FBI Crime Stats Are Unreliable, Shrouded in Mystery, and All Over the Place

The mainstream media has recently trumpeted the Federal Bureau of Investigation’s estimate that violent crime fell 3 percent nationally from 2022 to 2023. They have largely ignored, however, the latest iteration of the Bureau of Justice Statistics’ National Crime Victimization Survey (NCVS), which found a 9 percent increase in property crime in urban areas over that span—part of a 26 percent rise in urban property crime, alongside a whopping 40 percent surge in urban violent crime, from 2019 to 2023.

Set aside, for a moment, the media’s downplaying of BJS’s inconvenient urban-crime data. Are the FBI’s statistics really precise enough to make much of a reported 3 percent annual change in violent crime?

Trying to decipher how the FBI produces its statistics, and what those numbers even are, is a lot like combing through a crime scene and searching for clues. The agency’s processes, such as how it tries to “estimate” unreported figures, has long been a black box, even to the Bureau of Justice Statistics—the Department of Justice’s actual statistical agency. Things have become even more cryptic of late, as the FBI has struggled to implement its new crime-reporting system, adopted nationally in 2021. That new system makes year-to-year comparisons more challenging, in part because only 85 percent of agencies provided data for 2023—in other words, the FBI is capturing only a portion of crimes reported to police.

The FBI also has a history of offering multiple and conflicting crime figures, which makes precise cross-year comparisons difficult. For example: How many violent crimes were committed nationwide in 2020? The table linked to on the FBI’s Uniform Crime Reporting Program (UCR) website—its “Source Data”—indicates 1,272,812 such crimes committed nationally in 2020. But the bureau’s 2021 press release announcing the publication of 2020 crime data pegged the figure at 1,277,696. Then, in 2022, its press release highlighting the publication of 2021 crime statistics claimed the 2020 figure had actually been 1,326,600. Add in the FBI’s Data Discovery Tool, which currently lists the 2020 total as 1,304,574, and you have four different figures for one statistic.

Or take the number of murders (including instances of non-negligent manslaughter) in 2021. The FBI’s press release announcing the agency’s 2021 crime numbers declared that there had been 22,900 murders that year, a 4.3 percent increase from 2020. Today, however, the UCR’s “Source Data” claims that there were 21,462 murders in 2021—a 4.7 percent decrease from the prior year. Apparently, over 1,400 murders went missing between the time the FBI issued that initial press release and today. Moreover, the agency’s Data Discovery Tool lists the 2021 homicide total at 19,563—more than 3,000 shy of the originally released tally.

This is not to say that the FBI is doing anything nefarious, but rather, that the media are putting too much faith in the crime data released by an entity that isn’t a principal federal statistical agency. The bureau appears to have adjusted its numbers at different times in different ways, to compensate for missing data or to incorporate previously missing data. But its process is hardly transparent, its results are often inconsistent and poorly explained, and its numbers are frustratingly fluid.

A press corps eager to report a 3 percent drop in violent crime from 2022 to 2023, per the FBI’s “estimated” numbers, might want to think about how reliable such estimates are—given that one can find evidence from the bureau of either a 6 percent or 15 percent drop in its own reported number of murders committed in a given year, depending on the source. And that’s for murder, the easiest crime to count.

More fundamentally, the FBI cannot capture crimes that aren’t reported to police. Victims responding to the NCVS—which has been around since the Nixon administration and is one of the largest federal surveys on any topic—say that most violent crimes (55 percent) and property crimes (70 percent) aren’t reported to police. So, if victims are to be believed, the FBI’s data doesn’t capture most of the criminal activity committed in the United States.

Even if the FBI’s current, presumably fluid, estimate that violent crime fell 3 percent from 2022 to 2023 is roughly accurate—and even if that figure holds for the unreported crimes that the FBI can’t capture—Americans can be forgiven for thinking that this pales next to a 40 percent, four-year rise in urban violent crime, or next to a 26 percent, four-year increase in urban property crime. The press wants to wish this away to avoid reckoning with the human costs of lax law-enforcement policies.

Source: City Journal