How to read a research paper with the question still in view

A useful research note lets you explain how a paper reached its conclusion. You should be able to find the original source, name the experimental setting, describe the comparison and point to the measurement behind the claim. That is a more durable reading skill than remembering a striking sentence from an abstract.

The method below is intended for reading published literature. It is not an experimental protocol or a guide to using research materials. Keep in-vitro findings, animal-model findings and human research clearly identified throughout your notes. A paper's setting is part of its finding.

First, define the question you brought

Write one sentence before collecting references: “I want to understand what this study measured about…” Finish it with a specific subject. The question might concern a reported cell response, the interpretation of an analytical result, or the way a comparison was designed.

Give yourself a second line for what would count as an answer. For example, if your question concerns a particular measurement, you need its definition and results. A broad discussion of a pathway may provide useful background while leaving your question open. Recording that distinction helps prevent an attractive introduction from becoming your evidence by default.

NIH treats rigor as a property of the research process, spanning design, methods, analysis, interpretation and reporting. Use that wider frame while reading: a conclusion sits at the end of a chain that you can inspect. NIH, Rigor and Transparency.

Confirm the source and read beyond the abstract

Save the title, authors, journal, publication date and persistent identifier, such as a DOI or PMID. Open the full article and its supplementary files where available. Record whether the source is an original study, a review or a preprint; these labels help explain what kind of material you are reading.

PubMed provides publication-type information and routes to available full text. Its records can also link corrections, retractions and expressions of concern. Look for those updates before taking a finding into your own work, and record the version you actually read. Database filters help navigation, but some rely on indexing that may be incomplete for newer records. NLM, PubMed User Guide.

For your notes, use a simple availability label: “Full text and supplement read,” “Full text only,” or “Abstract only.” If a key detail is behind an unavailable file, mark it as unresolved. This prevents a later reader, including your future self, from assuming you checked material you could not access.

Build a map before judging the conclusion

On a first pass, locate the question, methods, figures, results and limitations. You can then return to the sections that carry the paper's main argument. The following map is an editorial reading aid, not a scoring system:

Leave space between observation and interpretation. Copy a figure or table location into your note so the observation can be checked without rereading the whole article. If an interpretation depends on several experiments, record those connections individually instead of treating the entire paper as one undivided result.

Keep the experimental setting explicit

Identify exactly what the authors studied. A cell-line result should keep its cell-line description. An animal study should retain its species and model. A human study should retain the population and design described by its authors. Avoid allowing a general phrase such as “research shows” to erase these details when you summarize the paper.

For animal research, the ARRIVE 2.0 guidelines provide a reporting framework. Their Essential 10 identify information readers need to assess reliability. Use that framework to locate relevant reporting; it does not turn the presence of a checklist into a guarantee about the conclusions. ARRIVE 2.0 overview.

A practical habit is to begin every evidence note with its setting: “In the reported cell model…” or “In this animal study…” If your next sentence reaches beyond that setting, pause and identify which additional source supports the extension. This guide makes no claim that a particular preclinical observation establishes a personal health outcome.

Ask what the sample number actually counts

A number beside a figure needs a definition. Find the experimental unit, then check the number of units allocated and analyzed. Repeated measurements and independent units answer different questions about the amount of information available.

ARRIVE's sample-size guidance asks authors to specify experimental units per group and distinguishes starting group sizes from the numbers ultimately analyzed. Differences can reflect exclusions or attrition that deserve an explanation. When reading, copy the author's definition of the unit beside each important value of n. ARRIVE, sample size.

In your own table, give “unit” and “number” separate cells. If a figure legend and methods section appear inconsistent, write down the discrepancy and its locations. Avoid resolving it by choosing whichever number makes the finding look stronger.

Examine the comparison and safeguards against bias

Identify the comparison that supports the result: which groups, what control condition and what measurement process. Then look for reporting about randomization, masking, data exclusions and repeated experiments.

NIH's preclinical reporting principles call for these details, alongside clear statistical reporting and a distinction between biological data points and technical replicates. They also encourage sufficiently complete methods and access to underlying data and materials. NIH, Reporting Preclinical Research.

Keep three separate labels in your notes: reported, unclear, and not located. The last label describes your reading, not proof that the researchers omitted a procedure. You may need to check a supplement or ask for clarification. Conversely, a statement that a safeguard was used still needs enough detail to understand where it applied.

Read the effect and uncertainty together

Look at the reported magnitude, units, direction and uncertainty before reducing the result to a significance label. Read the figure legend carefully: identify what its error bars describe and which analysis supports the comparison. Record a value in the form actually reported rather than adding precision or converting an interval into a single confident number.

The American Statistical Association explains that a p-value does not measure effect size, scientific importance or the probability that the hypothesis is true. It cautions against basing a conclusion solely on whether a threshold was crossed. These principles support reading the analysis in its full context. ASA statement on statistical significance and p-values.

A useful note therefore has room for both “what was observed” and “how precisely it was estimated.” If the uncertainty information you need is absent or unclear, say so. Statistical notation should help you describe the evidence more accurately, rather than becoming a substitute for understanding the measurement.

Separate planned questions from exploratory observations

Look for the study's stated primary question and outcome, then identify additional analyses. A paper can contain several interesting observations whose evidential roles differ. Record whether the authors present a result as planned hypothesis testing or as exploratory work.

ARRIVE's interpretation guidance emphasizes the original objectives and places findings alongside other relevant literature. It also cautions about treating exploratory additional outcomes as conclusive. Use that distinction when deciding how strongly to phrase your summary. ARRIVE, interpretation and scientific implications.

One helpful sentence pattern is: “The authors observed [specific measurement] under [reported conditions]; they interpret it as [their explanation], while [named limitation] remains.” Fill each part from the paper. When a part is missing, leave it missing rather than supplying a plausible story.

Finish with a note someone else can follow

Close the paper and try to write a short account using these prompts:

1. Which source and version did I read?
2. What question did the study address?
3. What setting, comparison and experimental unit were reported?
4. Which result answers that question, and where can it be found?
5. What uncertainty or limitation changes the interpretation?
6. What remains open, and which source would help next?

Return to the paper wherever your account becomes vague. This second pass often reveals exactly which definition, figure legend or supplementary detail deserves attention. Keep the final note short enough to revisit and detailed enough that its claims remain traceable.

For material documentation, continue with Reading a COA, field by field. For subject-specific reading paths, use the research library. Research publications, analytical records and product listings each need their own evidence and their own scope.