How to Read a Study Without Being a Scientist
Introduction
"A study showed that…" opens half the training and nutrition content in circulation. And it works, because it sounds authoritative and because almost nobody is going to check.
You don't need scientific training to evaluate whether a claim holds up. You need a handful of questions and the willingness to ask them before changing anything in your routine.
That a study exists doesn't mean it says what you're being told it says.
Not all studies carry the same weight
This comes first, and it organizes nearly everything else. From least to most weight:
Animal or cell studies. Useful for understanding mechanisms, useless for deciding what to do on Tuesday. If the headline ends in "in mice", application to humans is a hypothesis.
Observational studies. They measure associations in populations. Good for generating questions, unable to establish cause. The classic: people who take supplements are healthier — and they also exercise more, smoke less and see doctors more often.
Randomized controlled trials. Here evidence about causation begins: who receives what is assigned at random. It's the format that allows you to say "this produces that".
Systematic reviews and meta-analyses. They gather all available trials. The highest level, with one caveat: a meta-analysis of bad studies is still bad.
Practical rule: a recent meta-analysis outweighs the new study your feed showed you, however spectacular the headline.
The six questions
1. Who was it done on? The most important and the most ignored. A result in elite lifters may not apply to you, and neither may one in untrained beginners: beginners improve with almost anything, so a study in them can barely distinguish between methods.
2. How many people and for how long? Twelve participants over six weeks is a small, short study. Useful as a signal, not as proof. And in training, six weeks is short: much of the early improvement is technical learning, not adaptation.
3. What exactly did they measure? Here's the most frequent trap. Many studies measure an intermediate marker rather than the outcome you care about.
The clearest case: muscle protein synthesis measured hours after training. It's constantly used to sell protocols, and work by Mitchell and colleagues showed that this acute response doesn't predict the hypertrophy the same person achieves after months of training. A protocol can raise the marker and not change the result.
4. How big was the effect? "Statistically significant" doesn't mean "large". It means it probably isn't chance. A method can be significantly superior to another and that superiority can be 1 %, which changes nothing in practice.
5. Compared with what? A protocol that beats doing nothing hasn't shown it beats what you already do. It's the comparison most often omitted from headlines.
6. Who funded it? It doesn't invalidate anything on its own, and it's information. If a supplement study is funded by whoever sells it, waiting for independent replication is reasonable.
Two ideas that save arguments
Absence of evidence isn't evidence of absence. That something isn't demonstrated may mean it doesn't work, or that nobody has studied it properly. Those are different situations and shouldn't be treated alike.
A single study almost never changes anything. Isolated results contradict each other constantly; that's how science normally works, not a failure of it. Ioannidis put it bluntly in a famous paper: a large share of published findings don't hold up when replication is attempted. That's why weight lies in the body of work, not in novelty.
How to check a claim in five minutes
1. Find the original study, not the article commenting on it. The title is usually in the piece. 2. Read the full abstract, not the headline. Often the abstract itself is more cautious than whoever cited it. 3. Look at the method: who, how many, how long, against what. 4. Check whether a meta-analysis exists on the same topic. If one does, it wins. 5. Ask whether it would change your decision. If the answer is no, you're done.
Common mistakes
Citing the summary of the summary. The chain headline → post → outlet → study loses nuance at every step, almost always in the direction of exaggeration.
Confusing mechanism with result. "It makes physiological sense" is a hypothesis, not a finding. Training history is full of elegant mechanisms that produced nothing.
Using science as a weapon rather than a criterion. Hunting for the study that confirms what you already do is the opposite of informing yourself.
Dismissing everything because "science contradicts itself". Total scepticism is as comfortable and as useless as credulity.
Applying a study done in another population to your case. The single most frequent error.
Conclusion
Evaluating evidence doesn't require technical training: it requires asking who it was done on, how long it ran, what was measured, how big the effect was, and what it was compared against.
Try it with the next claim that makes you doubt your routine: before changing anything, find the study and answer those five questions. Most of the time you'll discover the finding is smaller, more specific or more provisional than what reached you — and that your current plan didn't need touching.
References
- Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8).
- Mitchell, C. J., et al. (2014). Acute post-exercise myofibrillar protein synthesis is not correlated with resistance training-induced muscle hypertrophy in young men. PLoS ONE, 9(2).
- Greenhalgh, T. (2019). How to Read a Paper: The Basics of Evidence-Based Medicine (6th ed.). Wiley-Blackwell.
- Schoenfeld, B. J., & Aragon, A. A. (2018). A commentary on the effects of protein supplementation. Journal of the International Society of Sports Nutrition, 15.
- Guyatt, G., et al. (2011). GRADE guidelines: A new series of articles in the Journal of Clinical Epidemiology. Journal of Clinical Epidemiology, 64(4).