Florida's analysis of crime in 1996 brought some good news: the state's overall crime rate declined again, allowing Arizona to take over the top spot among the most crime-ridden states. And for residents of Palm Beach County, the 6 percent drop in crime was much higher than the overall rate.

One big problem: both numbers were wrong. In its report, the Florida Department of Law Enforcement included in its calculations police agencies that had failed to submit a report or gave inaccurate figures.

Twenty such departments were counted as having no crime at all by FDLE, which skewed the overall rate. The police department for Hialeah, a city of 200,000 people in Dade County, was the largest agency to report "no" crime.

Palm Beach Post reporters also discovered that two local agencies turned in incomplete or inaccurate reports to the state.

One city said its crime rate declined 60 percent from 1995 to 1996; the reason was that officers were not turning in crime reports to a records clerk.

How the story was done

In advance of the state's report, police reporter Monika Gonzalez asked every police department in the county for a copy of the report they sent to FDLE. Our original intention was to compile a chart and run a story the day the report was to be released.

But when we compiled the numbers in Excel and did a few calculations, we noticed that several departments had huge swings in crime from 1995 to 1996.

So we called those departments. One city sent its six-month report to the state because the final version wasn't ready. Another sent a full report but knew it was incorrect. Two others didn't send a report. FDLE included all of them in its calculations.

To account for these agencies, we excluded them from both 1995 and 1996 totals and recalculated the crime rate. The county's rate actually increased during that time period.

Next we asked FDLE for 1996 crime data from every police department in the state, which we received on a 3.5-inch floppy disk and imported into Excel.

We found 17 other departments that were counted as having no crime. Using our paper copies of 1995 data, we found seven more who didn't report any crime in that year.

So we again excluded all of those agencies from both years' totals and recalculated the change in the crime rate. Instead of a 1.7 percent drop, Florida's crime rate edged up 0.41 percent.

Lessons learned

Obviously, the lesson here is to check the data.

Florida has hundreds of police departments and the state's report presents summary data but not detailed information. FDLE made no mention of the fact that these 20 agencies reported no crime - we simply looked at the numbers in Excel.

Eyeballing (and multiple sorting) led us to most of the major discoveries for our stories, and a few minor ones that would raise questions. One county said its crime rate dropped nearly 84 percent during the two-year period. We also found that agencies in other counties seemed to have a habitual problem with reporting, filing no reports in both years.

FDLE acknowledged that its report is only as good as the numbers it receives but said there were no plans to change the system.

But reporters, editors and, hopefully, readers now realize that the crime reporting system depends not just on the public, but also on police agencies who compile statistics.

And that sometimes, these agencies make mistakes.