Born in a Press Release, Raised by Aggregators: The Strange Life Cycle of a Viral Study

A Study Is Born at 6 A.M. Eastern

Every weekday morning, a small army of press releases lands in the inboxes of American journalists. Most die quietly. But every so often, one of them contains the magic ingredients: a number, a threat, and a quote from someone with 'Dr.' in front of their name.

Let's invent one, because the real ones are litigation-adjacent. Say a wellness company surveys 340 of its own customers and finds that 71% report feeling 'more focused' after using its product. The press release goes out under a banner reading 'NEW RESEARCH.' Note what's already happened before a single journalist typed a word: the sample is the company's own customer base, the comparison group is nonexistent, and 'feeling focused' is doing the heavy lifting that a cognitive test should be doing.

None of this is hidden. It's all in the fine print. It's just that nobody reads the fine print at 6 A.M. on deadline.

The Assembly Line of Identical Headlines

Here's where it gets interesting from a data perspective. Within about 48 hours, that single survey generates a remarkably consistent cluster of headlines across the country:

• 'New Study Finds Most Americans Feel Sharper After Morning Coffee Substitute'
• 'Experts Say This Trend Could Change How You Work'
• '7 in 10 Americans Report Better Focus — Here's Why'

Different outlets, different writers, one source. This is what researchers call citation laundering: a claim enters the ecosystem through a press release, gets repeated by enough outlets that it acquires the texture of consensus, and suddenly 'a new study finds' has been upgraded to 'experts say' — a phrase that implies a scientific community weighed the evidence, when in fact 340 people who already bought the product were asked a leading question.

The aggregator dynamics are the real engine. Once two or three mid-size outlets run the story, editors at other outlets face a simple calculus: every competitor is covering it, the story comes pre-packaged, and skipping it looks like a miss. Nobody's verifying anything because verification was never the assignment. The assignment was speed.

Why 'A New Study Finds' Is Journalism's Least Reliable Phrase

The phrase survives because it's technically true. A study did exist. It found something. What it doesn't tell you is the only thing that matters: how big was the sample, who was in it, and compared to what?

Run the numbers on our fictional survey. Three hundred forty self-selected customers is a sample of convenience — the researchers didn't select people, the product did. That's not a poll of Americans; it's a poll of people who already voted with their wallets. And without a control group, 71% feeling 'more focused' tells you nothing, because in uncontrolled surveys of anything people like, positive responses routinely run 60-80%. Ask people if they feel more focused after adopting any habit they chose, and the answer is yes. The baseline does all the work.

The deep irony, given our site's name: the entire problem is that n equals many sounded convincing. '7 in 10' feels like data. It has the shape of data. But 7 in 10 of a biased 340 is worth less than 7 in 10 of nothing, because at least nothing doesn't come with a marketing budget.

The Checkable Questions

The good news is that this particular failure mode is remarkably easy to catch. Before sharing any 'new study finds' story, ask five things:

1. Who ran the study? If the answer is a company selling the thing being studied, close the tab. You're reading an ad wearing a lab coat.
2. How many people? Under a few hundred, treat every percentage as an anecdote with decimal points.
3. Who's in the sample? Customers, volunteers, and online panel respondents are not 'Americans.' They're a very specific slice of Americans.
4. Compared to what? No control group means no finding. Full stop.
5. Can I find the actual study? If every one of those forty headlines links back to a press release rather than a published paper, that's not a study — that's a press release with a publicity team.

Notice that none of these require statistical training. They're the questions a decent editor would have asked in 1995, back when newsrooms had the time.

Forty Outlets, One Source, Zero Skeptics

The uncomfortable truth is that the system isn't broken. It's working exactly as incentivized. Press releases are free content, aggregators reward volume, and 'a new study finds' has never once been false advertising as long as some study, somewhere, found something.

But readers have one advantage the assembly line doesn't: we're not on deadline. The next time forty headlines tell you what 'experts say,' take five minutes and trace the claim back to its source. Very often you'll find the whole towering stack of consensus rests on a sample you could fit in a mid-size lecture hall — one that was assembled by whoever had something to sell.

The plural of anecdote is not data. And the plural of press release, it turns out, is not evidence either.