The first presidential debate is set for Thursday night, and there’s a good chance that the major party candidates will resume a recent squabble over crime statistics from the stage in Atlanta, Georgia.
The dispute started after President Joe Biden released a statement earlier this month touting a new quarterly report from the FBI showing dramatic crime rate reductions between 2023 and 2024. Experts say the decline is real, but that the numbers likely exaggerate the extent.
Former President Donald Trump was quick to criticize the data’s limitations, claiming crime is “so much up.” He cited a different data source that is only current through 2022, based on a survey of Americans to determine if they were victims of crime. And like Biden, Trump missed part of the story by zooming in to cherry-pick a statistic in his favor. Zooming out to see the big picture, the overall trend in crime victimization suggests that by 2022, crime had returned to pre-pandemic norms.
This messy dispute on crime data points to a more serious problem: The national crime statistics leave yawning gaps, and collecting better crime data should be a nonpartisan issue, said John Roman, director of the Center on Public Safety and Justice, NORC at the University of Chicago.
“The data itself shouldn’t be controversial,” Roman said. “The more we can improve the crime data, the less controversial it is, and the more we can have policy debates that are based on different approaches to solving the problem, rather than debates about what the facts are.”
Roman also chairs the Council on Criminal Justice’s working group on better crime data. Earlier this week, the group put out a report urging an increase to the Bureau of Justice Statistics’ budget to $93 million so it can, among other things, produce monthly estimates of national crime statistics.
But in lieu of better national crime statistics, if the issue comes up in the presidential debate, here’s what we do know about crime trends.
Biden says crime is way down. Where does that data come from?
The recent White House statement about falling crime in 2024 cited the FBI’s quarterly crime data, which included numbers from 11,000 police agencies around the country and trumpeted a 26% decrease in homicides. “Americans are safer today than when I took office,” Biden said in the statement.
Nearly 8,000 police agencies weren’t included in the quarterly crime statistics report, which is a relatively new offering, first issued in 2020. It was partly introduced because of the lag that has long plagued national crime reporting. Historically, the agency has only released data annually, about 10 months after the end of the year. While the quarterly reports are more timely, they are also preliminary because crime data often changes with police investigations, and fewer agencies submit information to the quarterly report than to the annual one. As a result, they sometimes contains errors, according to criminologists.
For example, the FBI’s quarterly data recorded 46 homicides from the New York Police Department, but the department’s own data showed 82 murders from the first quarter of this year. In Dallas, where the FBI’s data documented 33 murders, Texas’ crime data dashboard showed 50 such incidents.
A spokesperson from the NYPD said that the department is adjusting and adopting its data reporting system to the federal government’s new standards, and as a result, some of the data initially reported to the FBI might be rejected and not counted because “it may not entirely conform with these new data reporting standards.” Overall, 98% of the data that NYPD reported to the FBI was accepted, the spokesperson said.
But of the largest cities where The Marshall Project compared FBI’s statistics with the local statistics, the numbers are spot on or show little differences for most places.
It’s not uncommon for crime data to change, said Darrel Stephens, a former executive director of the Major Cities Chiefs Association. Sometimes the crime classification can change, and sometimes crime reported to the FBI can initially be incomplete.
“I don’t worry much about mismatches like that,” Stephens said. “If the information gives you a sense that there is an increasing or decreasing problem, or it can give you some insight into the nature of the problem, then whether it’s exactly precise doesn’t matter to me.”
Trump said crime rates are actually up — is he right?
As Biden and the Democratic Party attempt to use falling crime rates to boost voters’ confidence in the administration, Republicans are pointing to flaws in the data and accusing Biden of cooking the books.
“Biden said crime is down,” Trump said in a rally in Philadelphia last week, “crime is so much up.” Then, after a long digression about asylum seekers, political correctness, fake news and Biden’s age, Trump returned to the subject. “According to the much better crime victimization survey, there has actually been a 43% increase in violent crime since I left office,” Trump said in the speech that aired live on Newsmax.
Trump is right that the National Crime Victimization Survey is considered a more robust data source than the preliminary quarterly numbers from the FBI. However, the most recent NCVS is nearly 18 months out of date and says nothing about what crime rates look like right now. Trump also left out that the same year’s FBI crime data — the other national gold standard — showed a slight decrease in 2022.
That discrepancy between the the FBI’s Uniform Crime Reporting program and NCVS numbers was a mystifying phenomenon in the 2022 crime statistics, made possible because the two reports measure different things. The FBI crime data represents incidents reported to the police and is useful for understanding law enforcement’s view, while the victimization survey relies on interviews with over 150,000 families and is useful (albeit broadly) for understanding how people experience crime.
Criminologists have some hypotheses about why the Justice Department’s two pillars of crime statistics show such different year-over-year trends. For one, the victimization survey is historically much more volatile from one year to the next, as seen in the chart above — so it may be influenced by statistical noise. Second, crimes reported to the police are subject to factors like police staffing levels, police response time and overall levels of community trust.
Most importantly, the victimization survey captures a slightly different time frame. Using the 2022 reports as an example, while the FBI’s crime data collects information from the calendar year, the victimization survey covers victims’ experiences from July 1, 2021 to Nov. 30, 2022. Since the NCVS shows an increase in violent crime, it’s potentially because violent crime rates were higher in the latter part of 2021.
Despite the one-year differences, the longer trends in FBI’s crime data and the victimization survey both show that violent crime is returning to pre-pandemic levels: After the initial increase in violent crimes and decrease in property crimes at the beginning of the pandemic, now violent crime is going down, and property crimes are seeing an uptick, but not surpassing where they were before the pandemic.
Violent crime and victimization rates return to pre-pandemic level
OK so the best crime statistics are slow, and the faster ones are incomplete. Can we say anything with confidence about how things look today?
All the best available evidence we have today suggests that crime is indeed going down. But until the 2023 FBI crime statistics and the Bureau of Justice Statistics’s victimization survey are published in the fall, our national picture remains murky.
Locally, data from individual police departments can convey a more accurate — though patchy — picture about crime. Many dashboards, studies and media coverage relied on data compiled directly from police agencies. Most of them showed an overarching decrease in murders and shootings, though there are always outliers, and it’s hard to draw nationwide conclusions when data from suburban and rural police agencies are far less accessible than in big cities.
Here are a few examples of how non-government organizations are tracking crime statistics:
In all of these data collection efforts, the violent crime trend is the same, with fewer murders and shootings each year since 2022.
But the limitations of these data are just as clear: The majority of efforts for more timely crime data rely on data from major cities and police departments. They do not capture trends in the suburbs or rural areas.
“Only the federal government has the resources to capture everybody, rural areas, urban areas, non-metropolitan counties,” said John Roman. “This is the reason why we put so much emphasis on improving federal government reporting.”
Is the FBI “cooking” crime statistics?
When the FBI’s crime statistics started to show a decrease in violent crime in 2023 and 2024, Trump and his supporters cast doubt on the finding, going as far as claiming that the FBI is manufacturing the crime statistics in favor of Biden.
Several criminologists told us this claim is completely unfounded.
On the contrary, because so many police agencies already publish weekly — if not daily — crime data, and because so many non-government organizations are tracking crime statistics on their own, these sources provide another layer of verification for the FBI’s crime data.
And right now, every source points to a decrease in violent crime. “This means the FBI’s Q1 2024 data is incomplete, not inaccurate,” said Alex Piquero, a criminology professor at the University of Miami.
The preliminary quarterly estimates are usually good enough to see the big-picture trends, such as drops in violent crimes. But the scale of the trend is far fuzzier than the statistics cited by the White House.
“There’s no fudging of the numbers, and the drop is real,” Piquero said. “The question, of course, is how big that drop will be, and then how big that drop will be across crime types. That’s the thing that we just don’t fully grasp yet.”
Why is the federal government so slow at tracking crime data? Has anyone tried to make it faster?
The national crime collection program has historically lacked resources. And for a long time, the general assumption has been that local law enforcement agencies should be responsible for collecting crime data, and the federal agencies should just sum up the numbers at the end of the year. Increasingly, Roman said, that’s become “an antiquated view of the world,” because policymakers need more real-time data.
“When COVID happened, you saw how good the data were — the tracking of cases, the testing and the deaths,” said Piquero. “Violence is a public health issue, and we have to treat the data just as we treat public health data.” That position — one that’s gained in popularity in recent years — got a new high-profile ally this week when U.S. Surgeon General Vivek Murthy declared firearm violence a public health crisis.
Trump and Biden both inherited the data lag problem, but neither did much to improve the national crime statistics, and it only worsened during their presidencies.
When Trump took office in 2016, the FBI was already in the process of modernizing its crime data collection system, announcing that it would retire a century-old data collection by 2021. By Jan. 1, 2021, the FBI decided to go through with its plan, and thousands of police agencies could not report data to the FBI because of that.
As Biden took office, the administration had its hands full with the pandemic. “Is crime data reporting the highest priority of [the Biden] administration then? Probably not,” said Piquero, who served as the director of the Bureau of Justice Statistics from 2022 to 2023. “Should it be? In my mind, crime data should be just as important as any other data collection.”
In 2022, the FBI rolled back its decision and told police agencies that they could submit data through the previously retired system, a decision that continued into the following years. But in many places, the transition continues to cause confusion and errors in crime statistics.
The FBI is expected to release the 2023 national crime stats this fall, how good will they be?
By the end of 2023, over 15,000 police agencies reported their crime statistics to the FBI. Of those agencies, 13,770 submitted data through the National Incident-Based Reporting System, while almost 1,500 submitted data through the Summary Reporting System — an older, more limited system that had previously been retired by the FBI.
Together, these agencies covered nearly 85% of the U.S. population. That’s a significant improvement from 2021, when the FBI temporarily retired the older data collection system and received data from only 60% of police agencies in the country.
Most notably, in 2023 the New York Police Department started submitting data through NIBRS — the new, more granular system — meaning the largest police department in the country is now reporting data to the federal government.
While the majority of states saw nearly every police agency submitting data to the FBI, a handful of states like Florida, Mississippi, California and New York State still fell behind.
