Columbus Dispatch—Why Don’t Local, State and Federal Crime Numbers Add Up?

by Beth Burger
The Columbus Dispatch
October 17, 2017

Each of last year’s homicide victims had families and friends. Most had funerals and burials.

All have Columbus police homicide detectives assigned to their cases, pursuing their killers.

So it shouldn’t be difficult to figure out how many people were killed, right?

It depends on who you ask.

The FBI counted 91 people killed in the city in 2016. The Ohio Department of Public Safety recorded two more deaths, marking the total at 93. Columbus police, whose detectives are actually in charge of investigating the deaths, report that 106 people were killed.

When The Dispatch looked into the state’s tally of homicide numbers and cross-checked it with the city’s list, it found discrepancies, including:

• One woman was killed but counted twice as a victim.

• A woman who is still living was included on the list of homicide victims. (Her husband reportedly wanted to kill her but never followed through. A year later, they divorced, court records show.)

• A total of 14 homicides were missing from the state list. Among them are a grandmother who died after she shielded her grandchildren from an attacker. One young woman not on the list died of complications she suffered as a shaken baby in the 1980s. A man who was gunned down in his front yard isn’t counted.

“I would hope I never go under the knife with a surgeon who is as accurate as your murder numbers,” said Thomas Hargrove, chairman of the Murder Accountability Project, a Washington-based nonprofit group dedicated to studying homicide statistics.


The Dispatch pulled incident reports for the homicides missing from the state’s list. Each report shows the victim and the crime appropriately coded.

“I didn’t realize the state is showing different numbers,” said Dale Thomas, a programming analyst for the Columbus Division of Police in the Police Net Unit, which collects crime statistics from the software system PremierOne and sends a monthly report to the state.

Homicide isn’t the only crime category with discrepancies. Every federally tracked crime category is affected, a review showed.

The city and state use incident-based reporting statistics for crime. That means that if someone is raped, robbed and then killed, the crime data is counted three different ways — once for a rape, once for a robbery and once for a homicide. Uniform crime reporting would count only the homicide in that scenario because it’s the most serious crime. In either system, however, the homicide statistics should be the same.

In 2007, The Dispatch reported that the city had been using the incident-based reporting system for a few years and had similar discrepancies. A decade later, nothing has been done to correct the problem.

The state takes the data voluntarily submitted by each law enforcement agency and passes it on the FBI, which converts it to the uniform crime reporting statistics published each year.

“The validation process electronically applies edit checks to each incident reported to determine applicable data elements within each incident are correctly reported, including format and coding,” said China Dodley, a spokeswoman for the Ohio Department of Public Safety.

Thomas said the Columbus Police Net Unit routinely checks with the homicide unit to make sure the stats match. However, that same process doesn’t occur with the state, he said.

The accuracy of the state’s data depends on the law enforcement agencies that submit it, Dodley said.

“Any verification on the crime totals must be done through the reporting law enforcement agency, which in this case would be CPD,” she said in an email.

When accurate, databases can tell police whether their policing strategies are working. It can help them make the case for more funding to fight crime.

The city uses the state database when it generates statistical crime reports. The only exception is the homicide category because it’s always wrong, according to police. If the homicide totals are inaccurate, it’s unclear how off the other categories might be.

“Honestly, this matters. You can’t manage what you don’t measure,” Hargrove said.

Researchers, lawmakers, investors and real estate agents are just a few of the groups that sometimes use uniform crime reporting data from the FBI. Community members who are concerned about safety are given a snapshot of reported crimes through the statistics and the data is one way leaders can be held accountable.

Jeffrey Butts, director of the Research and Evaluation Center at John Jay College of Criminal Justice in New York, said that in some ways, law enforcement never has been best equipped for crunching the numbers.

Go to a police department and ask to be taken to the unit that does the crime reporting. Sometimes, it’s civilians with advanced degrees and computers, he said.

“Other places, you walk in and it’s full of uniforms,” he said, pointing out that some departments view crime reporting as a desk job for officers who no longer want to work patrol. At a department such as the New York City Police Department, there are civilian research analysts with doctorates.

Columbus sets preferred degrees for certain positions involving data, but they’re not required.

The Police Net Unit has two officers and a civilian who handle crime data and submit it to the state. A preferred degree for the unit is one in computer science. The division has a Research and Development Unit that uses the crime data to create internal reports citing statistics. A preferred degree for that position is a degree in English or statistics, Sgt. Dean Worthington said in an email.

When data-accuracy issues arise within police agencies, it can take years to correct. Other cities have dealt with similar issues, including in Chicago and the Pittsburgh area.

“Outside scrutiny is the only way these things get caught and improved,” Butts said.

Law enforcement agencies started collecting the data in the 1930s, when the FBI was tasked with collecting and maintaining the data. Uniform crime reporting, which is a voluntary reporting program, is still the most commonly used statistic, although there’s a push to switch to incident-based reporting.

Homicides are smaller in number compared with other reported crimes and can provide a benchmark.

“Homicides are supposed to be more controlled and standardized between jurisdictions,” he said. “It’s a real problem when you hear about errors in homicide reporting. It’s the one stat that allows comparisons across state and local borders.”

[ read the original ]