Crime Numbers

People claim crime statistics lie. With the 2020 numbers up, here's how to tell what to trust.

The FBI released the annual Uniform Crime Report yesterday. It shows an increase of about 30 percent nationally in homicides and about 21 percent in Georgia overall. Atlanta recorded 157 homicides in 2020, an increase of 58. The state overall had an increase of 99 homicides. Atlanta has about 5 percent of Georgia’s population, but 59 percent of its increase in homicides.

I write this newsletter because I’ve been vocal for years about how the media makes money from selling fear of crime — and particularly Black criminals. Up until last year, I would explain, over and over, that violent crime in America peaked around 1994 and has been in an unprecedented decline ever since. Most of America for the last 15 years has been as safe as Boomer-era Mayberry.

I would reference the Freakonomics chapter on crime and abortion, a quasi-eugenics argument explaining the fall in violent crime — which has since been revised. I would note the ongoing argument about the role of lead contamination in crime rates. I would talk about how high school graduation rates have been rising among the poor, driving down crime through better opportunities. And always, always, always, I would reference the FBI’s Uniform Crime Report data confirming the drop.

As often as not I’d be met with a shrug, a personal anecdote, and an assertion without further evidence that crime is getting worse not better. Many people don’t care about facts. They want their biases affirmed. But now that crime actually is rising substantially, I’m left in the unenviable position of trying to convince people who were already twitchy to approach the problem with rationality and empathy.

That starts with an honest understanding of the data.

After many years below the national average, Georgia is now at par for overall violent crime per capita with the rest of America. It takes us back to rate of 2016. The odds of any random person in Georgia being the victim of a violent crime that was reported to the police were about one in 250 in 2020.

The FBIs figures differ from those of the annual National Crime Victimization Survey, which lags UCR data by a few months. The Department of Justice surveys the public to measure how many were crime victims, rather than taking the reports made by police departments. The NCVS suggests that one’s odds of being the victim of a violent crime are closer to one in 130.

The NCVS doesn’t have 2020 figures yet. Based on the 2019 survey, however, less than half (41%) of violent victimizations are reported to police. The percentage of violent victimizations reported to police was lower for white victims (37%) than for black (49%) or Hispanic victims (49%).

Like the UCR, the NCVS has been tracking a long-term decline in violence. While one measure may be different than the other, they’re both saying the same thing.

The UCR numbers for Georgia this year may be a little funky. In 2020, 398 agencies provided numbers. In 2019, 122 fewer Georgia agencies reported crime data to the FBI. That’s because Georgia switched over in October to the National Incident-based Reporting System, or NIBRS ,in October 2019, but many smaller cities and towns weren’t prepared for the change.

“The FBI only publishes agencies that have 12 months' worth of good data; which also accounts for the decrease in 2019,” said Nelly Miles, spokesperson for the Georgia Bureau of Investigation. “The increase in 2020 is due to agencies updating their Records Management Systems to capture NIBRS data, training, and outreach to the agencies.”

NIBRS is a better system, flat out. Under the old SRS system, if a robbery and a murder happened at the same time and place, it would only count the murder in the figures — only the most major crime was tabulated. NIBRS also captures the day and time of the crime and the relationship of the victim to the offender. We also get more demographic details, location data, property descriptions, drug types and quantities, the offender’s suspected use of drugs or alcohol, the involvement of gang activity, and whether a computer was used in the commission of the crime.

It’s easy to accuse individual police agencies of fudging their numbers. The value in the FBI’s data collection process is that it makes it very difficult to massage the numbers for the country as a whole. Long-term patterns become clear. One has to argue that police agencies across the country, large and small, are all engaged in some kind of conspiracy to either inflate or deflate their statistics in exactly the same way to argue that the overall count is substantially incorrect.

Not that it keeps people from making that argument.

Nonetheless, I tend to focus on homicides when looking at crime statistics because there’s almost no subjectivity. A fistfight might be a simple assault or an aggravated assault, depending on the arresting agency, someone refusing an ambulance and whether Saturn is in estrus or something. Robberies and burglaries can fluctuate depending on whether victims bother to report the crime. Most rapes go unreported.

But with a murder, there’s almost always a dead body at your feet, and it usually has bullet holes in it. Three-quarters of homicides in Georgia are committed with a firearm, usually a handgun.

Georgia’s homicides have regressed back to the murder rate of 1996. The reason for the disparity between the spiking murder rate and relatively tepid change in overall violence is that robberies have plummeted. A digital economy makes a street robbery less profitable, and the incident rate for “violent crime” counts a street robbery the same as a rape or a murder.

Now, many, many more people die of accidents every year than are murdered. About as many people die in car accidents as are murdered, for example. Twice as many people fall to their death as are gunned down. About three times as many are poisoned. But the numbers are steady. If police were misclassifying murders as accidents, we would see it in the data, unless every agency was doing so in exactly the same way.

I suppose it’s possible that some of those deaths are actually murders and we are living in a Hobbesian homicidal hellscape of cut brake lines, forced drownings and — oh, I don’t know — deliberate injections of bleach or imbibing ivermectin in poisonous quantities. But you’d have to prove that to me, and it’s not in the numbers.