A single study makes the news, a friend shares it, and suddenly everyone "knows" that coffee cures something or a common food causes cancer. But research papers are written for other researchers, not for you — dense, hedged, and full of jargon. The good news: you don't need a science degree to read one critically. You need to know which parts to read, and which questions to ask.
Start with the abstract — but don't stop there
The abstract is the paragraph at the top that summarizes the whole paper. It's the fastest way to learn what the study was about and what it claims to have found. But abstracts are also where authors put their most optimistic spin, so treat it as the headline, not the verdict. The real story is in the methods and results.
The four parts of almost every paper
Most studies follow the same skeleton, often called IMRaD:
- Introduction — why they did the study and what question they're trying to answer.
- Methods — who or what they studied, how many, and what they measured. This is the most important section for judging trust.
- Results — what they actually found, usually with numbers.
- Discussion — what the authors think it means. This is interpretation, not fact — read it skeptically.
The five questions that decide how much to trust it
You can size up almost any study by answering these, mostly from the methods section:
- Who was studied? People? Mice? Cells in a dish? A result in mice is a long way from a result in humans.
- How many? A study of 40 people tells you far less than one of 40,000. Small samples produce flukes.
- Compared to what? The strongest studies compare a group that got the treatment against a similar group that didn't (a control group). No comparison group means weak evidence.
- How big was the effect? "Statistically significant" doesn't mean "large." An effect can be real but tiny enough to be meaningless in daily life.
- Was it random? A randomized controlled trial — where people are randomly assigned to groups — is the gold standard. An observational study (just watching what people already do) can find links but can't prove cause.
Read the limitations — the authors already told you the weaknesses
Near the end of the discussion, most honest papers include a limitations paragraph where the authors list what could be wrong with their own study: too few participants, a short time frame, factors they couldn't control for. It's the most useful paragraph in the paper, and the one press coverage always skips.
Two quick trust checks
- Is it peer-reviewed? A preprint hasn't been vetted by other scientists yet. Here's what that distinction means.
- Who paid for it? Industry funding doesn't automatically make a study wrong, but a soda-funded study showing soda is fine deserves an extra raised eyebrow. Check the "funding" and "conflicts of interest" notes.
The one trap to always watch for
A study can find that two things happen together without one causing the other. "People who drink wine live longer" doesn't mean wine extends your life — maybe people who can afford wine also have better healthcare. That's the correlation-versus-causation trap, and it's behind most misleading health headlines.
The bottom line
Reading a study isn't about understanding every statistic — it's about asking: who did they study, how many, compared to what, how big was the effect, and what did the authors admit they couldn't rule out? Answer those, and you're already reading more carefully than most of the headlines about it.
Staring at a paper full of jargon? Plainly is a free Chrome extension that explains a research paper in plain English — what they found and how much to trust it — right on the page. Open the PDF and click "Decode this Document." Your files are never stored.