Of all the feats Mayor Martin O'Malley promotes as he runs for governor, none sounds more impressive than his claim of crime-fighting prowess.
It goes like this: Violent crime in Baltimore has declined nearly 40 percent since he took office in December 1999, a rate outpacing every other big city in the nation.
The claim is central to O'Malley's gubernatorial strategy of convincing Maryland voters that, despite the city's stubbornly high homicide rate, his leadership on crime reduction has ushered in a Baltimore renaissance by stopping population loss, spurring investment and attracting employers.
But several criminologists say O'Malley's "nearly 40 percent" claim, prominently featured in his campaign literature, is an inflated assessment based on an inaccurate calculation of statistics.
The main source of skepticism from criminologists is an O'Malley-commissioned audit in 2000 that found violent crime had been under-reported in 1999, leading to an upward revision of city's violent crime rate for 1999, the year the mayor was elected. That is the benchmark against which O'Malley measures progress.
Yesterday, The Sun learned that Gov. Robert L. Ehrlich Jr.'s Office of Crime Control and Prevention is auditing a year's worth of statistics for the city and four other large Maryland jurisdictions to determine whether they have underreported crime more recently.
O'Malley says the city's progress on violent crime -- homicide, rape, robbery and aggravated assault -- cannot be denied and that the Police Department's crime statistics are reliable despite recent gaffes in reporting.
"We police our own numbers pretty darn well," the mayor says.
Nevertheless, O'Malley says he expects attacks. "If the state police wanted to do [an audit] they could. ... I hope your political antenna goes up if Ehrlich wants to do it this year."
The crime experts, including an FBI agent working as consultant for the city, say a more accurate estimate would put the reduction in violent crime in Baltimore from 1999 to 2004 at less than half of O'Malley's claim, which would make it significant, but not nation-leading.
Before the audit O'Malley ordered when he took office, the violent crime rate in Baltimore in 1999 was that decade's lowest. After the audit, 1999 became one of the city's most violent years in two decades, reversing a four-year trend of declining violence, records show.
The extensive audit found that police had been mistakenly classifying violent crimes, especially aggravated assaults, as lesser and technically nonviolent offenses. Some criminologists question some of the reclassification techniques used in that audit six years ago, such as calling crime victims to get their recollection of events.
"He may be right in correcting what happened in 1999," says David B. Muhlhausen, a senior policy analyst specializing in criminal justice issues at the Heritage Foundation. "But I would be very suspicious that the year he comes into office he corrects the numbers and then says he drastically reduces crime."
Several criminologists say O'Malley should not declare such a steep reduction in violent crime unless current crime reports undergo the same comprehensive scrutiny applied to the 1999 data.
"A better comparison would be the unadjusted numbers for 1999," says James Nolan, a West Virginia University criminologist and former FBI expert who was hired by the city to help devise the 1999 audit.
By that measure, O'Malley's "nearly 40 percent" decline -- actually 37.4 percent from 1999 to 2004 -- would drop to a 23.5 percent reduction in violent crime, the sixth-largest decrease among the nation's largest cities in that period.
"The only real accurate way to [determine the rate of reduction] is to do another audit," Nolan says.
O'Malley says he encourages anyone to obtain police records and conduct a review. He says a comprehensive audit similar to the one in 1999 is unnecessary because the scrutiny applied six years ago helped establish a more accurate and thorough process for reporting crime.
The audit established more training on crime classification, implemented checks on reports at several levels within the Police Department and started an inspections unit for spot-check audits, says Kristen Mahoney, chief of technical services for the department.
One such audit of 2003 crime reports, not nearly as exhaustive as the 1999 review, increased crimes reported to the FBI by 1.6 percent.
Mahoney says many critics mistakenly think officers and their immediate supervisors are solely responsible for classifying crimes. The process is far more complicated, she says, and includes a string of staff reviews starting with details in 911 calls.
Internal checks are conducted by FBI-trained employees at the department's central records division, which operates independently of the district commanders, who answer for their crime data each month.
Nolan and other criminologists agree that the implementation of the more rigorous measures is likely to avoid mistakes found by the 1999 audit.
"They're probably not making the same mistakes," Nolan says. But, he adds that the only way to determine that the errors have not resurfaced is with an audit.
Michael D. Maltz, a criminologist at the University of Illinois at Chicago and Ohio State University, says the city should audit its figures every year in the same manner it did in 1999.
"Comparing an audited account of crimes to an unaudited account of crimes is meaningless," Maltz says, recommending that any such audits be conducted by an agency "not connected to the hierarchy of the Police Department."
City elected officials including Del. Jill P. Carter and City Councilman Kenneth N. Harris Sr. say the mayor should counter criticism by inviting an independent audit, noting that the mayor is facing the same scrutiny that O'Malley, as a councilman, leveled at Mayor Kurt L. Schmoke.
Harris says O'Malley could "remove any type of suspicion," especially in an election year, by authorizing a new audit.
Some of the recent reporting mistakes that Harris, Carter and others point to include:
Police said last year that violent crime in 2004 was lower than in 2003. But when the FBI published the city's data, violent crime was shown to have increased 4 percent. The reversal surprised city officials, who primarily attributed the change to a computer problem caused by flooding at police headquarters.
In December 2003, a Police Department review found that the city had underreported rapes to the FBI. After the review, the city's rape total for 2002 was increased from 178 to 211. But databases with the FBI and the Maryland State Police still show 178 rapes. A state police official, Denise Scherer, who is responsible for collecting the data and reporting it to the FBI, says city officials alerted her office to the change but never delivered the formal paperwork.
O'Malley administration officials say all of the questionable incidents were corrected by the Police Department's systems, demonstrating that the city is capable of accurately policing its data.
The mayor acknowledges the increase reported in the 2004 crime data and attributes the problems in recordkeeping to former Police Commissioner Kevin P. Clark. O'Malley fired Clark -- his third police chief -- in November 2004.
"We were struggling that year," O'Malley says. In 2004, the city recorded 276 homicides, the most during any year of the mayor's tenure and far exceeding his goal of 175.
The mayor's critics have used the stubbornly high homicide rate to argue that O'Malley's crime-fighting efforts have had little effect.
Many criminologists say homicides can be a more reliable barometer of a city's violence because they are almost always reported to police and there is little discretion in classifying the offense. But they also say homicides do not present a complete portrait of a city's safety because they account for a small percentage of violent crime.
In the six years since O'Malley took office, homicides have dropped 11.8 percent, from 305 in 1999 to 269 in 2005. Just getting Baltimore's annual homicide count below 300 -- something that didn't happen during the 1990s -- was an accomplishment, O'Malley officials say. His campaign Web site says O'Malley's public safety reforms have resulted in "more than 250 people being alive today and 24,000 fewer people being victims of violence."
Beyond statistics, administration officials say, anecdotal evidence among many community leaders illustrates a safer city. Even the Rev. Gregory Perkins, an O'Malley critic, says drug dealers are less brazen.
Still, polls conducted by the Gallup Organization and Landor Associates have found that residents and visitors still widely perceive Baltimore as dangerous.
As a candidate in 1999 and as a city councilman, O'Malley routinely raised questions about the city's crime reporting. One of the reasons for the audit was to establish an accurate baseline.
Schmoke says there was never any reason to doubt the statistics the audit studied.
"I announced in 1998 that I wasn't running for re-election, so what incentive would there be to pad the number?" Schmoke says. "No [federal] government agency ever accused us of fudging our numbers on crime in the 12 years I was mayor."
The 1999 audit made no such accusation either. Instead, it mainly attributed mistakes to a lack of training and inadequate internal reviews. The audit was conducted by the Police Department and a private consultant, Maple and Linder Group, which was paid about $160,000.
The audit examined a large sample of crime reports from the first six months of 1999 by comparing what officers recorded as the facts of an incident to the classifications that they determined were most appropriate. The auditors also called victims to determine whether their recollections matched the reports.
The audit then took the rate of erroneously classified crimes from the January-to-June review and applied that rate to the entire year using a formula provided by Nolan, the FBI consultant.
The review boosted total crime in 1999 by 14.5 percent, from 66,015 to 75,587. The number of crimes classified as violent rose by 22 percent, from 15,245 to 18,630, after the audit.
The most frequent change involved assaults. Before the audit, the city reported 7,138 aggravated assaults. After the audit, the number was increased to 10,452, a 46 percent increase.
That was the most aggravated assaults reported in Baltimore since 1975. The city has never exceeded 10,000 aggravated assaults. In 1993, the most violent year since 1975, 8,577 were recorded.
Criminologists agree with city officials that aggravated assaults -- attacks intended to cause serious injury or committed with a deadly weapon -- are the easiest to downgrade to common assaults.
"Police departments have a history of fudging numbers, especially the aggravated assault numbers," says Muhlhausen of the Heritage Foundation.
No such accusations were made against the police during the Schmoke administration. O'Malley officials defend the audit's changes by saying that the FBI endorsed the study.
But the FBI does not keep records for Baltimore's 1999 crime statistics in any of its databases, say two FBI officials with the Uniform Crime Reporting Program office in West Virginia. And the FBI's crime report for 2000 provides a footnote for Baltimore's statistics that reads: "Due to changes in reporting practices, annexations, and/or incomplete data, figures are not comparable to previous years' data."
Nolan is cited in the audit as giving the study an FBI stamp of approval, but the West Virginia University professor says he did not entirely agree with the way the audit was conducted. He says he believes the audit was generally accurate but that the city should not have changed reports based on calling victims to review their recollections of crimes.
The majority of changes were made after reviews of narratives written by officers. That, Nolan says, is the best way to determine mistakes in classifications.
"I think the error rate would be much less in 1999 if you take out the callbacks," Nolan says.
Other criminologists question whether it was appropriate for the audit to restate crime for all of 1999 by applying an error rate based on mistakes found in a sample from the year's first six months.
"Certain types of crimes may be seasonal," Maltz says. "It would be better if they could do an audit for the entire year."
Nolan agrees but says practical concerns prevented a review of the full year's statistics. "From an academic point of view, it could be a concern," he says. "Realistically, it's probably not."
Political experts differ on whether the debate will get voters' attention.
But O'Malley's political opponents -- Republican Gov. Robert L. Ehrlich Jr. and Montgomery County Executive Douglas M. Duncan, a Democrat -- have used the issue to support claims that the mayor exaggerates his record in Baltimore.
"The statistics that he touts are suspicious," says Audra Miller, a spokeswoman with the Maryland Republican Party. "What the mayor is trying to sell the people of the state and the city is counter to the facts."
Duncan, who is vying with O'Malley in the primary to challenge Ehrlich, criticized O'Malley last year after the mayor said in a June 6 speech in Montgomery County that his administration had reduced violent crime in Baltimore by 50 percent since 1999.
"Martin O'Malley has a habit of exaggerating in order to divert attention away from a city still plagued by crime, a struggling school system and his failure to meet his own goals," a Duncan campaign flier says.
Sun reporter Andrew A. Green contributed to this article.