How to read a research paper: a guide to getting started

1. Choose the question that brings you to the paper

Write one line about what you want to find out. You might be interested in what researchers investigated about a habit, how they measured a behavior or what happened in a comparison. That question gives you a way to decide which parts to explore first.

In their rules for reading a scientific paper, Carey, Steiner and Petri recommend adapting your reading to your goal. Our practical starting point is to put your question at the top of your notes, so you can return to it when the text leads in several directions.

2. Identify the kind of article you are reading

Original research presents data and analysis; a review brings together earlier work. There are also commentaries and articles about methods. The kind of text changes what it makes sense to look for: do not expect every review to recruit new participants or an opinion piece to report an experiment.

Find the article type, title, date and source before taking notes. If it is a review, record which studies it draws on. The NCCIH guide to research methods helps distinguish several approaches. The route below focuses on an article reporting research results.

3. Use the abstract as a map

An abstract generally summarizes the question, approach, results and conclusion. Read it to get oriented and note the claim you want to check in the rest of the text. The NCCIH explanation of article sections helps you recognize that map.

Then look through the headings in the full text. Locate the methods, results and discussion. You do not need a fixed reading order, but you do need to know which section each idea in your notes comes from. If you could only access the abstract, record that: your reading does not yet include the details of the study.

4. Return to the methods with specific questions

The methods explain how the study was done. Look for who or what was studied, how they were selected, what was compared and what was measured. An observed measurement is different from a questionnaire response: record how the information was obtained.

You can leave technical terms as open questions while locating those pieces. If the conclusion you care about depends on a technique you do not understand, mark that point for further reading or help before repeating it as an established fact.

• Who or what was studied?
• Was there a comparison, and how were the groups organized?
• Which variable was measured, with what instrument and over what period?
• Which part of the procedure can I not yet explain?

5. Separate findings from interpretation

In the results, look for the data that answer your question. In the discussion, look for how the authors explain those data and which other explanations they consider. Keeping those in separate lines helps you avoid turning an interpretation into a measurement.

The NCCIH guide to results notes that a statistically significant result may have little practical importance and that a single study rarely settles a question. Ask how much the measured outcome changed, what uncertainty is reported and whether the design supports talking about a cause or an association.

6. Read a figure together with its caption

Choose the figure or table that accompanies the central result. Read its caption, axis or column names, units and group labels. Return to the paragraph that refers to it. This is a useful route for checking which data a sentence describes.

If there are bars, colors or symbols you cannot interpret, record the question rather than assigning them a meaning. The reading rules from Carey and colleagues also emphasize working through figures and tables; they are more than illustrations to skip over.

7. Finish with a short reading note, rather than a headline

We suggest this note format to organize your reading. It is not a quality scale or a validated method for scoring studies. It is a way to keep both what you can explain and what remains open in view.

• Question: what the researchers wanted to find out.
• Design and sample: how they studied it, with whom or using which data.
• Result: what they observed or measured.
• Interpretation: the explanation the authors propose.
• Limit: what this work cannot answer.
• Open question: a term, method or detail I need to revisit.
• Source: the article link and the date I read it.

If you lack the full text or get stuck

In biomedicine, a PubMed record does not always include the whole article. The official PubMed help explains how to follow publisher or PubMed Central links when available. You can also check access through your library.

For an unfamiliar word, look for its meaning in the context of the method and return to the original sentence. If you use an automated explanation, compare it with the cited passage. A clear-sounding answer does not replace checking the source. If a question changes the meaning of the conclusion, you can leave your note open.

Try the route with a familiar question

You can start with our reading Two studies, two timelines: 59 and 66 days for a habit to become automatic. Use it as an entry point and then open its original study. Try completing the note: what question was asked, what kind of work it is, where the result comes from and which limit you would need to keep when explaining it.

PaperFlow publishes short readings with access to their sources to support that discovery. It does not carry out a specialist appraisal of a study on your behalf. You can explore the open reading library or return to an unanswered question in your note; your first pass does not need to become a final judgment.