Just Because There's a Study, Doesn't Mean It's True

"Studies show..." is one of the most trusted phrases in the English language, and one of the least reliable. It sounds like the end of an argument. In reality, it's often just the beginning of a different question: which studies, funded by whom, designed how, and reviewed by who?

This isn't a theory. It's documented, repeated, and measurable — and once you see how it works, you can't unsee it.

1. Who Paid For It

In the 1960s, growing evidence was starting to link sugar to heart disease. The Sugar Research Foundation — an industry trade group — needed that concern to go away. So it paid three Harvard scientists the equivalent of roughly $49,000 in today's money to write a literature review. The review hand-picked which studies to scrutinize, tore apart the ones implicating sugar, and concluded that fat and cholesterol were the real problem. It was published in the New England Journal of Medicine in 1967 — one of the most respected medical journals in the world — with no disclosure of who had paid for it, because journals didn't require that disclosure until 1984.

That single review helped redirect decades of public health guidance, government dietary advice, and food industry marketing away from sugar and onto fat — a shift some researchers now believe contributed directly to the obesity epidemic that followed.

Same substance. Same science. Wildly different conclusions, depending entirely on who was cutting the check.

2. Is It Observational, or Is It Controlled?

Not every flawed study involves corruption. Sometimes the study design itself is the problem, and it takes an honest scientist to catch it.

For decades, observational studies — the kind that simply track large groups of people and note what happens to them — suggested that hormone replacement therapy reduced a woman's risk of heart disease by 40 to 50%. It looked like solid, consistent, repeated evidence. Then a large randomized controlled trial, the Women's Health Initiative, tested it properly: one group given hormone therapy, one group given a placebo, assigned at random. The result was the opposite of what decades of observational data had suggested — the therapy modestly increased heart disease risk, rather than lowering it.

What went wrong wasn't fraud. It was the difference between watching what happens to people who happen to choose something, versus actually testing it. Women who chose hormone therapy on their own were, on average, healthier, wealthier, and more health-conscious to begin with — and that difference, not the hormones, was what the observational studies were really picking up.

3. Peer Review Isn't a Guarantee

People often assume "peer-reviewed" means "verified true." What it actually means is that a small number of other researchers — often two or three — read the study before publication and didn't object enough to block it. Peer review checks whether the methodology is described clearly and whether the conclusions loosely follow from the data presented. It does not re-run the experiment. It does not audit the raw data. It does not require the authors to disclose who paid for the work, unless the journal specifically demands it — and as the sugar industry example shows, for decades many of the most respected journals in the world simply didn't ask.

Peer review is a filter, not a guarantee. It catches some bad science. It has let plenty of bad science through, especially when the flaw is in who funded it or how the question was framed rather than in the math.

4. Read It Yourself

The single most protective habit is also the simplest: don't stop at the headline, and don't even stop at the abstract. Headlines are written by journalists or marketers, not researchers, and they're optimized to be shared, not to be accurate. Abstracts are written by the same researchers who ran the study, and they tend to describe the results in the most favorable possible light.

The actual data is often in a table, several pages in, and it frequently tells a smaller, messier, more uncertain story than the headline promised. A study might say "linked to" when it means a weak correlation in a small sample. It might say "may reduce risk" when the actual effect size was too small to matter in real life. None of that shows up until you look.

None of this means science can't be trusted. It means a single study, a single headline, or a single "expert" citing "a study" is never the end of the conversation — it's the start of a few honest questions. The people selling you the next diet, supplement, or fitness trend are counting on you not to ask them.