Odessa police have reported inaccurate crime statistics to the FBI for at least a decade.
An Odessa American investigation into the Odessa Police Department’s Uniform Crime Reporting (UCR) numbers discovered police had misreported their own crime data in terms of incidents, clearances and arrests. Although generally making their results appear far worse than they should have been, police were unaware of the problem before the OA’s inquiry.
The investigation also uncovered the unsolved 2008 homicide of an infant, never publically reported, not only through the UCR but also to the media.
More than a month after discovering the errors, the software company, city technicians and Odessa police are still working to sort everything out in the massive, convoluted world of thousands of pieces of data and federal guidelines.
“For us, the headache is just starting,” Police Chief Tim Burton said last week, more than a month after starting to delve into why the numbers were off.
UNIFORM CRIME REPORTING
The Odessa Police Department is one of more than 17,000 agencies contributing to the UCR program, which began in 1929 as a way to provide a national view of law enforcement statistics despite varying local laws.
Sometimes UCR guidelines go against common sense.
For example, by the UCR definition, a male can’t be a victim of rape and crime clearances are counted in the year of the clearance, not year of the crime. In 1984, Odessa police cleared 16 homicides in a year where 13 occurred. However, all agencies must deal with this.
“The numbers are imperfect, but they’re imperfect in an equitable way,” Burton said.
He — like the FBI itself — warned against comparing agencies because of the numerous variances possible.
“Your first instinct is to compare yourself to others to see how you’re performing,” Midland Police Chief Price Robinson said. “But communities are different.”
For example, despite similar populations, the average Midland County household has about $10,000 more in income ($59,925) than the average Ector County household ($49,345), according to Demographics USA County Edition 2009.
But looking at the 2009 report released in early September, some agencies had more reason to be pleased than others.
Midland police outperformed departments of their size, dealing with fewer crimes but clearing more cases in 2009, the same year Midland’s Crime Stoppers received an award for helping Midland police clear 324 cases.
The Ector County Sheriff’s Office dealt with more crime reports in 2009 than 2008, but cleared more in 2009. The Midland County Sheriff’s Office cleared 10 percent of its crimes, lowest in the area, but experienced the lowest rate of violent crime reports.
Odessa police could point to very few positives in their reports except when compared against their own numbers from previous years.
They had 8.9 percent more crime reports per 100,000 people than the state average and 60.5 percent more violent crime reports. They reported 28 percent of all the murders, rapes, robberies and aggravated assaults they handled cleared in 2009 — the rest of the state reported clearing 44 percent. They reported 176 fewer arrests than Midland police even with more than 500 more reported crimes. Odessa police reported one arrest out of 36 rapes. It was their only reported rape arrest in 10 years.
Odessa police especially lagged behind their peers in clearance — either arresting someone or knowing who to arrest and being stopped by something out of law enforcement hands such as death of the suspect.
In September, Burton explained the draw of the oilfield during the boom times meant more officers left for economic opportunity and new recruits weren’t as experienced.
Burton said the results of that “double whammy” weren’t a surprise.
“There was such a large turnover, and we were so drastically understaffed for such a long period of time,” Burton said. “The clearance rates are a reflection of the amount of law enforcement resources to devote to crimes.”
For UCR data, Odessa police reported having 154 commissioned officers, or about one officer for every 650 people. Midland police reported having 170, a ratio of 1-600. Ector County Sheriff’s Office had 70 nonjail commissioned officers, or one for about every 500 people outside city limits, although only six deputies are on duty at a time.
The results were not desirable but were correct, Burton said. They made sense for modern Odessa, with its transient population and place in the international drug trade.
“Are the numbers being reported by our records department accurate?” Burton said in the initial interview in September. “Yeah, I believe they are accurate,” he said. He also expressed confidence in records manager Connie Sims, who he said was responsible for their accuracy.
The following day, however, police contacted the Odessa American to say they had found a problem with the numbers and requested a planned story about area UCR numbers be held until they could produce a corrected report.
On Oct. 14, the problem still hadn’t been fixed, but Burton said police had found something was wrong with how software was handling the numbers put into it. Both Burton and Sims stressed that the numbers being input were correct, and the crime incidents and rates were probably correct, but clearance rates were off.
“I have every confidence it’s all been done correctly,” Sims said. “We had every confidence in this software … Obviously it’s not picking up these cleared cases.”
Representatives of the city’s information services division said the city had been using a case-management system program by Tiburon Inc. for the past 14 years and had been using v. 7.42 for the past five years. City spokeswoman Andrea Goodson said annual maintenance for the software, which encompasses all of the city’s law-enforcement records, runs about $100,000.
Goodson said when information technicians began to look into it, they identified what they thought was the problem, but the output remained incorrect, so they contacted Tiburon.
“At this time, the analysis has indicated the records management system is operating properly, and there was no risk to the public or Odessa Police Officers at any time,” a Tiburon representative said in an e-mail.
Odessa police Cpl. Sherrie Carruth later said the same.
“The Records Management System is working fine. After initial diagnostics, a technical support request was submitted to Tiburon. A Tiburon engineering team is working to resolve the discrepancies in the UCR module,” Carruth said. “The problem is in the CANNED report, which is a standardized report that comes with the module.”
“Our data audits verify that the records management application continues to operate properly,” a Tiburon representative said in an e-mail. “We continue to be pleased with the progress we have made and defer to the Odessa PD for timing of the release of the updated and accurate report.”
Tiburon Inc. did not respond to questions of whether there was any human error, which police said had been the problem with incorrect 2008 UCR rape statistics.
THE HUMAN ELEMENT
Burton, who joined OPD as police chief in fall 2008, said that in early 2009 when he reviewed the past year’s numbers, he noticed the unlikelihood of only two rapes having been committed in the past year.
Sims, who began at her position in March 2009, said turnover in the records department meant not everyone was trained in coding data for UCR, and the human and programming errors were fixed for 2009 (there were 36 incidents of rape reported that year).
Burton said he also saw one of seven murders had been cleared in 2008 but assumed the cases just hadn’t been processed yet.
“I should have picked up on this in a more timely fashion,” Burton said.
But he pointed out that OPD’s mistakes weren’t intentional, malicious or advantageous.
“We’ve shortchanged ourselves,” Burton said. “We obviously need to do a better job of auditing,” something Burton said OPD would institute.
On Oct. 14, police officials twice confirmed only the October 2008 fatal shooting of Ray Sauceda Jr. and August 2009 stabbing of Roger Lee Velasquez Jr. remained uncleared by UCR guidelines during the past two years.
But a follow-up inquiry Oct. 15 for Odessa police’s internal homicide numbers discovered a third uncleared homicide in the past two years, that of infant Dustyn Bailey Wright.
Born Feb. 7, 2008, Dustyn died March 16, 2008, in an El Paso hospital after being transported from Medical Center Hospital. The crime was first categorized by Odessa police as injury to a child, described as a skull fracture and fractured rib in the police report. When the infant died, it became a murder case, but that fact was never upgraded for UCR purposes.
UPON FURTHER REVIEW
Odessa police last week provided revised figures using a more labor-intensive method. In that report, 35 of 39 murders have been closed from 2000 to 2009, a result which outperforms the state and national average; in reported UCR data, Odessa police originally gave 18 of 37 cleared.
One cleared case, according to UCR guidelines, is the May 2008 stabbing death of Ryan Adam Moon. The man arrested in the case, however, was acquitted. Still, it is a cleared case because enough probable cause was established for prosecution, and Odessa police said they consider the case closed.
Police said the September 2001 shooting of 70-year-old Lynn Floyd Moore was the fourth uncleared case.
The other unreported murder appears to have occurred in 2001 when police reported four of the six homicides as UCR murders. The July 2001 death of Leticia Gomez’s newborn child was first classified as tampering with evidence before the charge was upgraded to murder. Gomez eventually pleaded guilty to manslaughter in 2002 after the murder trial led to a hung jury.
A grand jury no-billed Odessa officer Chris Myers in the Feb. 4, 2001, shooting homicide of Dewayne Robert Barbee, 31, who was armed with a gun and sword. Myers was never arrested and as a justifiable homicide by a peace officer, this would be counted elsewhere.
The problem of faulty UCR submissions, whether resting with human error or a software glitch, is not confined to Odessa.
A Scripps News investigation showed homicide clearance rates had declined from 91 percent of all murders in 1965 to 64 percent in 2008. Scripps discovered cities from Florida to Michigan to California made serious errors in their reporting, for a variety of reasons.
Closer to home, Sweetwater Police Chief Jim Kelley wasn’t sure which program they used to submit their UCR information, but he said he knew it wasn’t accurate.
“The UCR doesn’t mean anything. I mean really, in the great scheme of things,” Kelley said. “I may get in trouble with the FBI (for saying this), but that’s the way it is.”
Kelley said his department knew its own clearance rate and although it had reported eight of 22 murders cleared during the past 25 years, 21 cases have been solved.
“It’s more important being out here working than looking at a book,” Kelley said. “Obviously, this book has some inaccuracies.”
Kelley wasn’t aware of any funding connected with the reporting.
Nor, initially, was Burton, but after looking into it, Burton said funds from the Edward Byrne Memorial Justice Assistance Program were in part reliant on UCR participation. The city of Odessa received $67,584 from the Department of Justice through that JAG program, of which almost $23,000 was shared with the Ector County Sheriff’s Office.
Michelle Person, a spokeswoman with the Office of Justice Programs, said the JAG program requires at least three years of data on violent crimes (murder, rape, robbery and aggravated assault) be reported through the UCR program to the FBI to be eligible for direct JAG local-formula funding.
Person said the Bureau of Justice Statistics calculates for each state a minimum base allocation, which can be enhanced by the state’s share of the national population and the state’s share of the country’s violent crime statistics, according to a formula.
“Once the state funding is calculated, 60 percent of the allocation is awarded to the state and 40 percent to eligible units of local government,” Person said.
Unless intentionally manipulated, clearance rates and accuracy have no effect on funding because funding is based on reporting crime. Even then, federal agencies will only send auditors to help get the numbers right.
“The FBI does not take any punitive action” for misreporting, said Stephen Fischer, a spokesman with the FBI’s Criminal Justice Information Services division.
According to protocol, the numbers should be reviewed multiple times by other agencies to root out errors, each month as they are submitted, and at the year’s end.
Nancy Carnes, a CJIS spokeswoman, said the FBI reviews information if it notices a change of more than 15 percent in any category in any direction. Carnes said state agencies also review each monthly report.
“There’s a mission that we report a reliable data set,” Carnes said. “If we find numbers are not correct, we may not publish them.”
Carnes said the FBI in part relies on state agencies to do a level of fact-checking.
Tom Vinger, a Texas Department of Public Safety spokesman, said the state’s UCR department also looked over Odessa’s numbers and never found anything wrong.
“Odessa is current in their UCR reporting, and there are no issues with their reporting,” the records department said.
“As I speak with you, I can’t attest to the accuracy of the UCR report,” Burton said last week.
That’s why the department has spent the past month rechecking numbers. And that work will continue — the headache Burton referred to earlier.
Burton stressed the UCR numbers are only one tool law enforcement has of counting and reporting crime numbers. But he said police continue to work to find out what all numbers have been affected by the software problem, and how to fix the problem so that accurate UCR data can be produced, first for 2009 then for the monthly 2010 reports that have already been turned in.
“The review of the UCR numbers is an overview of where we stand on the aggregate,” Burton said earlier this month. “We want to be sure it is accurate.”