Exploring the Chasm Between How Democrats and Republicans Perceive Crime Trends

Partisanship is a helluva drug.

I wrote a piece last December about how Americans are bad at perceiving crime trends. In it I tried to connect the dots between why 77 percent of Americans said crime was rising in 2023 when it clearly was not.I came up with five main explanations: “Crime” is poorly defined, the questions aren’t easy to answer definitively, we live in a data vacuum, the media doesn’t cover the planes that land (well, most of them), and partisanship.

Americans Are Bad At Perceiving Crime Trends

A year has passed and Gallup released its 2024 survey recently showing that the share of Americans who said there is more crime in the US now than a year ago fell a fair amount to 64 percent. The FBI reported a small drop in violent crime in 2023 and early data points to an larger decline in both violent crime and murder occurring in 2024, so perhaps perceptions of crime are starting to align with reality. That’s the lowest share since 2019 and pretty accurately reflects murder and violent crime being largely at the same relatively low level now as it was five years ago.

Or perhaps not.

Breaking down the survey by political party though shows an enormous gap between how Democrats and Republicans perceive crime trends. While the overall assessment of crime graph looks relatively normal, the partisanship graph shows just how broken Americans’ perceptions of crime really are these days.

That’s a spit-take of a graph right there.

Partisanship has certainly played a role in peoples’ perceptions of crime over the last 25 years with the party that holds the presidency tending to think crime was climbing less than the party that didn’t hold it. Still, the gap wasn’t that large until 2021, and the gap between Republicans and Democrats went from a very high 34 percentage points in 2023 to an astounding 61 percentage points in the 2024 survey.

To be clear, neither of the nation's crime reporting systems point to rising crime in 2023 or 2024. Both FBI and NCVS showed a slight decline in violent crime with the former showing a very big drop in murder and the latter showing a tiny increase in property crime. FBI data and reporting from numerous other sources point to declining reported crime nationally in 2024 though we won't be able to compare that to NCVS 2024 for a while. Crime isn't plunging (other than murder) either, but the above results show perceptions are farther from reality than they've been since at least when Gallup began this survey.

The explosion of partisanship as a leading cause of how people perceive crime trends is relatively new and is likely primarily driven by political factors that aren’t really at the heart of this newsletter. But I believe the problem is exacerbated by three non-political factors that allow for the gap to reach this point, and it’s those that I want to explore for the rest of the piece.

What is ‘Crime’?

Gallup has been asking Americans whether there is more or less crime than a year ago for more than 30 years, but ‘crime’ can mean a lot of things to a lot of people. Does crime mean violent crime as defined by the FBI’s Uniform Crime Report Summary Reporting System program? That’s murder, rape, robbery, and aggravated assault. Does it mean property crime? Does it just mean murder?

Murder is falling at the fastest rate ever recorded, but murder also makes up around 0.2 percent of all UCR Part I offenses recorded each year. Property crime accounts for around 80 to 85 percent of all UCR Part I offenses but property crime usually carries less societal cost and also tends to be underreported more frequently.

It’s also plausible that one person’s perception of crime is influenced by property crime while another is thinking just of murder. If Republican respondents are thinking of crime as motor vehicle thefts and Democrats are thinking of it as murder then they were both right in the 2024 survey given that murder is plunging while motor vehicle thefts have surged. And crime trends are not static so there’s no reason to believe that one person’s impression of what makes up crime in 2004 is driven by the same definition of crime in 2024.

There’s also disorder which Charles Lehman has written and talked a bunch about recently and makes sense as (generally) non-criminal factors that people may be lumping into their perceptions of ‘crime’. Charles writes:

Disorder is not measured like crime—there is no system for aggregating measures of disorder across cities. But if you look for the signs, they are there. Retail theft, though hard to measure, has grown bad enough that major retailers now lock up their wares in many cities. The unsheltered homeless population has risen sharply. People seem to be controlling their dogs less. Road deaths have risen, even as vehicle miles driven declined, suggesting people are driving more irresponsibly. Public drug use in cities from San Francisco to Philadelphia has gotten bad enough to prompt crack-downs.

Unfortunately, we are left with more questions than answers without a better understanding of exactly what people think of when they think of ‘crime’ and the usefulness of the survey is lessoned without those issues being addressed.

The Survey

It’s also important to note that the survey does not have every year, making it more challenging to evaluate how perceptions of crime declined in the 1990s. The survey is missing 1993 through 1995 and 1999. The stretch from 1993 to 1995 is when the US great crime decline really began, but it also accelerated in 1996 when the survey picks back up.

Violent crime fell 7 percent between 1992 and 1995 and murder fell 9 percent, but violent crime fell more than 6 percent while murder fell over 9 percent in 1996 alone. How perceptions of crime changed during those critical years would be really nice to know, but those years are missing.

The mid and late-90s is the last time that murder fell so suddenly as it’s falling now, so not being able to really compare year-to-year changes in perceptions then to now is not ideal.

The Changes Are Smaller

Finally, crime is changing less each year than it did in the 1990s which undoubtedly makes it harder for average citizens to accurately perceive the real trends. There were 506,000 fewer violent crimes reported by the FBI in 1999 compared to just seven years earlier in 1992. The FBI estimated that there were just 168,000 fewer violent crimes in 2022 compared to 1999. Obviously the US has grown considerably in the last two decades leading to a lower rate, but the ups and downs of violent crime have been far more muted over that span than they were in the 1990s.

Murder is falling fast in 2023 and 2024, but accounting for a very large decline in 2024 puts the number of murders in the US in 2024 at roughly the same number as were estimated 25 years ago. The murder rate will be considerably lower thanks to a population of 50 million more people, but the year to year changes over the last 25 years have been much smaller than when the FBI estimated 9,000 fewer murder victims in 1999 compared to 1993.

Reported violent crime fell a LOT in the 1990s, but the decline has been much more gentle over the last few decades. Reported violent crime is still declining, but it may be less perceptible given that the slope is not as steep now as it was the mid and late-90s. Reported violent crime has still gone up and (mostly) down from year to year, but for the most part violent crime today is about as common as it was 6 years ago. The violent crime rate has been between 363 and 390 per 100k every year since 2011 (excluding 2021’s flawed estimates).

Consider this graph showing the share of Americans saying crime rose over the last year along the X-axis versus the change in violent crime totals in each year of the survey compared to the total 5 years prior. This graph doesn’t include 2024 data since the crime data isn’t out yet (so no 2024) and I’m using the FBI’s original estimates for 2021 which were flawed but less so than the revised figures in my opinion (the graph doesn’t really change excluding 2021 and 2022 though).

The relationship is a little less strong when using murder instead of violent crime, but it’s still there. There’s very little correlation when using property crime.

In my opinion, this more muted change in both murder and violent crime over the longer term opens the door for extreme partisanship to take an outsized role in impacting whether people think crime is going up or down. And that it has in the last few years with things accelerating in an election year.

Americans rightly recognized when violent crime was surging in the late 80s and early 90s, recognized it when violent crime was plummeting in the late 90s, and has largely hovered around 60 to 80 percent saying it’s rising regardless of what actually was reported ever since. Absent a major change in the pattern Americans have defaulted to mostly thinking it’s going up.

Detaching perceptions of crime from realities of crime makes a bad recipe for smart policymaking. One cannot learn from successes if you always think you’re failing, and policymakers will constantly feel an incentive to respond harshly to rising crime even when it’s falling. The gap between reality and perceptions is part of the reason that projects like the Real-Time Crime Index exist, to close the data gap and help people figure out what is real and what is wrong.