Reading the research
Most of the time you shouldn't have to do this — we read the studies so you don't have to, and we leave the trail on every one. But sometimes a post lands in your group that starts with "studies show," and you want to know for yourself whether it holds up. Here's how to check. These are the same questions we ask of every source on this site, and not one needs a science degree.
One note before we start: this page teaches you how to weigh evidence. It doesn't tell you what to believe about any single claim. Where you land stays yours. We're just handing you the questions.
A real study was written by named researchers, published in a journal, and checked by other scientists in the field first — that check is called peer review. It's a floor, not a promise. A blog post, a podcast clip, or a product page has no floor at all. Quick test: can you click through to the actual study? Look for a journal name, authors, a date, and a link to somewhere like PubMed. A bold claim travelling with nothing you can click is your first warning.
One study is a starting point, not a fact. Findings get stronger when separate teams keep landing in the same place. The strongest evidence is never a single dramatic paper — it's a stack of them pointing the same way. Some studies exist only to gather all the research on a question and weigh it together; they're called systematic reviews or meta-analyses. If you can find a recent one, start there.
A finding is only as good as the people it came from. Forty people in one survey is a very different thing from a hundred thousand. And it matters who they were — a result from another country, or only in adults, or only in mice, may not reach your own kitchen table. None of that makes a study worthless; it tells you how far the finding can travel.
Read what the researchers actually looked at, not what the headline says. The gap is often wide. The biggest trap is one small phrase: linked to. When a study says two things are linked, or associated, or correlated, it means they showed up together — not that one caused the other. Ice cream sales and drownings both climb every summer; the heat drives both. Proving cause is much harder, and far fewer studies do it.
Two questions hide inside every result. Statistically significant sounds impressive, but it only means the finding probably isn't a fluke — it says nothing about size. A change can be real and still be tiny, technically detectable and useless in daily life. Look past whether something changed and ask how much. A finding that's real but small doesn't earn a headline, even though it often gets one.
The pattern is simple: reaching for the most dramatic number instead of the most honest one. A study reports a range and the post quotes only the scary end. A risk is said to "double" when the thing it doubled was already almost nothing. One study that happens to fit gets shared a thousand times while ten that disagree sit in silence. When a number feels built to alarm you, look for the plain version underneath.
Money doesn't always bend a result, but it always earns a second look. A study of a supplement, paid for by the company that sells it, isn't automatically wrong — it's one you read harder. Researchers are usually required to say who funded the work, under a heading like funding, disclosures, or conflicts of interest, often near the end. When it's missing, that absence is information too.
Not every journal is what it looks like. Some will publish almost anything for a fee, with little real review, and a flimsy study can be dressed up to look serious. Two quick checks: is the study listed on PubMed, the main public index of medical research? And has it been retracted — pulled back after publishing? A thirty-second search tells you which one you're dealing with.
This is the tell that separates careful work from a sales pitch. Good researchers state their own weak spots out loud — the sample was small, the study couldn't prove cause, the result needs repeating. That honesty is a good sign. Be more suspicious of the study that promises certainty, not less. Anything offering a breakthrough, a cure, or one simple answer has usually skipped the part where it tells you what it doesn't know.