The National Crime Victimization Survey 2023 Is Out

Some thoughts on the latest NCVS.

The Bureau of Justice Statistics released its annual National Crime Victimization Survey (NCVS) for 2023 yesterday which marks the start of a series of crime data releases. NCVS and the FBI’s Uniform Crime Report (UCR) each form one half of the nation’s crime measurement systems. They’re usually in agreement, but UCR showed a decent sized drop in violent crime in violent crime in 2022 while NCVS pointed to a massive increase that year.

I wrote about how I reconcile this disagreement, saying:

“In my opinion, NCVS and UCR are telling the same story which is obscured by differences in methodology which add confusion to the story. Both measures tell us that the nation’s violent crime rate in 2022 was substantially lower than it was in the 1990s, largely in line than where it was over most of the last 15 years, and likely slightly higher than where it stood in 2019.

NCVS and UCR disagree mostly about what 2020 and 2021 looked like which — in my opinion — is largely irrelevant at this point. This is especially true given that NCVS doesn’t measure murder and the major change occurring in 2020 and 2021 was a historic surge in murder.”

The new data from the BJS seems to affirm this approach as reasonable.

There are four main caveats that people must be aware of when thinking about what NCVS shows in terms of US crime trends. I went into depth about these caveats in the piece linked above, but I’ll reiterate them here as a reminder.

1. NCVS is a survey and so it carries margins of error. Those are reported in the report's appendix, but if you go by the baseline number without considering the margins of error then you’re doing yourself a disservice.

2. NCVS doesn’t estimate murder counts. Murder skyrocketed in 2020 and has been falling ever since while other forms of reported crime have been largely even (with the notable exception of auto thefts).

3. NCVS measures the year of the survey, not the year of the crime. If you’re interviewed for NCVS they’ll ask you whether you have been the victim of a crime within the six months prior to the month of interview. That means that the July 2022 robbery you tell your interviewer about in January 2023 gets counted as a 2023 offense. This usually isn’t a huge deal over the long run, but it does obfuscate when exactly changing trends occurred. It also should be recognized if someone wants to say that crime did X in Y year per NCVS.

4. 2020 and 2021 were particularly challenging years to survey. There was a pandemic, it made in-person surveys difficult. We decided to potty train toddler triplets at the start if quarantine. It was a mistake.

With those caveats in mind, NCVS estimates that violent crime declined a bit in 2023 (I used a BJS tool called N-Dash to get the data for the below charts). The decline was clearly within the 2022 confidence interval so BJS calls it “consistent” between the two years and that seems wise. The 2023 estimate’s confidence intervals are largely in line with most of the decade from 2010 to 2019 as shown below.

BJS notes the overall drop in violent crime since the 1990s while summing up violent crime over the last five years by saying: “While the 2023 rate was higher than those in 2020 and 2021, it was not statistically different from 5 years ago, in 2019” (emphasis mine).

My general approach is to throw out the 2020 and 2021 figures. The argument that violent crime skyrocketed in 2022 and 2023 requires one to argue that violent crime was at the lowest level since the 60s and 70s in 2020 and 2021, murder surged nationally those two years as violent crime otherwise plummeted, and violent crime now has increased to remarkably high levels not seen since…2019. The more logical explanation is that the years that were uniquely hard to survey produced weird results which shouodn’t influence our conception of the nation's current crime trends.

So what else is in NCVS?

My favorite part of each year’s report is the quantification of the percent of victimizations that were not reported to the police each year.

The share of violent crimes reported to the police went up in 2023 though it was largely in line with recent years when considering the margin of error. It doesn’t prove that violent crime is suddenly being reported more often, but it does seem to run counter to any narrative that underreporting of crime is responsible for our current crime trends. NCVS doesn't seem to point to any major recent changes in crime reporting patterns.

As to the reporting of individual crimes, most were similar in 2022 and 2023 though the share of auto thefts being reported to police fell over 8 percentage points and the share of rape/sexual assaults that were reported more than doubled. The 46 percent of rape/sexual assaults estimated to have been reported to police is the highest percent reported since 2010.

I’m not sure there’s a huge takeaway from the change in reporting patterns other than that these figures can bounce around a bit from year to year. The margin of error for auto thefts getting reported, for example, is between 64 percent and 81 percent. The margin of error for rape/sexual assaults is +/- 8.2 percent, much larger than last year’s 3.5 percent.

NCVS is an incredibly useful tool for outlining US crime trends and helping to measure what does and does not get reported to the police each year. All crime data is imprecise, but NCVS — as a survey — carries an additional (and measurable!) imprecision that should be kept in mind when discussing its trends.

NCVS tells us that violent crime in the US probably fell a small bit in 2023 and is roughly where it was in 2019. That was my guess about what the FBI will say last week, and nothing in the 2023 NCVS changes that guess.

We’ll get the FBI’s Uniform Crime Report for 2023, the second quarter FBI 2024 statistics, and the July update for the Real-Time Crime Index all in the next few weeks.

NCVS, therefore, is the appetizer and official kickoff of Crime Data Fall. Let the crime data commence!