When math becomes misinformation
The Associated Press just proved that sometimes the most dangerous propaganda uses accurate numbers to bury the truth
Here’s a journalism pop quiz:
If Group A commits violence against Group B at a rate of 15%, and Group B commits violence against Group A at a rate of 12%, are these roughly equivalent?
If you answered yes, congratulations — you’re qualified to write for the Associated Press.
In response to the viral stabbing of a Ukrainian refugee on a Charlotte train, the AP published what might be the most statistically dishonest piece of journalism this year. Not because they lied — every number they printed was technically accurate. But they arranged those numbers like a three-card Monte dealer, shuffling facts so fast that readers couldn’t possibly follow the actual story.
The AP editors even knew how inflammatory the piece was because they served up different headlines depending on the audience. For member newspapers, the actual AP wire went with “Train stabbing spurs outcry over Black-on-white violence, but data shows such occurrences are rare.” But on the public APNews.com website, a woman’s murder was dismissed as “fuel for old racial tropes in politics and media.” Evidently, journalistic bias comes in different strengths.
The piece itself opens by acknowledging a brutal murder caught on video, then immediately pivots to warning readers about “bitter racial and political rhetoric” and “age-old harmful narratives about Black criminality.” Fair enough — sensational crimes shouldn’t drive policy. But then the AP commits journalistic malpractice so egregious it makes those “harmful narratives” look like doctoral dissertations by comparison.
Let’s follow their shell game. First, they tell us “Black offenders were involved in about 15% of violent victimizations of white people” while “White offenders were involved in about 12%” of crimes against black victims.1 See? Basically the same!
Except buried 500 words later, they casually mention that black Americans comprise 13% of the population while whites make up 56%.2 They even quote a criminologist explaining that “there are just more white people that could be potential offenders.”
Did you catch that? Let me do the math the AP refused to do:
When you adjust for population, an individual black person is approximately 23 times more likely to commit a violent crime against a white person than vice versa.3
That’s not a rounding error. That’s a canyon-sized disparity the AP buried under an avalanche of obfuscation.
They quote someone saying it’s wrong to use this crime to “resurrect racist talking points about the Black community.” But they never tell us whether those talking points are factually accurate. The AP doesn’t even attempt to dispute the accuracy. They just label discussing accurate data as “harmful narratives” and move on.
If racism means believing false things about racial groups, then accurate statistics cannot be racist by definition. Math isn’t racist. Per capita crime rates aren’t racist. They’re just data. But by labeling accurate statistics as “racist talking points,” the AP has created a category where truth itself becomes problematic.
This is textbook “malinformation” — a newly invented censorship word for information that is true but deemed harmful because of how it might be used. The AP is essentially admitting these statistics are correct but declaring them too dangerous to state clearly. They’re telling us that stating “15% vs 12%” without population context is responsible journalism, but noting the 23x per capita difference is “harmful.”
Statistical malpractice doesn’t just misinform, it radicalizes — in two directions. When mainstream media outlets twist themselves into pretzels to avoid uncomfortable realities, they don’t make those realities disappear. They just convince readers that the media can’t be trusted with fourth-grade math.
But there’s an even more insidious problem. The readers who believe the AP’s false equivalence — who walk away thinking interracial violence is symmetrical (or worse, the reverse of what it actually is) — are being radicalized too, based on misleading (to be generous) narratives.
The AP seems to think that discussing actual crime statistics is bigotry, that noticing patterns that really exist makes you racist. Once you’ve declared truth itself problematic when it’s inconvenient, you’ve abandoned any pretense of caring about reality.
If the supposedly objective press can’t present basic statistics honestly, how can we expect activists and politicians who have actual, admitted bias to debate solutions across the same set of facts? When the AP — which feeds thousands of newspapers — can’t do simple division, it poisons every downstream conversation about crime, justice, and policy. We end up with two Americas living in completely different mathematical realities.
The most revealing moment: “Some criminologists caution against relying on raw count crime numbers as it relates to the race of victims and offenders because population size matters.”
Some criminologists? As opposed to the ones who think per capita analysis is optional? This is like writing “some mathematicians believe multiplication exists.”
I don’t know why the disparities exist between white and black crime rates. Perhaps it’s poverty, education, historical injustice, cultural factors or some complex interaction of all of the above and then some. But we can’t even begin that conversation when our media gatekeepers are playing statistical Three-card Monte with basic facts.
The AP piece ends by intoning that violence “isn’t the domain of Black Americans alone.” Of course it isn’t. No one claimed it was. But when the per-capita rates differ by a factor of more than 20x, pretending they’re equivalent isn’t journalism — it’s gaslighting.
We’ve reached a point where doing arithmetic correctly is treated as more dangerous than doing it wrong. Where propaganda with footnotes is preferred to accurate statistics. Where avoiding racism apparently requires denying reality itself.
The stabbing happened. The statistics are what they are. Calling people racist for noticing patterns that actually exist doesn’t make them disappear — it makes “racist” a meaningless term and “journalism” an empty promise.
We can’t solve problems we refuse to measure. And we can’t trust institutions that think we’re too stupid to handle the truth, no matter how uncomfortable it might make us.
The data comes from a Bureau of Justice Statistics report, specifically Table 2 that shows the “Number of violent incidents, by victim and offender race/Hispanic origin, 2017–21.” What’s particularly egregious here is that the total number of black-on-white violent crimes is approximately 2.4 million, while the total number of white-on-black violent crimes is about 370,000 — even before adjusting for population size, black-on-white crimes outnumber the other way around by 6.4x, despite whites outnumbering blacks by 4 to 1 in the population.
The AP cites “2024 census estimates” for population percentages while analyzing crime data from 2017-2021 — a temporal mismatch that any freshman statistics student would catch. Using the actual 2020 census figures from the relevant period (13.54% Black, 59.6% non-Hispanic white) makes the per capita disparity slightly worse, not better. But even giving the AP every benefit of the doubt with their cherry-picked numbers, the math still reveals a 23-fold difference they desperately tried to obscure.
Black offenders were involved in 15% of violence against white victims, while white offenders were involved in 12% of violence against black victims. Using the AP’s numbers of black Americans comprising 13% of the population and whites 56%, the per capita likelihood ratio = (15%/12%) × (56%/13%)² = 1.25 × 18.56 = 23.2. This assumes equal victimization rates across groups, which the BJS data confirms (19.8 per 1,000 for whites, 19.4 per 1,000 for blacks). Using the raw incident counts (2,382,400 Black-on-white vs 371,540 white-on-Black) yields similar results.