Every year, Pennsylvania officials make weighty pronouncements about statewide crime trends, backed by stacks of statistics to document the ups and downs.
But inherent flaws in data make those statistics unreliable and possibly misleading because thousands of crimes every year go unreported by police departments.
The problems had the largest impact in 2001, when a switch to a new Internet-only collection system led to a serious decline in reporting which is voluntary. Even though reporting dropped by more than 50 departments including three cities the state publicized a 6 percent reduction in serious crimes from 2000 to 2001 without accounting for crimes missed in those areas.
The drop in crime probably was closer to 2 percent, according to an analysis by The Morning Call that compared only departments that reported numbers in both years.
Data problems make it nearly impossible for anyone to home in on trends occurring in geographic regions smaller than the state as a whole. And larger departments that do not report disqualify themselves from state grants to be used for fighting crime.
''The voluntary numbers really are meaningless if you want to look at the true nature of crime across the state,'' said Judy Yupcavage, public policy and information manager for the Pennsylvania Coalition Against Domestic Violence.
Among the Uniform Crime Report program's most serious problems:
A third of the state's 1,100 municipal police
departments routinely fail to participate in the data-reporting process, meaning the final report misses thousands of crimes. Another 15 percent or so of police departments report crimes some months but not others. In total, 2.7 million Pennsylvanians live in areas where crimes go unreported or under-reported to the state.
Officials who collect the data make little attempt to track the annual fluctuations, which leaves them unprepared to explain peaks and valleys in the data.
Little effort is made to alert the public of the report's statistical shortcomings, other than a disclaimer that reporting is voluntary and not all departments submit numbers.
These problems undermine the program's stated mission: to provide government officials and the public with information about crime in Pennsylvania and to provide law enforcement administrators with criminal statistics.
The combination of missing data and failure to disclose problems in the Uniform Crime Report calls into question its accuracy and can make the statistics misleading, according to experts.
The Pennsylvania State Police, who oversee the UCR process, defend it, saying as long as they receive data from the major metropolitan areas, they are able to track broad crime trends.
''As long as we're getting Philadelphia, Pittsburgh and the state police which are accounting for a large percentage of the crimes you're able to look at trends,'' said Jack Lewis, state police spokesman.
But the system's flaws even can derail such broad comparisons, as happened in 2001. The data excluded cities as large as Scranton, Wilkes-Barre and York.
Credible scientific studies include information about data collection problems and how those problems affect the results. But state police said such routine reporting analysis is beyond the mission of the state UCR program, which state police have overseen since 1973.
''We stand by the decisions that were made when we implemented this program that we disseminate the information as we know it, whatever data we have,'' said Sgt. Mike Troxell of the research and development unit.
The underlying problem: Pennsylvania doesn't require police to file UCR numbers and many departments say they don't have the time, resources or incentive to do so.
''I'm a one-man department,'' said Chief Jeffrey Tapler of Lower Milford Township. ''We haven't done it here, ever.''
Though he said he would like to start reporting, Tapler said he has no secretarial help and doesn't have the time to get up to speed on the UCR.
Filing UCR data isn't just a matter of transcribing local crime figures. Differing criteria for specific types of crimes can cause confusion. State law defines certain crimes differently than the FBI standards upon which the UCR is based.
Still, thousands of police departments across the country report routinely. Pennsylvania is one of only a handful of states feeding crime statistics to the FBI's UCR that doesn't require police to report statistics.
In New Jersey, monthly reports have been mandatory since the program was created by state law in 1967. Every one of the state's 484 departments files, said Sgt. Alyson Honrath of the state police Records and Identifications Bureau.
In Pennsylvania, most departments that do not report are smaller departments, those with fewer than 12 officers. Many say they lack the computer capabilities necessary to gather and file the data.
In the past, that didn't matter as much, because many departments submitted paper records to state police every month. Larger departments generally had software that compiled the numbers, sent through the mail on diskette.
But in January 2001 the state switched to an Internet-only filing system and stopped accepting reports on paper or diskette. Despite having been warned of the change about six months in advance, many departments failed to adjust.
That year 111 more departments failed to report a 33 percent jump over the prior year. Many were mid-size departments that were unready for the switch in computer systems.
State police say they anticipated the temporary slippage but are confident reporting will increase as departments adjust to the new system. They believe reporting rates will be back to normal if not better within a year.
While he plans to look at the number of departments reporting over the next few years as the transition to the Internet-only system continues, state police spokesman Lewis said there are no plans to incorporate those counts into the annual report or look any deeper at reporting trends.
New Jersey aggressively checks its numbers for accuracy. Twice a year, it looks at six-month trends across the state as compared to the same period a year earlier, double-checking for inconsistencies.
''Just because goes down, we don't assume all is good and well,'' Honrath said. ''We still go out and make sure that things aren't being missed.''
Pennsylvania's unwillingness to demand all departments file and then check for errors weakens the accuracy of the data and increases the likelihood people will draw false conclusions about crime, experts say.
Under-reporting can cause serious problems, especially when comparing trends in smaller areas, said Iain Murray, an authority on statistics in the Washington, D.C., area who has written extensively about crime reporting.
''If you are looking for local figures, to be able to hold a locally elected sheriff in account, this is a problem,'' Murray said.
Even larger trends can be skewed, as the 2001 report shows.
''Our efforts to fight crime continue to pay off for Pennsylvania residents,'' state police Commissioner Paul J. Evanko announced in a news release last year, announcing the erroneous 6 percent drop. ''The numbers show that the commonwealth is the safest it has been in years.''
Lewis took the blame, saying he didn't think of checking whether reporting had dropped before writing news releases touting the apparent drop in crime as the real thing.
Comprehensive reporting would allow crime groups such as the Pennsylvania Coalition Against Domestic Violence to zero in on trouble spots, target their public service messages and better judge whether crime prevention programs are working in a particular area, Yupcavage said.
But because the UCR is so unreliable, Yupcavage said, she relies instead on anecdotal evidence, such as a newspaper-clipping service that looks for domestic violence-related murders.
''You would have to go police department to police department to get the actual numbers in any given year,'' she said.
There are no initiatives in Pennsylvania to mandate reporting. Experts say there are two ways to boost reporting dictate or entice. States that do the best job gathering data usually tie reporting to grants that target smaller departments as well as larger ones.
That's the method preferred by many state law enforcement personnel, who generally oppose mandates that could cut into the already-thin police budgets.
''Instead of making it mandatory, make it an incentive'' by tying it more closely to funding, said Amy K. Corl, executive director of the Pennsylvania Chiefs of Police Association in Harrisburg.
A few state grants depend on reporting to the UCR, but not enough to make a major difference in reporting especially to smaller departments, Corl said.
For example, federal Juvenile Accountability Incentive Block Grant money is distributed to states based on the statewide UCR violent crime reporting to the FBI. The state then distributes about $7 million a year to local governments, also based on their reported violent crimes. Only counties and larger municipalities are eligible.