A look at the Associated Press Review of the Afghanistan Civilian Casualty Count

On 11 February 2002, the Associated Press put out a "review" concerning civilian casualties in Afghanistan. In it, it purports to rebut other estimates that have been put forth to date. The AP tally is 500 to 600. Using the numbers given in the article—it is uncertain if they are the only numbers involved in the range but they are promoted as being accurate counts as far as the AP is concerned—the number is 526. The article brings up some interesting things. First, of course, is the assumed impeccability of the AP sources and implied weak credibility of anyone else. This is found throughout the piece.

An interesting bit from the report that gives a glimpse into the way casualties were counted can be found in a later paragraph (remember this for later):

Kama Ado, a village of 300 people living in mud huts in the Tora Bora region of eastern Afghanistan, was bombed on Dec. 1. An AP reporter who visited five days later saw widespread destruction and 44 graves, but villagers said it was so difficult to dig up the hard ground that some of the graves contained the fragmentary remains of four or five people. Witnesses put the village's dead at 155.

If such a case is true—this is winter in Afghanistan—then one must reconsider just how accurate the tallies of the AP are (and they are implied to be so). The AP did assign 155 casualties in the final analysis, but the fact that they were "fooled" at least initially should be noted and calls into question just how much their claim to accuracy can be trusted.

If the AP is able to question others' ability to accurately assess the numbers (which it does), it should be held to the same standard. And despite the numerous caveats that the number "will surely rise as more exhaustive tallies are compiled by independent bodies" and that accounts are "by no means complete," the AP numbers are being presented as currently the most accurate. (Early in the piece, the AP notes one instance where an "independent aid group" gave a casualty count of 67, while they gave it 70—just to "show" that some unnamed group has even lower numbers, therefore the AP is obviously fair and necessarily not undercounting.)

The story mentions economist Marc Herold (who gave the December estimate of 3000 to 5000 and currently offers 3100 to 3800). More than anything else, it is his numbers that are being challenged. As for credibility, the article states that he has "written essays on social and demographic issues in Afghanistan," making him no mere rogue economist. He also admits his sources are secondhand (so are many of the AP sources, unless they have some unprecedented access that no one else has) but ones he trusted. The implicit assumption being that the AP sources are necessarily trustworthy and his inherently suspect. And Herold's numbers are what the AP is attacking.

It seems the majority or all of the reporter quotations in the story are from the official news agency of Afghanistan, talking about the way the Taleban inflated casualty numbers. Even if they are fully trustworthy (and they very well may be), where are the other western journalists who were also eyewitnesses? They were there and they were not always led around by the Taleban or mindlessly reporting spoon-fed information. Journalists from Britain and other nearby areas who have spent a good deal of time in the region, who know the people, customs, even languages in some cases.

Perhaps only AP reporters can be trusted and those from various European papers or other sources are inherently biased or too naive or stupid to take accurate estimates or view Taleban counts with necessary skepticism.

Back to Herold. It is also suggested that he used Taleban or Taleban influenced sources. Not explicitly though, that would be too easy. The article brings up the Project on Defense Alternatives study (whether they lean to the right or the left should be irrelevant if the data are accurate), stating that its "counts were significantly lower," adding that "the study was based on selected Western media; it discounted reports based on Taliban figures." Since the AP story calls these sources into question, just what are they and what do the reports say?

The PDA report claims it used a "stricter criteria to screen these sources and correct for likely reporting errors and distortions" and does so primarily by using only western news sources: Reuters, the Associated Press, Agence France-Presse, BBC News, the Independent, The Times, and the Guardian." This is because "these sources seemed more attuned to the issue of civilian casualties than were US newspapers, while also being disinclined to accept on face value official Taliban reports or accounts from the Pakistani press." That the Guardian's numbers (see below) exceed even Herold's at the high end of the range is interesting to note.

Rejecting the Pakistani press out of hand (while Taleban tallies were usually reported through Pakistani sources, complete rejection seems overly strict) and other reporting from nonwestern sources (note: the AP is an American news organization that was founded in and is based in New York City). Though it held eyewitness testimony as usually reliable, it often rejected any reports of deaths from non-family members (they "were not taken at face value"). Since the criteria used were deliberately strict and deliberately left out much testimony from refugees and others when the deaths did not occur in the family, a lower count would be clearly expected (at 1000 to 1300, a count still at least twice the AP current estimate—and released over three weeks earlier; updated a week later).

Also, a companion piece to the study ("A critical appraisal of Operation Enduring Freedom and the Afghanistan war"), the PDA concludes that there were a "minimum of 3000 civilian deaths attributable to the impact of the bombing campaign and war on the nation's refugee and famine crises." They admit to be "seeking to err on the conservative side." Based on NGO reports and reporters who visited the camps (post-Taleban), I'd say this is a seriously low count. It also puts the civilian casualty count, according the PDA at 3000 to 4300. Though indirect deaths seems not to enter into the other reports, making it more of an interesting contextual note in examining the AP article ("why only count deaths?" seems an appropriate question).

Herold used a great number of sources: Indian daily papers, three Pakistani dailies, the Singapore News, British, Canadian, and Australian papers, the Afghan Islamic Press, Agence France-Presse, South African Broadcasting Corp. News, Pakistan News Service, Reuters, BBC News Online, Al-Jazeera, and US papers. He also used NGOs (non-governmental organizations) like RAWA (Revolutionary Association of the Women of Afghanistan) and Emergency Italia.

He was hardly indiscriminate in using his sources and looked for corroborating reporting from other sources. Most of the listings for the "daily casualty count" have multiple citations. Unless one is to believe that all of most of his sources are tainted and have been duped and lied to by the Taleban (reporting the "information" without question), it seems the charges are, at best, exaggerated. That many of the sources match with those used by the PDA report is notable. A report that, again, gives higher numbers than the AP (it seems the report is only being used to discredit Herold's numbers, anyway).

But since he does use Pakistani sources (rejected), other Arab sources (apparently necessarily unreliable), and Indian sources (which are unreliable why?), much of his data do not count. Since it is Herold's numbers that the AP report is calling into question, the fact that the PDA contradicts its own numbers—which goes without comment—is suggestive.

One of the cases noted is an 11 October 2001 bombing at Karam village. Three days later, on the fourteenth (the report doesn't even give the date of the actual bombing, the original story gave the day of the week), Taleban took journalists to view the site where it claimed two hundred civilians were killed. According to the AP, "reporters counted 35 graves, and villagers said 20 others had been taken to ancestral villages for burial," giving them a count of fifty-five.

Recall the excerpt from above about multiple bodies in graves. Remember that it was already three days later. Even more, note that someone from BBC News interviewed villagers who claimed two hundred were killed and that "Taleban radio asserted that the US was deliberately targeting civilians and had totally destroyed Karam, killing 170 people." (Seems like a lot of "rounding" of numbers going on.) Perhaps they were all lying. Perhaps someone else destroyed the village.

I tracked down the original story from the AP. It gave no figure other than the claims and that "it was difficult to assess claims of casualty figures three days after the attack. Muslims traditionally bury their dead quickly." It does mention doctors at the hospital in Jalalabad had been treated twenty-three victims of the bombing. At least that's "what they said" (according to the AP story).

At what point the AP came up with their "official" count is unclear. And the only sources they seem to cite are themselves, begging the question as to just what are acceptable sources to the AP?

And all the counts should be expected to be low. Remote areas have been left out of all the counts. Even with Herold, much of refugee testimony cannot be tallied since, like so many reports, estimates are lacking specific numbers (even within a range) and often contradictory (which does not mean there were no deaths). Also none of the studies take into account deaths indirectly related to the bombing campaign, let alone any non-death casualties.

There are other (unmentioned) sources like the London Guardian, which "says experts and informed sources put the total deaths of innocents at between 2,000 and 8,000" (I'd have to say the high end is a bit extreme, if we are only discussing "deaths"). And while the United Nations has not put out any official figures, a source from Kabul said that "It is definitely in the four figures" (both quoted in the Taipei Times).

Let's make some estimates of our own, recalling the 500 to 600 range given by the AP (also recalling that the numbers from the article already put the number at 526). The article's figures for Herat seem to be just plain wrong. It lists 18, while

According to official data of local demining organizations and the Regional Hospital in Herat, 38 deaths and an unknown number of injured people due to cluster bombs have been registered so far. However, some doctors in Herat Regional Hospital believe this number is much higher. In the village of Qala Shaker near Herat city alone, 12 people died and more than 20 were injured due to cluster bombs.

From the Médecins Sans Frontières website—these numbers only reflect casualties from the cluster bombs. Though perhaps once the bomb has landed, later casualties no longer count. Or maybe they weren't sanctioned AP sources and simply can't be trusted. Add the difference to the "official" Herat count and that would knock the AP tally to 546.

On 20 December 2001, US planes attacked a convoy of trucks that they claimed intelligence told them was transporting al-Qaeda and Taleban leaders. It wasn't. They were tribal elders on the way to Kabul. As many as fifty to sixty may have died on the road and surrounding village (some only count the dead in the convoy). According to other sources (the Guardian, for instance). The AP puts the count at twenty-seven ("at least"). Let's arbitrarily drop it to forty. Now the AP "tally" hits 586.

On 27 December, while bombing the house of a Taleban leader, as many as twenty-five villagers died. Being that Reuters was quoting the Afghan Islamic Press, I suppose we're supposed to dismiss it. The list in the AP article doesn't include it. Earlier reports gave as many as forty dead. Say we cut it to fifteen. 601.

Then there was the bombing at Qalaye Niazi on 29 December where people had gathered for a wedding ceremony. Retreating Taleban had told the people to store some weapons there. When the regime fell, the villagers notified the "authorities" about the cache. No one came to pick it up.

When the bombers came, over hundred friends, relatives, and villagers were there, many still asleep. Various reports of casualties came out, the lowest (almost) being from a villager who guessed thirty-two. Staff at a nearby hospital said it was 107. The UN report gave it as fifty-two—including ten women and twenty-five children.

The initial claim by the US was "Follow-on reporting indicates that there was no collateral damage." The AP assigns "at least 18...confirmed." It also said they "saw 35 destroyed houses in the area, but some of those living in them had already fled." Which certainly suggests a higher count. Going with the UN count, which falls in between, that puts it up to 653.

It was another case where US intelligence was misled. That should not mitigate the circumstances. To do so is to stop considering them people and making them the nonhuman ciphers of euphemisms like "collateral damage." When one wields weapons of mass destruction (which a bomber most certainly is, whether or not using the catch-phrase for a non "official enemy" makes some cringe), one had better be a lot more certain of what is being aimed at and what will be hit.

That leads to another interesting part of the AP article (quoting General Tommy Franks, the US war commander): "This has been the most accurate war ever fought in the nation's history." This is a claim we've heard before and later found to be inaccurate. One might bear in mind the International Committee of the Red Cross buildings destroyed and that the first four confirmed civilian casualties of the US bombing campaign were people working for the United Nations for the purpose of land mine removal. I wonder if the AP counted them in their estimate. If not, 657. 658, counting the woman from Mazar-e-Sharif who was killed when "an aid package of wheat, blankets, and cold weather gear crushed her house with her inside" (sid).

Also on the accuracy angle: the PDA study (which is really a study on accuracy, comparing the 1999 NATO bombing campaign over Kosovo and Serbia with the campaign in Afghanistan) concluded that "initial field reports suggest that the Afghanistan bombing campaign directly claimed more than 1000 civilian lives—or more than one for every 12 bombs or missiles expended." It gave the Balkans estimate as one per forty-six.

Another incident left out of the AP article is the count from the raid that took place on 23-24 January 2002. Between fifteen and twenty-one people were killed (the AP article says nineteen) and twenty-seven people taken into custody (said to be al-Qaeda and Taleban but released later when it was determined they were neither). Another intelligence error (either just wrong or misled again) by the military and the CIA. The article seems to be reluctant to add the count, saying that the "Pentagon is now investigating the possibility that its team received bad information and killed the wrong people." Since it's claimed they were anti-Taleban fighters, it would be considered "friendly fire" rather than dead civilians. If they can be considered "civilian," it'd put the tally up to 677.

There also is the three disputed deaths after a remote controlled plane shot two missiles at what were initially called al-Qaeda leaders meeting on a hill—even optimistically claiming one might have been Osama bin Laden. Later, local people claimed they were peasants out in search of scrap metal. The government promised to investigate. The attack occurred on 4 February 2002. On 11 February (interestingly the day the AP article was put out), the Washington Post ran a story on the missile attack. It reported that one of its own reporters tried to view the site and get information. He was "held at gunpoint by U.S. soldiers and prevented from entering the site." He was told "this is an ongoing military operation," by the soldiers' commander ("who would not identify himself"), and "if you go further, you would be shot."

He was also not allowed to visit the village where the alleged peasants were supposed to live. I was going to have to put this in the "we don't know, yet" pile but today (17 February) the New York Times ran a story confirming that it was three civilians. 670 (counting the January incident).

Those numbers exceed the AP review even before it was released. And only look at certain incidents and situations. There are numerous other reports and claims that haven't been fully reviewed and the noted numbers of people that undoubtedly died in areas where there were no journalists or reporting, people who were buried preceding the fleeing of relatives, stories that may be accurate from refugees. In the end, it appears that the AP tally is woefully low (even with its "caveats").

Should it be as high as Herold's or the range reported in the Guardian? Very possibly. Or maybe its closer to the PDA study. It isn't wrong to view the December estimate with skepticism, but it seems the AP was either using some form of ultra-skepticism or deliberately ignoring some of the plausible claims in order to push the "preferred" estimate that is much lower.

Why? It certainly is a count that would make its American readers more comfortable and maybe even its own reporters (I'd certainly like to believe the numbers were that low or lower but the evidence suggests otherwise). That can hardly be a criterion for such a "review," though. It's difficult to say just why the article chose to accept certain numbers while rejecting others and it would be overextending to claim it was deliberate or even propagandistic. But it does seem a bit curious.

Unlike both Herold and the PDA report, there is no explanation of sources or what was considered valid sources. The only explanation given is that it was arrived at through "examining hospital records, visiting bomb sites and interviewing eyewitnesses and officials." Other journalists did the same and came up with different numbers—but the AP story got all the press.

It can probably be agreed to by everyone that complete and fully accurate counts are far from made and that a better picture will evolve in the coming weeks or months (though true counts will probably never be known). It does seem a bit self-serving to play the report as if it the most accurate account while downplaying and/or ignoring most others. That the others are even mentioned is primarily to suggest their unreliability or to discredit each other, while pushing the AP numbers on the public.

Finally, it appears that all the counts (not just the three mentioned in this discussion) seem to be concentrating on only those who have been killed and ignoring or giving less weight to the many who were injured and wounded. As noted, indirect deaths (which would include wounded dying later from injuries or complications) are not counted, either. Nor, it would appear, family members of the Taleban who often live(d) in close proximity to where bombs fell. Or the ongoing starvation and exposure deaths—a humanitarian disaster prior to September 2001—that have been made far worse by the bombing campaign. A "casualty" includes them, too.

(When President George W. Bush claimed that the US has "saved a people from starvation" during the State of the Union Address, I felt sick. I'll be eagerly awaiting his live Easter feed when the people at Maslakh serve him an expansive banquet now that they've been saved from starving to death. I'm sure they are looking forward to it, too. He's eating pretzels in the comfort of his home watching a football game while these people are living in holes or under tarps, waiting hours for gruel or bread. It must feel great to have been "saved.")

(Sources: the AP report is widely available, I used www.washingtonpost.com/wp-dyn/articles/A58253-2002Feb11.html;
Herold report main page, www.cursor.org/stories/civilian_deaths.htm;
PDA reports from www.comw.org/pda/0201oef.html, www.comw.org/pda/0201strangevic.html;
MSF article from www.msf.org/countries/page.cfm?articleid=E755D9DB-D547-42F5-888C17D60135AE38;
AP "Karam" article from www.dominionpost.com/af/usstrikesback/2001/10/15/an;
"Karam" article from the BBC News, news.bbc.co.uk/hi/english/world/south_asia/newsid_1602000/1602179.stm;
On Qalaye Niazi, www.guardian.co.uk/Archive/Article/0,4273,4330175,00.html;
On the 4 February missile attack, www.washingtonpost.com/wp-dyn/articles/A55268-2002Feb10.html and www.nytimes.com/2002/02/17/international/asia/17TALL.html;
Also www.taipeitimes.com/news/2002/02/13/story/0000123850, www.alliancesouthasia.org/sept11/civilian_casuaties/casualties.htm and dozens of other news sites and wire reports)