“The crime rate isn’t going down,” moaned a Twitter user. “The big cities are purposefully not reporting crime.”

Neither of the above is accurate, but there is a crime reporting problem, in that some people may be misinterpreting data … for political gain.

Take-aways

• There is significant, general nonreporting into the FBI’s new NIBRS system.
• Some well-documented hot counties with large populations are not reporting, which masks their gun crime and street gang data, including gun homicide rates.
• The systems that the Federal government has in place to encourage reporting into NIBRS are ineffective.

NIBRS Mayhem

For decades, the FBI orchestrated the Uniform Crime Reporting system (UCR). This was a nationwide program whereby all law enforcement agencies – city, county, state, campus, et cetera – were to report what crime incidents they processed. These were largely monthly counts of major categories of crime (robbery, murder, being a politician).

And it was nearly robust.

Both the FBI and the National Archive of Criminal Justice Data (NACJD) would interpolate crime data for regions where reporting was incomplete. This was acceptable because for these major summaries, the reporting participation rate was in the mid- to high-90% range, and the laggards were often small cities and counties with not much crime anyway.

But the system was not capturing enough detail, and computer technology had advanced significantly (for the geeks reading this, UCR was, to the end, encoded in EBCDIC and the data portions were referred to by the FBI as “cards”). So, the FBY devised the National Incident Based Reporting System (NIBRS) to replace UCR, testing it on several states for years. They spent a significant amount of time and effort notifying and educating the rest of law enforcement that their reporting into NIBRS required them to be ready… in 2021 – during the pandemic, when law enforcement agencies were losing staff and had COVID outages, and as with all big government projects, few agencies were ready.

Using the 2022 NIBRS database, we see that compared to the Wide-ranging ONline Data for Epidemiologic Research  (WONDER) database from the Centers for Disease Control (CDC), law enforcement reported fewer gun homicides than the CDC… in some geographies.

But these shortages are not evenly distributed by state, nor are they evenly distributed within a state.

The National State View

We used the CDC data to compare against the NIBRS data to see where the law enforcement agencies were behind on their reporting.

The CDC gets their data from the medical community, including from coroner departments. This is different than from the FBI getting data from yet-to-be-trained and COVID-absent clerks. That major gaps exist between the two, and that the CDC gun homicide counts are consistently higher than NIBRS counts, are indicators that a significant problem exists.

click to enlarge and share

This map shows the percentage difference for the states… and immediately shows a tricky anomaly, that some law enforcement agencies are reporting more gun homicides than the CDC does. Since NIBRS is new, there remains the possibility that local agencies enter data incorrectly, that agencies (i.e., a city within a county) are double-counting some homicides, or that in a mad rush to catch up, some 2021 homicides were lumped in with 2022 murders. The old UCR system allowed agencies to report previous year homicides anytime in the future, to be reconciled by the FBI afterwards. So the FBI might not have had control of the data inflow, or with only two years of national rollout, the post-processing either.

In short, it is a mess.

Much to the chagrin of party partisans, there is no Red/Blue divide in this data. Of states reporting at least 25% fewer gun homicides via NIBRS than via CDC, we see places distinct as Arizona and California, and Florida and Nebraska, though New York and New Jersey are significantly negligent too.

The National County View

Drilling deeper into the county level, we see a lot of problems that need explanation.

click to enlarge and share

click to enlarge and share

The CDC suppresses county-level data when the numbers are low (less than 10, but not zero). This means that we have an incomplete comparison between the CDC and NIBRS data, as this map shows (gray regions had both CDC suppressed data and no NIBRS reports of gun homicides). This left Gun Facts with some tedious spreadsheet work to substitute NIBRS for CDC data when the CDC labeled it “suppressed” (NOTE: data is not suppressed by the CDC at state level because the amalgamation of all county-level data masks any potential privacy issues vis-à-vis low numbers of instances in a locality).

Trying to visualize deficient counties on a map is difficult because the range of underreporting is wide. But this map gives a general idea, and you can see two realities that are concerning.

1. The number of counties where the CDC suppressed data disallows a comparison.
2. Some states have counties that are seriously underreporting.

But let’s go back to the first map that showed state-level data which does not suffer from CDC suppression. We see handful of states (Arizona, California, Florida, Louisiana, Mississippi, Nebraska, New Jersey and New York) that reported at least 25% fewer gun homicides than the CDC. On that point there may be some worthwhile insights. We’ll consider three of the worst of them (New York with 37% of CDC count, Mississippi with 26%, Florida with 30% and California with 39%) and look for patterns.

California

Yikes!

I get to say this because I lived in Cali for 22 years, mainly in Alameda County. I worked in downtown Oakland, and I know the turf, including its earned reputation as a gang city.

It is surprising that in the year of our investigation, the CDC recorded 118 gun homicides and the FBI’s NIBRS system noted only 6.

click to enlarge and share

click to enlarge and share

click to enlarge and share

For Los Angeles county, home to “south central” and the epicenter of the legendary Crips/Bloods war, the CDC knew of 530 gun homicides and the FBI only 35.

Also note, the gray colored counties are where there is no viable reporting from either the CDC (due to a low enough gun homicide count that they suppress the data) or the FBI (due to no reporting from the counties at all).

Florida

Bad, but in a different way. But… yikes!

I get to say this because I grew up in Brevard County, have relatives all over the state, and so I know the turf. Like California, some  counties and cities in the Hurricane State (or as antagonists say, the state that blows) have gang problems.

But underreporting in hot counties is secondary to no reporting in most counties (76% of counties had CDC suppressed data and no reports into NIBRS).

That said, some of the most likely counties to have gang hotspots are not comparatively delinquent. Duval (Jacksonville) underreports by 25%. Dade (Miami) is low by half (CDC reports 202 gun homicides, FBI logged 98).

But Broward… home to Fort Lauderdale (or as spring break visitors called it, Fort LiquorDale)… is seriously underreporting. Broward is Florida’s second most populous county after Miami-Dade County, and it is the 17th most populous in the United States. The CDC gathered reports on 126 gun homicides, and all the agencies in Broward noted just one.

New York

Yikes-squared!

New York has both the California and Florida diseases (insert your own joke here).

Seventy-four percent of New York counties have unusable data due to low, and thus suppressed, gun homicide counts by the CDC, and no reporting into FBI’s NIBRS.

But that isn’t the BIG problem.

Three counties – Kings, Queens and New York – all part of the New York City metro area, and with 33% of the state’s population – did not report any gun homicides into NIBRS. The CDC notes 84 gun homicides in Kings, 45 in Queens, and 35 in New York. Add to this the 86 gun homicides in Bronx. Then there were 23 in Suffolk and 10 in Nassau (together, Long Island with 15% of the state’s population) for which the FBI was given no data.

What to Make of This Mess

We’ll stop here because some patterns keep showing up:

1. Local police agencies are not reporting at a sufficiently high enough rate to give statistical comfort in the FBI’s data. The degree of interpolation necessary to fill in these blanks makes the FBI’s annual summary reports invalid.
2. In some high-population states, underreporting is tragic, perhaps even criminal.
3. In some states, well-known hot counties with ample street gang activity are not reporting at all.

And it won’t get better without congressional intervention.

The federal government cannot legally force state, county or local governments to do work for them. ¹ As such, reporting is voluntary. Under the old UCR system, reporting levels were “high enough” but still spotty.

Congress has tried to tighten things up, but the results are lackluster. There once was the Local Law Enforcement Block Grant funding where each jurisdiction was reviewed for missing data, but that vanished in 2006, being replaced by the Justice Assistance Grant.

But that is only a carrot… no stick.

The Justice Assistance Grant (JAG) program cannot force law enforcement agencies to report crime data to NIBRS. All it does is offer money if an agency has reported at least three years of data (either UCR or NIBRS) during the 2012–2021 period. But, as with most federal programs, the bureaucracy to (a) get NIBRS ready, and (b) apply for the grants, is daunting, especially for agencies that are already under-staffed.

To encourage participation in NIBRS, the FBI and Bureau of Justice Statistics provide money to state and local agencies to “get on board.” But the $120 million made available does not appear to have created minimally acceptable reporting.

This leaves researchers in a rough spot. As we demonstrated in our “Top 15 Murder County” analysis, gun crime is highly localized. The CDC data would be useful if they did not routinely suppress gun homicide data at the county level. Until either (a) the CDC admits that suppressing gun homicide stats is dumb or (b) the federal government finds a functional means for inducing high crime reporting participation levels, we have to accept that gun crimes are indeed being underreported.