INQUIRER STAFF WRITERS

A review of Philadelphia's crime statistics by the City Controller's Office has

concluded that the police failed to report between 13,000 and 37,000 major crimes in 1998.

The review, released yesterday, relied on an FBI inspection of 1,000 randomly selected police reports.

In addition, investigators with the Controller's Office interviewed more than 300 citizens who filed crime

reports with police to compare their recollections with the actual reports.

The controller's 51-page audit, nearly two years in the making, drew an angry response from Police

Commissioner John F. Timoney, who took office in March 1998.

Timoney, in a statement attached to the report, blasted the review as overblown, statistically flawed,

and "of no help to the department."

He said the analysis ignored his reforms in police record-keeping and was based on a year, 1998,

before they had a chance to take effect. Timoney has made reforming police statistic a major priority,

creating the force's first internal auditing year and disciplining captains thought to have tampered with

reports.

Timoney reacted coolly to the two major changes urged by Controller Jonathan A. Saidel, saying only

that he would consider them. The controller urged the department to:

Routinely call back victims to test the honesty of police reports.

Create a central office to review crime reports.

Those reports are now overseen by district captains, whose performance is often measured against the

level of crime in their areas.

Saidel launched his audit after a series of articles in The Inquirer detailed how police had failed to count

thousands of crimes ranging from rapes to thefts in previous years.

Yesterday's audit indicated that the overall problem of crime-report "downgrading" was far larger than

ever before reported.

The police reported 106,000 major crimes in 1998. Timoney at one point said that year's figure would

be a "benchmark" for accurate statistics. He later said that last year's figures were a more reliable count

as his reforms took hold.

With cooperation from Timoney, the controller subjected the Police Department's figures to the first

such sweeping outside review in the nation.

The auditors reviewed the 1.4 million calls made by citizens to 911 for which police filed a report.

About 200,000 involved crimes and the rest were other requests for police service.

The auditors randomly selected 1,053 of the 1.4 million reports and sent their case-files to the FBI for

review. The agency agreed to serve as referee for the audit.

The FBI rejected 34 of them as improper downgrades - about 3 percent.

The controller then applied that 3 percent "error rate" back to the original group of 1.4 million 911

reports, concluding that as many as 42,000 crimes were "downgraded" in the group.

Further statistical refinement, including a margin-of-error computation, led to the conclusion that the

range of uncounted major crimes was between 13,000 and 37,000.

The controller's sample and methodology were designed with the assistance of Alan J. Izenman, senior

research professor in the Department of Statistics at Temple University.

In a duel of academics, Timoney said that he had contacted the chairman of the University of

Pennsylvania's mathematics department, Dennis DeTurck, for a separate appraisal.

DeTurck found the controller's approach flawed, Timoney said. He estimated that the department

undercount was about 8,000 crimes.

In a separate interview, Clifford Zukin, director of Eagleton's Center for Public Interest Polling at

Rutgers University, said the controller's sampling of 1,054 police reports appeared to be large enough

to provide accurate results.

"If the sample was selected properly, it's a valid inference," said Zukin, a political science professor. "It

sounds fine to me."

According to the controller's audit, some felony assaults in 1998 were classified as minor beatings, and

in at least one case, "as a contempt of court offense instead of an aggravated assault," the controller

said.

In one example, the report said a victim, who was blind, reported to police that she had heard someone

enter her home and then went downstairs to find her house ransacked and her ATM card missing.

Philadelphia police listed the case as one of "lost and found property." The FBI said it was a burglary.

Sometimes, the report said, robberies, aggravated assaults and thefts were simply left out of the city's

crime count.

The controller researchers also attempted to directly contact the citizens who had called 911 to report

the 1,053 crimes used as the random sample.

It was able to locate 384 of them. Of them, 29 - or 7 percent - told the auditors that police officers had

dropped key facts from reports in order to reduce the severity of an offense.

Performing another complex calculation, the controller extrapolated that as many as 39,000 additional

crimes in 1998 might have been dumped in this fashion. The auditors noted, however, that they had not

interviewed the patrol officers for their side of the story. For that reason, the controller stopped short of

a flat conclusion that those crimes, too, had been downgraded.

Crime statistics have assumed increased importance in recent years as a yardstick for evaluating police

captains and plotting crime-fighting strategies.

Under Timoney, the department has made accurate figures key to the tracking of crimes and

deployment of police throughout the city.

But the controller's report said that department has been using flawed measurements, in part because

the police district captains themselves oversee the tallying of their own statistics.

"District commanders who can influence the numbers should not oversee personnel who code and

score crime reports," the report said. "A conflict of interest exists that exposes police management to a

greater degree of risk for 'downgrading' crimes."

The controller's office said that police in Dallas, Houston, Baltimore, Seattle and San Jose, among

others, centralize all record-keeping to avoid such conflicts of interest.

In his written response to the audit's recommendations, Timoney said the department had already made

a series of changes, including installing better computers that were less prone to errors.

"In reality," Timoney wrote, "the Controller's Office chose to audit a reporting system that no longer

exists."