Your Opinion Is Not Your Own
J. Jonah Jameson ran a newspaper. Whatever Spider-Man did, the Daily Bugle found the angle that made him a menace. Saved the bus? Reckless endangerment. Stopped the robbery? Vigilante. The man had a conclusion and went shopping for evidence.
He’s a caricature, and everybody laughs, because everybody recognizes him.
This is not new
The idea that news used to be neutral and recently stopped being neutral gets the history backward. Partisan papers were the norm in this country for most of its existence. Party organs, press barons, circulation wars fought with invented atrocities. “Objective journalism” was a mid-twentieth-century professional ideal that held, imperfectly, for a few decades.
What changed isn’t that outlets acquired an agenda. It’s the delivery. It arrives continuously, it’s shaped to what you clicked yesterday, and it’s paid for by how long it holds you. Propaganda used to arrive once a day on the porch, if you could afford it, and sometimes you didn’t even read it.
What isn’t there
In 1968, two researchers interviewed a hundred undecided voters in Chapel Hill and compared what those voters called the important issues against what nine news outlets had emphasized. The two lists matched at .967, where 1.0 would be a perfect match. What people thought mattered was, very nearly, what the press had covered most.
That’s agenda-setting, and the mechanism isn’t persuasion. Nobody told those voters what to conclude. What the coverage did was set the list of things worth having a conclusion about. The correlation alone doesn’t prove which way the causation ran. But the effect is the one you notice once you look for it: a story that isn’t covered isn’t a story you have a wrong opinion about. It’s one you don’t have an opinion about at all.
Two examples with no party attached.
When someone goes missing, coverage doesn’t track who goes missing. A 2016 study compared FBI missing-persons data against news coverage. Black people were about 35% of missing persons and roughly 13% of the subjects of coverage, weighted by intensity. White women were about half of all article subjects.
And people are afraid of flying. In 2023, commercial aviation ran 0.003 deaths per 100 million passenger miles. Cars and light trucks ran 0.53, about 176 times higher. One of those produces a week of coverage when it goes wrong. The other kills 40,000 Americans a year in increments too small to report.
Pick your number
Guns are the cleanest illustration of the next move, so bear with me while I take apart my own side’s favorite number along with everyone else’s.
In 2023 there were 46,728 gun deaths in the United States. Of those, 27,300 were suicides and 17,927 were homicides. Which number you lead with decides what the reader thinks the subject is. Lead with 46,728 and the subject is guns. Lead with 17,927 and the subject is violent crime. I’d argue suicide is a different problem with different causes and belongs in a different conversation, and you should notice that I just made a framing choice in front of you rather than behind your back.
Now the other direction. Defensive gun use is counted in ways that barely resemble each other. The Bureau of Justice Statistics recorded 184,300 incidents of victims threatening or attacking with a firearm during nonfatal violent crime across 2018 through 2022, about 37,000 a year, with property crimes counted separately. The Kleck and Gertz survey put annual defensive uses at 2.2 to 2.5 million. That isn’t a margin of error. Those are different instruments asking different questions of different samples, and the National Research Council called the disagreement a controversy in the field. Each side quotes the end it likes, and neither mentions the other end exists. A defensive use also isn’t the same as a life saved, because no survey can establish what would have happened otherwise.
Both hands
The same move, from each direction. These are claims worth checking rather than measurements of what the coverage did to anyone.
In a survey run in 2020 and reported the following year, 53.5% of respondents classified as very liberal on both economic and social questions estimated that police killed 1,000 or more unarmed Black men in 2019. The report set those answers against database counts of 13 fatal shootings and 27 deaths by any means, while acknowledging the records are incomplete.
Nuclear power causes 0.03 deaths per terawatt-hour generated. Coal causes 24.6, counting accidents across the supply chain and, mostly, air pollution, before you get to the carbon. Nuclear’s waste problem is real and unsolved. It is also visible, contained and counted, which is exactly what coal’s is not.
From the other side: one review found 31 credible cases of in-person voter impersonation between 2000 and 2014, against more than a billion ballots cast. And a study of Texas arrest records from 2012 to 2018 found undocumented immigrants arrested at lower rates than native-born citizens, for violent, drug and property felonies alike.
Then the finding that makes this structural rather than partisan. In 1985, researchers showed 144 students the same 36 minutes of network coverage of the massacre at Sabra and Shatila, where Phalangist militiamen killed hundreds of Palestinian civilians in two Beirut refugee camps. Viewers rated the coverage on a single nine-point scale running from hostile-to-Israel at one end to hostile-to-the-Arabs at the other, with 5 as neutral. Pro-Israeli students put it at 2.9. Pro-Arab students put the same tape at 6.7. Each group landed on the side of the scale meaning this was slanted against us. Asked to count the references, one group remembered 57% of them as unfavorable to Israel; the other remembered 26%.
Same tape. Same room. And the people most certain the coverage was slanted were the ones who knew the most about the subject.
Doing the arithmetic
Here’s where I tried to settle something and couldn’t, which turned out to be the useful part.
The claim I hear constantly, and have made myself, is that the big outlets lean left. Run down the national news organizations and the ratings agree. AllSides puts CNN, ABC, CBS, NBC, NPR, the New York Times news desk and the Washington Post news desk at Lean Left, and the Guardian, the Atlantic, the New Yorker and MSNBC at Left. The Associated Press moved to Left in November 2024 and back to Lean Left in May 2026. Of the outlets at that scale, Reuters is the only one rated Center. Further down the list a few smaller names land at Center as well.
The charts have limits worth saying out loud. AllSides rates political lean only, and only for online written content, which matters for every network named above. Ad Fontes scores reliability alongside bias, so the two aren’t measuring the same thing. Neither is peer-reviewed, neither publishes an overall tally, and critics argue they manufacture a false center and hand readers unearned confidence. The ratings are judgments with uncertainty attached: the Post’s editorial board was recently rated Lean Right by a panel in which four of six members said Center.
A rating alone tells you nothing about influence, though. Lean Left on a site with 30,000 readers isn’t Lean Left on the evening news. So weight it by reach.
Same measure, roughly comparable. The broadcast evening newscasts averaged about 7.6 million, 6.0 million and 4.2 million viewers in the 2024–25 television season. Roughly 18 million people. Fox News, MSNBC and CNN in primetime — 2.65 million, 915,000 and 573,000 — come to about 4.1 million for calendar 2025. Different reporting periods, so treat it as approximate.
Different measure, stated as such. Local TV early-evening news averaged about 3.1 million in Pew’s 2022 figures, but that counts television sets tuned, not people, so it can’t be laid beside the others. Worth naming anyway: it’s a top-tier audience that essentially never gets a bias rating attached to it.
Where people actually get news. 56% of adults often get news from digital devices, 32% from television, 11% radio, 7% print. Facebook 38%, YouTube 35%, Instagram and TikTok 20% each. Those are delivery channels, not sources, and they overlap heavily. Watching the CBS newscast on YouTube counts as both.
And the inversion. Fox leads the three cable networks named above, out-rating the other two together. A 2007 study of 257 news/talk stations owned by the five largest commercial owners found 91% of weekday talk programming conservative, and the attempt at a liberal equivalent, Air America, shut down in 2010. Among news influencers, Pew’s coded sample found 27% right-leaning to 21% left-leaning, counting the influencers rather than their audiences.
So: can I multiply bias by reach and tell you which direction dominates American news exposure?
No. The measures aren’t interchangeable. Nielsen counts average simultaneous viewers, Pew counts percentages of adults, Similarweb counts visits, and Pew’s local figure counts television sets. The audiences overlap in ways none of these capture, and a platform tells you nothing about the politics of what crossed it. These figures cannot produce that number, and anyone who hands you one built from them has stopped measuring and started arguing.
What holds up is narrower, and comes stated separately, because that’s how the evidence comes. The written output of the big national organizations rates left of center. Their broadcasts hold the largest audiences that can be compared with each other. Fox leads the cable networks listed here. Talk radio and local news carry real reach that is measured differently, or not measured for lean at all. The combined ideological exposure of the country remains unmeasured.
Why the skew exists has two answers, and both have evidence. Gentzkow and Shapiro measured slant across 433 daily papers covering three quarters of US circulation and found consumer ideology explained about 20% of the variation, against about 4% for owner ideology. Papers lean the way their readers lean. Then Martin and McCrain studied Sinclair’s 2017 acquisitions, using a dataset covering 743 stations: after the acquisition, those stations ran about 25% more national political coverage, less local political coverage, and content that had moved right by roughly a standard deviation, while viewership barely budged. That one runs from the top down. Sinclair reaches over 40% of US households; Nexstar about 70%.
There’s a supply side too. Journalists identifying as Republican went from 25.7% in 1971 to 3.4% in 2022, with independents now the majority at 51.7%. That survey had an 11% response rate, which is a real caveat, and party identification isn’t the same as story selection.
And the country being covered, per Gallup: in 2024, 37% of Americans called themselves conservative, 34% moderate, 25% liberal. Separately, 45% now identify as political independents, a record, which is party identification rather than ideology.
So part of the answer is that the national outlets aren’t mirroring the country. They’re serving the slice that pays them — subscribers, members, donors — and that slice isn’t the country either. The other part is that ownership moves coverage whether or not the audience asked.
Spot the tells
Since none of this resolves cleanly, the practical question is how to tell a borrowed opinion from a built one. Mine included. The tells:
They can’t put it in their own words. The phrasing arrives intact, the same unusual construction from three different people in a week, and asking for it another way produces the same sentence again.
There’s no floor under it. Ask where the number came from and the answer is that they heard it. Everyone has heard it. Nobody can name who counted it.
They can’t state the other side. Not agree with it. State it, in a form someone who holds it would recognize as their own position.
Disagreement gets explained by motive. Not “here’s why that’s wrong” but “here’s why you’d say that.” Once the argument is about your reasons for speaking, nobody is examining the claim.
We all sort by team, and there’s a measurement of it. In a 2018 Pew test, people sorted statements as factual or opinion, meaning could-this-be-checked rather than is-this-true. Take “Immigrants who are in the U.S. illegally are a very big problem for the country today.” That’s an opinion. 51% of Republicans called it factual, against 19% of Democrats. Now take “Increasing the federal minimum wage to $15 an hour is essential for the health of the U.S. economy.” Also an opinion. 37% of Democrats called it factual, against 17% of Republicans. Same test, same failure, opposite directions.
Why it takes
Nobody audits the foundation of an opinion they don’t know they were handed. Who said it, what exactly did they say, where did the number come from, when was it measured, why this framing rather than another, and how was it counted. Six questions, and they go unasked, because asking them assumes there was a source. A thing that simply arrived doesn’t feel like it came from anywhere. It feels like something you think.
Borrowed vocabulary is the tell inside the tell. Words arrive with conclusions already packed in them, and using the word means accepting the packing.
And before anyone forwards this to the relative it’s obviously about: across 106 studies, people consistently believe media influences other people more than it influences them. The gap is largest for messages they consider harmful. Everyone reading this has someone in mind. That’s the finding, working on you, right now.
Which is the point of the title, and worth saying plainly: an opinion can be your own. It becomes yours when you’ve gone and looked, when you can say where the number came from and what’s wrong with it, when you can state the opposing case without sneering. That’s why the arithmetic above is on the page instead of asserted. You can check it, and you can tell me where it fails.
The cornucopia
The Fruit of the Loom logo has an apple, grapes, currants and leaves. A great many people also remember a cornucopia behind the fruit, a brown wicker horn. The company says flatly there has never been one.
That’s the part worth sitting with. Certainty that solid doesn’t require anyone to have lied to you.
Jameson didn’t need to invent the photograph either. He only needed the right caption.
I’m not sure how I got transported into this reality. But where I come from, it was “Spiderman,” and there was definitely a cornucopia.
Discover more from Perhapsery
Subscribe to get the latest posts sent to your email.
