IMPEACH BUSH

The U.S. military's claim that violence has decreased sharply in Iraq in recent months has come under scrutiny from many experts within and outside the government, who contend that some of the underlying statistics are questionable and selectively ignore negative trends.

Reductions in violence form the centerpiece of the Bush administration's claim that its war strategy is working. In congressional testimony Monday, Army Gen. David H. Petraeus, the top U.S. commander in Iraq, is expected to cite a 75 percent decrease in sectarian attacks. According to senior U.S. military officials in Baghdad, overall attacks in Iraq were down to 960 a week in August, compared with 1,700 a week in June, and civilian casualties had fallen 17 percent between December 2006 and last month. Unofficial Iraqi figures show a similar decrease.

Others who have looked at the full range of U.S. government statistics on violence, however, accuse the military of cherry-picking positive indicators and caution that the numbers -- most of which are classified -- are often confusing and contradictory. "Let's just say that there are several different sources within the administration on violence, and those sources do not agree," Comptroller General David Walker told Congress on Tuesday in releasing a new Government Accountability Office report on Iraq.

Senior U.S. officers in Baghdad disputed the accuracy and conclusions of the largely negative GAO report, which they said had adopted a flawed counting methodology used by the CIA and the Defense Intelligence Agency. Many of those conclusions were also reflected in last month's pessimistic National Intelligence Estimate on Iraq.

The intelligence community has its own problems with military calculations. Intelligence analysts computing aggregate levels of violence against civilians for the NIE puzzled over how the military designated attacks as combat, sectarian or criminal, according to one senior intelligence official in Washington. "If a bullet went through the back of the head, it's sectarian," the official said. "If it went through the front, it's criminal."

"Depending on which numbers you pick," he said, "you get a different outcome." Analysts found "trend lines . . . going in different directions" compared with previous years, when numbers in different categories varied widely but trended in the same direction. "It began to look like spaghetti."

Among the most worrisome trends cited by the NIE was escalating warfare between rival Shiite militias in southern Iraq that has consumed the port city of Basra and resulted last month in the assassination of two southern provincial governors. According to a spokesman for the Baghdad headquarters of the Multi-National Force-Iraq (MNF-I), those attacks are not included in the military's statistics. "Given a lack of capability to accurately track Shiite-on-Shiite and Sunni-on-Sunni violence, except in certain instances," the spokesman said, "we do not track this data to any significant degree."

Attacks by U.S.-allied Sunni tribesmen -- recruited to battle Iraqis allied with al-Qaeda -- are also excluded from the U.S. military's calculation of violence levels.

The administration has not given up trying to demonstrate that Iraq is moving toward political reconciliation. Testifying with Petraeus next week, U.S. Ambassador to Iraq Ryan C. Crocker is expected to report that top Shiite, Sunni and Kurdish leaders agreed last month to work together on key legislation demanded by Congress. If all goes as U.S. officials hope, Crocker will also be able to point to a visit today to the Sunni stronghold of Anbar province by ministers in the Shiite-dominated government -- perhaps including Prime Minister Nouri al-Maliki, according to a senior U.S. official involved in Iraq policy. The ministers plan to hand Anbar's governor $70 million in new development funds, the official said.

But most of the administration's case will rest on security data, according to military, intelligence and diplomatic officials who would not speak on the record before the Petraeus-Crocker testimony. Several Republican and Democratic lawmakers who were offered military statistics during Baghdad visits in August said they had been convinced that Bush's new strategy, and the 162,000 troops carrying it out, has produced enough results to merit more time.

Challenges to how military and intelligence statistics are tallied and used have been a staple of the Iraq war. In its December 2006 report, the bipartisan Iraq Study Group identified "significant underreporting of violence," noting that "a murder of an Iraqi is not necessarily counted as an attack. If we cannot determine the sources of a sectarian attack, that assault does not make it into the data base." The report concluded that "good policy is difficult to make when information is systematically collected in a way that minimizes its discrepancy with policy goals."

Recent estimates by the media, outside groups and some government agencies have called the military's findings into question. The Associated Press last week counted 1,809 civilian deaths in August, making it the highest monthly total this year, with 27,564 civilians killed overall since the AP began collecting data in April 2005.

The GAO report found that "average number of daily attacks against civilians have remained unchanged from February to July 2007," a conclusion that the military said was skewed because it did not include dramatic, up-to-date information from August.

Juan R.I. Cole, a Middle East specialist at the University of Michigan who is critical of U.S. policy, said that most independent counts "do not agree with Pentagon estimates about drops in civilian deaths."

In a letter last week to the leadership of both parties, a group of influential academics and former Clinton administration officials called on Congress to examine "the exact nature and methodology that is being used to track the security situation in Iraq and specifically the assertions that sectarian violence is down."

The controversy centers as much on what is counted -- attacks on civilians vs. attacks on U.S. and Iraqi troops, numbers of attacks vs. numbers of casualties, sectarian vs. intra-sect battles, daily numbers vs. monthly averages -- as on the numbers themselves.

The military stopped releasing statistics on civilian deaths in late 2005, saying the news media were taking them out of context. In an e-mailed response to questions last weekend, an MNF-I spokesman said that while trends were favorable, "exact monthly figures cannot be provided" for attacks against civilians or other categories of violence in 2006 or 2007, either in Baghdad or for the country overall. "MNF-I makes every attempt to ensure it captures the most comprehensive, accurate, and valid data on civilian and sectarian deaths," the spokesman wrote. "However, there is not one central place for data or information. . . . This means there can be variations when different organizations examine this information."

In a follow-up message yesterday, the spokesman said that the non-release policy had been changed this week but that the numbers were still being put "in the right context."

Attacks labeled "sectarian" are among the few statistics the military has consistently published in recent years, although the totals are regularly recalculated. The number of monthly "sectarian murders and incidents" in the last six months of 2006, listed in the Pentagon's quarterly Iraq report published in June, was substantially higher each month than in the Pentagon's March report. MNF-I said that "reports from un-reported/not-yet-reported past incidences as well as clarification/corrections on reports already received" are "likely to contribute to changes."

When Petraeus told an Australian newspaper last week that sectarian attacks had decreased 75 percent "since last year," the statistic was quickly e-mailed to U.S. journalists in a White House fact sheet. Asked for detail, MNF-I said that "last year" referred to December 2006, when attacks spiked to more than 1,600.

By March, however -- before U.S. troop strength was increased under Bush's strategy -- the number had dropped to 600, only slightly less than in the same month last year. That is about where it has remained in 2007, with what MNF-I said was a slight increase in April and May "but trending back down in June-July."

Petraeus's spokesman, Col. Steven A. Boylan, said he was certain that Petraeus had made a comparison with December in the interview with the Australian paper, which did not publish a direct Petraeus quote. No qualifier appeared in the White House fact sheet.

When a member of the National Intelligence Council visited Baghdad this summer to review a draft of the intelligence estimate on Iraq, Petraeus argued that its negative judgments did not reflect recent improvements. At least one new sentence was added to the final version, noting that "overall attack levels across Iraq have fallen during seven of the last nine weeks."

A senior military intelligence official in Baghdad deemed it "odd" that "marginal" security improvements were reflected in an estimate assessing the previous seven months and projecting the next six to 12 months. He attributed the change to a desire to provide Petraeus with ammunition for his congressional testimony.

The intelligence official in Washington, however, described the Baghdad consultation as standard in the NIE drafting process and said that the "new information" did not change the estimate's conclusions. The overall assessment was that the security situation in Iraq since January "was still getting worse," he said, "but not as fast."

Staff writer Ann Scott Tyson contributed to this report.