01Turn the abstract into a question to investigate

Write down the main claim in plain language, preserving the model and outcome. A phrase such as improved cellular response should prompt the questions: which cells, which response and compared with what?

ARRIVE’s animal-research checklist asks for an accurate abstract summary and separately specifies the fuller design, outcome, analysis and result information needed in the manuscript. Its structure illustrates why a summary cannot substitute for the details supporting it.ARRIVE guidelines 2.0 — Complete English checklist (opens in a new tab)

This is a reading route, not a requirement to read every paper in the same order. When a figure is central to the claim, starting with its caption and tracing backwards to the relevant methods can be especially useful.

02Map the claim to its actual experiment

In an original hypothetical abstract, a peptide is described as increasing cellular resilience. The full result turns out to be a change in one fluorescent readout after a defined exposure. The evidence directly concerns that readout in that model.

The next task is to locate any additional observations connecting the readout to resilience. Did the study measure survival or function independently, or does the broader wording rely entirely on interpretation of the one signal?

Record the independent units, control condition, time point and numerical result for the relevant figure. If several experiments appear in one panel, keep their roles separate rather than combining them into one undifferentiated claim.

03Read the caption and supporting material together

A caption often defines n, error bars, normalisation and statistical comparisons. The methods may explain the material identity and measurement process, while a supplementary file contains the full data or excluded observations.

Check that these pieces describe the same experiment. A method used elsewhere in the paper is not automatically the method behind the headline figure, especially when several models or molecular forms were studied.

An article may also cite a previous method rather than describe it fully. Follow that citation when the missing detail affects interpretation, and record any modification the authors made. A citation to a procedure is not proof that every original condition was retained.

If supporting files cannot be accessed, state the resulting limitation. You can still describe the available evidence, but should not imply that raw measurements or full methods were inspected when only the summary was available.

04Write a bounded conclusion from the full record

A useful evidence note separates three statements: what was tested, what was observed and what the observation can support. For the hypothetical example, that means describing the fluorescent response first and then explaining whether independent functional evidence was provided.

Include limitations that change the conclusion, such as an uncertain material identity, a small number of independent units or a readout that does not directly measure the claimed function. Avoid a generic list of every possible weakness.

Do not assume that the absence of a detail in the abstract means it was absent from the study. Equally, do not assume a missing detail exists somewhere in the full paper. Mark it as found, unclear or not reported after checking the relevant sections.

The resulting summary may be narrower than the abstract, but it will be easier to verify. It also makes later comparison with other papers possible because the experimental claim has been defined precisely.