Geoff Dougherty

---

Math problems: Many numbers used to back a campaign for more deputies were wrong. The errors were inadvertent, officials say.

By Geoff Dougherty | St. Petersburg Times Staff Writer

“EMERGENCY,” said the deputy depicted on the cover of the brochure. “Send Backup I NEED HELP”

Below was a picture of a dispatcher and her response: “I’m sorry ... all units are busy!”

Inside was more: “Our deputies need our help! When their backup is 15-18 minutes away, our safety, as well as the safety of our deputies, is at stake.”

Pasco Sheriff Lee Cannon hoped the brochure would generate support for his 1998 proposal for a law-enforcement tax. Over 10 years, the plan would have added 220 deputies and cost property owners up to $ 80-million. Some voters undoubtedly backed the tax because Cannon’s statistics painted a frightening picture of a department critically understaffed.

They might have voted differently if they had known the facts.

And facts, a Times investigation shows, were in short supply in Cannon’s campaign for more deputies. Many of the key numbers on which Cannon based his case are wrong:

Wrong about the time to get backup to a deputy in trouble. (The Sheriff’s Office keeps no statistics on that issue).

Wrong about response times to life-threatening calls. (The data is so poor that an accurate calculation can’t be made.)

Wrong about the number of calls for service his deputies must handle. (It’s less than half of what Cannon said.)

Wrong about the percentage of time his deputies spend answering calls. (It’s a third of what Cannon said.)

Wrong about Pasco’s crime rate. (Cannon used outdated figures.)

Each time Cannon used an incorrect number, it lent more support to his political position than an accurate number would have.

And his erroneous numbers influenced more than just ordinary voters. Many of them were repeated to county commissioners – who set the sheriff’s budget – in writing and at official meetings.

Although the tax plan was defeated at the polls, Cannon’s top aides say the campaign elevated the profile of the staffing issue. In the two budgets passed since the tax vote, county commissioners (who Cannon had blamed for his staffing problems) gave the sheriff 48 additional deputies – the biggest increase in 12 years.

“We were shot down in September 1998 (on the tax vote) and in October 1998 we get these new positions,” said Capt. Thomas Brooks. “I think it opened up a lot of peoples’ eyes.”

County Commissioner Steve Simon, while noting other factors played into the new spirit of cooperation, said the tax campaign made commissioners more willing to give Cannon the deputies he sought.

“In part, you could attribute that to the campaign,” Simon said.

Law enforcement experts say the problems with the Sheriff’s Office statistics raise questions about leadership and credibility.

Fred Shenkman, a University of Florida criminologist who advises police departments, said the inaccuracies may lead some to wonder whether the Sheriff’s Office was honest in producing its statistics.

“Either they are lying to you or they are so unsophisticated they don’t realize their mistakes,” said Shenkman, who has no firsthand knowledge of Pasco’s situation. “Which is worse? It’s sort of scary.”

Sheriff’s officials said they fell victim to an inadequate computer system that gave them bad statistics. And they denied the inaccuracies were intentional.

“There was no intention on our part to artificially inflate any of the statistics we used,” said Harold Sample, Cannon’s top aide. ” ... In looking at what we faced on the street and trying to put analytical comparison or quantification to it, we examined data that was available and felt that it helped to demonstrate these existing deficiencies and we used them.”

It’s not unusual for law enforcement agencies to have trouble gleaning useful data from their computer systems, according to George Sullivan, a police management consultant hired by the Times. In fact, Sullivan founded his business to deal with precisely that type of problem.

“Having poor systems or ineffectual practices are not management shortcomings, per se,” Sullivan wrote. ” ... The state of police management and utilization of information is far behind the private sector.”

Priority 1 response times

Cannon based the tax campaign on the department’s slow response times – an average of 15 minutes, 52 seconds in potentially life-threatening cases. Cannon also repeated that claim in a speech at a Board of County Commissioners meeting in May 1998.

But the Times analysis turned up a statistical anomaly: Two-thirds of the department’s priority one calls are answered in less than the 16-minute average. The average is raised by a small group of calls with extraordinarily long response times. (See chart.)

Sheriff’s Office officials explained that most of the delayed calls did not turn out to be truly urgent incidents.

Although some 911 calls initially are listed as urgent and keep that classification until a deputy arrives on scene, that’s not true for others.

In those instances, dispatchers or deputies obtain additional information about the incident and assign it a lower priority. Dispatchers are supposed to log those changes but sometimes don’t get around to it.

That means the system may show it took an hour to respond to a priority one call that became a far less serious call – and one that merited a slower response.

Although acknowledging the record-keeping problems, Cannon stood by his numbers in the days before the tax vote.

After the vote, the department hired its own consultant and he raised many of the same questions that the Times did in its original story. Only then did sheriff’s officials say that data problems invalidated their response-time statistic.

“I cannot tell you what our response time is” to an urgent call, community relations director Bill Kontyko said in an interview Monday.

Despite that, the department insists its response times are too long and the problem is short staffing.

In the absence of data, how does the department know that?

“Based on actual experience,” Sample said.

To justify their statements about understaffing, sheriff’s officials point to the department’s unusually low ratio of deputies to residents. That number is one of the few that is accurate and supports their position.

But many police experts describe the deputies-to- residents ratio as a meaningless number.

“If they are talking about deputies per thousand citizens, the whole issue is moot,” said Sheldon Greenberg, a former police officer who teaches law enforcement leadership at Johns Hopkins University. “The number ... is absolutely useless.”

Did Sample know about opinions like Greenberg’s before the brochure was published?

“Yes. But the numbers were so far off, as far as any comparison with other counties, that the need was apparent,” Sample said. “We are not talking about (the difference between) 1.8 to 1.9. We are talking about less than 1.0 (in Pasco) compared to a state average of 2. We are talking about a huge disparity.”

Emergency backup

The Times also found problems with the department’s claim that understaffing places its deputies in danger. The brochure said it took, on average, 15 minutes, 52 seconds to respond to deputies in trouble. In reviewing the department’s data, the Times found no indication that such response times were tracked.

In an interview, Kontyko conceded that point.

“We can’t tell you what our average response time for that type of call would be,” he said.

He said he assumed that the response time to backup calls would be the same as for other high-priority calls.

Brooks said that assumption is not correct. A deputy calling for help is one of the highest-priority calls the department handles, Brooks said, and is unlikely to suffer from the type of dispatching delays that affect other kinds of calls.

“Other units hear the call on the radio and they respond without being dispatched,” Brooks said.

But using information from the sheriff’s dispatch system, the Times calculated a similar measure – the average time a deputy spends alone on the scene of a high-priority call before backup arrives. That average, just under four minutes, provides an adequate margin of safety, experts say.

“Four minutes is very, very good,” said John Doherty, a retired police captain and criminologist at Marist College in New York. “They are probably very proud of that, and they should be.”

How busy are they?

Police experts say an adequately staffed sheriff’s department spends only about a third of its time handling residents’ calls for service. Any more than that leaves too little time for activities like community-oriented policing and report writing.

The brochure claimed Pasco deputies spend 78 percent of each shift answering calls for service. Cannon made the same statement at a May 1998 County Commission meeting.

But a Times computer analysis of 1998 dispatch logs shows that Cannon’s road deputies spend 26 percent of each shift handling calls from residents. Further, the newspaper found the Sheriff’s Office figure included time spent not on law-enforcement calls, but on jail transports and off-duty jobs like guarding the Port Richey Hooters restaurant. The Sheriff’s Office figure was inflated by an arithmetic error, which the department acknowledged in an interview.

When concluding that deputies spend 78 percent of their shift on calls for service, the department added up the time spent by every member of the department on all the incidents for one year. That figure included work by road deputies, detectives, prison transport deputies, school resource officers and others. The department then divided that number by the total time worked by only road deputies.

The mistake is akin to calculating Wade Boggs’ batting average by adding up all the hits for the entire Devil Rays team and dividing by the number of times Boggs went to bat. It yields a high – and inaccurate – number.

In an interview shortly after the tax vote, Kontyko acknowledged the error.

“That’s what I did,” he said. He added that his approach surely would increase the percentage of time spent on calls, but he couldn’t estimate by how much.

Calls for service

In the brochure and in budget documents submitted to the Board of County Commissioners, the department said it responds to nearly 300,000 calls a year for service.

The Times analysis turned up a much lower figure: 101,000.

The Sheriff’s Office, while admitting many of the statistics used to support the tax district were wrong, insisted the department responded to about 300,000 calls for service.

Lt. Robert W. Stone, who oversees the dispatch center, said Monday he was certain of that number because he and an assistant had hand-counted the number of calls the department answered each day. They then added those figures to get a yearly total.

That insistence came after Sample and others said the department had learned its lesson about statistics and would subject its numbers, in the future, to a rigorous review.

“At this point, will we be doing a more in-depth analysis of those numbers? You bet your sweet bippie,” Sample said on Monday.

On Tuesday, Kontyko called the Times to admit the 300,000 number was wrong.

“We do not respond to 300,000 calls for service,” he said.

He said that the high number possibly was a count of every time a police car arrived on the scene of an incident, rather than a count of the incidents themselves.

The crime rate

The brochure says violent crime increased 29 percent between 1988 and 1996. But when the brochure was printed, the Florida Department of Law Enforcement already had released its 1997 crime figures. If Kontyko had used the most current figures, the brochure would have said violent crime was down 2.7 percent from 1988.

In making the case that the Pasco Sheriff’s Office gets less funding per resident than other counties, the brochure said the Hernando Sheriff’s Office had a law-enforcment budget in 1996-1997 of $ 13.6-million.

But Hernando’s chief deputy, Don Shields, said that number is at least $ 500,000 too high.

“I don’t know where they got that number,” he said. “I don’t have a clue where they got their figures.”

How it happened

In an interview Monday, Sample and Kontyko took responsibility for the department’s inaccurate statistics.

Sample said the Sheriff’s Office had never tried to calculate response times or other figures from its $ 1.14-million computerized dispatch system before 1998.

At that time, Sample said, he and others were certain the department suffered from critically low staff levels. Although the department didn’t know what its average response time was, officials knew it was long and that it was a symptom of having too few deputies.

They turned to the computer system in an effort to justify Cannon’s tax. The numbers they found appeared to support their beliefs about staffing.

“Everything was done in good faith,” Sample said. “We searched for statistics to bolster what we knew was happening.”

There is no law against unintentionally providing false information to voters or public officials.

However, it is a crime to try to mislead a public official – such as a county commissioner – by intentionally submitting false written statements.

Sample and Kontyko blame the computer system for producing reports that weren’t accurate. Kontyko said he is dissatisfied with the system and the way it was programmed.

Many of the inaccuracies in the sheriff’s brochure did not, in fact, spring from the computer system. Only one of those numbers, the response time, was directly attributable to computer errors.

The response time to backup calls was the result of an incorrect assumption on Kontyko’s part. An arithmetic error led the department to portray itself as busier than it truly was. The inflated number of calls for service resulted from a misleading description about what the department had counted to arrive at a total of 300,000.

Law enforcement experts said the Sheriff’s Office’s approach to statistics is a common one among police agencies.

Police rarely have much occasion to work with statistics, said Greenberg, the Johns Hopkins professor.

Greenberg recently surveyed 70 police executives and found that only 10 had any formal training in how to set staffing levels.

“Resource allocation in many police agencies consists in many cases of the chief or sheriff making a plea to the council and the council saying: “Here’s how much money we have,’ ” Greenberg said. “It’s not based on anything more than that.”

That approach, although common at police agencies, wouldn’t fly elsewhere.

“You would be laughed out of most executive boardrooms if you looked for additional resources with as little information as the police compile,” Greenberg said.

January 23, 2000