hospitals to present exaggerated numbers." [AP 22nd April].
But the Falluja medics in question have yet to alter their tallies when directly interviewed - see for instance AP 30th April, Newsday 2nd May, and Knight-Ridder 9th May:
Dr. Rafe Hyad al Esawi, director of the Fallujah General Hospital, said the number of people killed exceeded 800 with more than 3,000 wounded. U.S. authorities say those figures are exaggerated, but al Esawi stood by his numbers and said they are growing because many people could not reach the hospital during the fighting to report deaths. "Some of the families buried their dead in their gardens," al Esawi said. "Now they are starting to come to the hospital to register."
Until recently both sets of numbers stood unrescinded by either party, making it difficult to choose between them without making subjective judgements (although the weight of evidence has always tended to support the local count, not least because of its plausible growth during the siege, but also the size of several temporary mass graves, as can be reviewed in the IBC Falluja Archive). It was also difficult to obtain a useful IBC estimate that integrated both numbers: the two sets of statistics diverge not just three-fold on the total, but six-fold on the number of women and children killed (301 vs. 52), which has meant that any range estimating the possible civilian count (including male civilians) could diverge even further, resulting in a final IBC estimate on the order of between "80 - 600."
A recent report however indicates that the IHM no longer includes any Falluja statistics in its official counts for April 2004. The IHM official who had produced the tally of 264 dead in Falluja had in the same interviews during April told reporters of statistics gathered for the same period in Baghdad (235 dead) and the rest of the country (57), making a total of 556 throughout Iraq between April 5th and the date of his first interview on the 21st.
But recently released statistics from the IHM show only 344 violent deaths from the 5th to the 30th of April throughout Iraq (except for three relatively untroubled Kurdish provinces), as revealed in a detailed 25th September Knight-Ridder report [2] providing monthly breakdowns of IHM statistics. We judge it inconceivable that this number of 344 still includes the data claimed for Falluja. Our judgement is based on three converging sources of evidence, which are analysed in detail in Note [3] below.
The sole conclusion consistent with all the evidence is that for some undeclared reason the IHM's April "Falluja count" no longer appears in their statistic of 344 for that month, and only the better-documented figures from Baghdad and elsewhere remain. It may be that the original figure, which was conceivably produced in haste by an office which had only just been brought into existence, is now considered unreliable, incomplete or inconclusive; in any event, as that number no longer appears to contribute to official records IBC cannot use it in its calculations (just as we would remove from our database any entry that had been solely based on a number later withdrawn by its source).
The only figure that remains viable for use is therefore the consistent and unaltered count which ultimately reached 800 and came from local medical sources. This is however a record of total deaths - IBC, given the priorities of its work, still needed to establish an estimate of the civilian dead among that number.
3. The calculation assumes that civilian males died in direct proportion to the reported totals of deaths of adult women.
Of the 800 reported deaths, a breakdown giving the number of women and children killed was available for the first 600 up to April 12th, most of whom were killed before there were any real opportunities for them to evacuate the city. Because the proportion of women and children killed was exceptionally high, we conservatively assumed that civilian adult males would have been as badly affected as their female counterparts, and accordingly assigned to them the same number of deaths as recorded for women (160 each). This is conservative because there are reasons other than participation in combat for men to be killed in higher numbers than women and children, including that they more commonly venture outdoors where they may be exposed to cross-fire, and are more likely to be mistaken for combatants by military forces. This still left a "surplus" of 139 adult males killed - some 23% of total deaths, and 46% of all adult male deaths. These we designated as having been insurgents, whom we therefore excluded from our count.
This left 200 "undefined" deaths which were recorded in Falluja from April 13th onwards. Six unambiguous deaths of women and children recorded in the weeks after the 13th were identified in the news stories collected in the IBC Falluja Archive, and subtracted from the 200, leaving 194. For the remaining "undefined" 194 deaths we assumed that the proportion of fighters among the dead remained at at least 23% of all deaths to the end of the siege, but because areas of the city were being emptied of women and childrenin a mass exodus (but not "military age males", who were forbidden to leave), we held that the proportion of fighters may have climbed toward 46%, their proportion in relation only to other adult males killed.
Reducing 194 by 46% and separately by 23% produced 105-149 as an estimate, expressed as a range, of non-combatant civilians in the "undefined" 194. To these numbers the six deaths of women and children known to have occurred after the 13th of April were then added, producing 111-155 as the range of civilians among the 200 whose deaths were recorded after the 13th. These could then be added to the 461 earlier extracted from the pre-13th count of 600, resulting in a final IBC Min-Max for x360 and the siege of Falluja of 572 - 616.
Additonal Notes:
[1] That Iraqi hospitals often exaggerate casualties is a standard claim by US military officials, but this is merely asserted or conjectured, and has never been demonstrated nor supported by evidence of any kind. In fact the usual "inflater" of casualty statistics is the US military, although these are always only of the "insurgents" or "bad guys" killed - the LA Times reported on 29th April (IBC link), for example, that "U.S. officials believe that the Fallouja insurgents have raised their profile militarily, but have set themselves up for a larger defeat... [and that] the fighters already have suffered 1,500 to 2,000 deaths, by U.S. military estimates." We judged the claim by Iraqi Health Ministry officials, despite the political context of their statements, to require taking more seriously.
[2] These may be the last such reports we see from the unit responsible for collating casualty statistics within the IHM: immediately after the date of the detailed and extensive statistics obtained by Knight-Ridder (all of which, from May onwards, are consistent with other reporting, and which had also been widely released in summary form to other major media during September), the unit was told to stop speaking directly to the press. [AP 23 Sept.])
[3] Reasons for discounting the IHM estimated total of casulaties for Falluja in April 2004:
Firstly, the IHM September total of 344 cannot include both Falluja and Baghdad, the other major location for deaths during April. We have already mentioned that the official who had released the count of 264 cited another 235 in Baghdad, and a total of 556 country-wide by April 21st. A day later his superior, Health Minister Khudayer Abbas, told the press that according to the IHM's nation-wide compilation, 576 "insurgents and civilians" had been violently killed in April throughout Iraq, Falluja included [AP 22nd April]. So, between April and September, more than 230 deaths have been removed from the April count without explanation.
Secondly, the Baghdad city morgue alone recorded more than 300 deaths due to gunfire and explosions throughout April - not unusual for this violence-wracked city since the invasion - and the morgue figures would be well-known to the Health Ministry. A 24th May AP survey of Iraqi municipal morgues from which this figure is taken, and which was able to obtain data from just four of Iraq's 18 provinces (partially dealt with in IBC database entries x334 and x350), including Baghdad but not Al Anbar province, where Falluja is located, counted 413 deaths in April. There are thus, even within a partial survey encompassing just 4 Iraqi provinces excluding Al Anbar, more recorded deaths than IHM's September-released statistic of 344 for April which covers 15 provinces including Al Anbar. These IHM totals must be missing Falluja.
Thirdly, the cessation of hostilities (albeit temporary) in Falluja immediately after the withdrawal of US ground forces at the end of April means that the city's contribution to IHM statistics for May would be negligible. However instead of a reduction in the number of deaths, the recently published IHM numbers jump more than two-fold from 344 in April to 749 in May [Knight-Ridder 25 Sep]. This is inconsistent with all other reporting. May, too, was violent, and the statistic of 749 is not unrealistic, but April is widely acknowledged as one of the most violent of the entire occupation - not least because of the siege of Falluja. In fact, if we subtract the IHM's highest April count of 280 for Falluja from the newer IHM statistic of 344 for April - leaving 64 - and accept that Falluja contributed little or nothing to IHM statistics during May, then the rise in violence throughout Iraq (excluding Falluja) from the end of April to the end of May is more than 11-fold. There is no independent evidence which would remotely support such a sharp rise. A tally comparing April 2004 with May 2004 derived from the media-reported incidents in IBC's public database suggests a ratio of approximately 1:1 between these two months, rather than the 1:11 which would be implied by the IHM figures.
If details later emerge that are of relevance to any of the calculations above, IBC will adjust its figures accordingly.