How to read a scientific paper without understanding all of it

The first scientific paper you try to read will be miserable. The sentences are long, the words are unfamiliar, and the whole thing seems designed to keep you out.

Some of that is genuinely necessary precision. Some of it is habit. Either way, here is the thing nobody tells you early enough:

Nobody reads these front to back. Not researchers, not professors, not the people who wrote them. There is a standard order for reading a paper, and it is not the order it is printed in.

The order to actually use

1. Title and abstract. The abstract is a summary of the whole paper in a paragraph. You are deciding one thing: is this worth my next twenty minutes? Most of the time it is not, and putting it down is a skill, not a failure.

2. Figures. Skip to the pictures. In most papers, the figures are the result — the text is explaining them. Read the caption under each one, which is usually written to stand alone, then look at the graph and ask what it shows. You can often get the main finding from figures alone.

3. The last paragraph of the introduction. This is reliably where the authors say what they set out to do. The rest of the introduction is background you may not need yet.

4. Methods. Here is the one that separates people who read papers from people who skim them. The methods are where you find out whether to believe any of it. How many subjects? Was there a control group? Who was excluded, and why? For how long?

Methods are boring on purpose and they are where the problems live.

5. Limitations. Usually near the end of the discussion. Good papers state plainly what their study cannot show. Read this before the conclusions, because it tells you how much weight the conclusions can carry.

6. Conclusions, last. By now you have your own view. Read theirs and see if you agree. Sometimes you will not, and that is allowed.

Three questions that work on any paper

Whatever the field, whatever the jargon:

How many? Twelve people and twelve thousand people are completely different kinds of evidence.

Compared to what? Issue 3 covered this. If there is no comparison group, be very careful.

What would change their mind? A paper that could not possibly have come out any other way is not reporting a result. It is reporting an opinion with references.

On not understanding things

You will hit sentences you cannot parse. You will hit statistical methods you have never heard of. That happens to everybody, permanently, including experts reading outside their own narrow area.

The move is to keep going. Mark it, read on, and see whether the paper still makes sense without it. Usually it does. If the whole argument turns on the part you could not follow, that is when you go and look it up — and now you have a reason to, which makes it stick.

Waiting until you understand everything before reading anything is how people never start.

Try this

1. Go to PubMed: https://pubmed.ncbi.nlm.nih.gov
2. Search for a condition you care about, and filter to Free full text.
3. Pick one. Read it in the order above — abstract, figures, methods, limitations.
4. Then write three sentences: what they did, what they found, and one thing the study cannot tell you.

Those three sentences are a real summary. If you can produce them, you have read the paper, whether or not you followed every line.

One question to take away

Why would a scientist write a limitations section at all, when it only makes their own work sound weaker?

The answer to that is most of what separates science from advertising.