The short answer
Start with a question you need the paper to answer. Preview its argument and structure, then read the evidence behind the relevant claim carefully. Leave with a short note recording the claim, evidence, limitation and page—not just a highlighted PDF.
- Choose the depth of reading for the task: screening a source, preparing a seminar and critiquing a method are different jobs.
- An abstract helps you decide where to look; it is not enough to support every claim you might cite.
- Separate what the authors observed from what they infer or recommend.
- Use AI to clarify a passage only after locating it, and verify the response in the original.
Explore this guide6 sections · jump to what you need
Decide what you need from this paper
Before reading, finish the sentence: “I need to know whether…” For an essay, that might be whether a study supports one explanation over another. For a seminar, it might be the author’s central argument and one objection. For a methods assignment, you may need to understand the design in much greater depth.
Duke’s reading guidance recommends previewing scientific articles and then returning to the sections your task requires. You do not need to understand every unfamiliar term before finding out whether the paper is relevant.
If the paper was assigned for detailed reading, do not use screening as a reason to skip it. The purpose is to choose your route through the text, not to pretend that an abstract replaces the required work.
Find the claim and the evidence behind it
Empirical papers often separate methods, results and discussion. Other disciplines organise arguments differently. George Mason University’s article guide explains these differences; a humanities article may build its interpretation through close reading rather than a labelled results section.
| You need to find | Look for | Question to ask |
|---|---|---|
| The problem | Introduction or opening argument | What question is unresolved? |
| The approach | Methods or explanation of the evidence | How could this approach answer that question? |
| The support | Results, figures, quotations or close analysis | What does the evidence actually show? |
| The interpretation | Discussion or developed argument | What extra assumptions connect evidence to conclusion? |
| The boundary | Limitations and the evidence itself | Where would this conclusion stop applying? |
Read back and forth when necessary. If the discussion says a change was “substantial”, inspect the result behind that word. If a figure makes no sense, read its caption and the method that produced it before trying to paraphrase it.
Worked example: write a claim the evidence can support
Imagine a fictional survey of 200 students at one university. Students who reported using a revision app also reported higher marks. The authors suggest the app may support studying.
A weak reading note says: “Revision apps improve grades.” It loses the study design, the sample and the uncertainty. A better note is:
In this fictional single-university survey, reported app use was associated with reported marks. The design does not establish whether the app caused better performance; prior attainment or study habits could help explain the relationship. Useful as an example of an association, not proof of an intervention’s effect.
In a real paper, add the exact page or table and check how the variables were measured. Was app use a yes/no question? Were marks verified? Were other factors considered? You do not have to find a fatal flaw. You have to match the strength of your claim to the strength of the evidence.
For an interpretive article, the parallel task is to inspect the passage or example that carries the argument. Is the interpretation consistent with the text? What competing interpretation does the author address—or leave open?
Use a reading note you can return to
| Field | What to record |
|---|---|
| Reference | Author, year, title and a stable link or identifier |
| My question | Why I opened this source |
| Main claim | One sentence in my own words |
| Evidence | The relevant result, passage or example, with page/table |
| Limit | One specific restriction on the conclusion |
| Use in my work | Support, counterargument, method or background |
| Unresolved | What I need to clarify before relying on it |
Mark direct quotations clearly and keep them separate from paraphrases. A note that merely copies the abstract is hard to use later: it records what the paper is about, but not why it belongs in your argument.
A reference manager such as Zotero can store the record and note. See our student research websites guide for finding and organising sources.
What to do when the methods are beyond you
Identify the minimum explanation needed for your task: what was compared, what was measured, and what would count as evidence for the claim. Look up a key term in your course text or a suitable reference. Keep a list of unresolved assumptions rather than silently treating them as understood.
If your assignment requires evaluating the method itself, you may need a tutorial or help from the teaching team. A polished summary cannot stand in for that understanding. Ask a specific question, such as “Why is this comparison appropriate given how the sample was selected?”
Allow longer for foundational papers and unfamiliar methods. There is no responsible promise that every research paper can be read properly in ten minutes.
Use AI to investigate a passage, not to invent a reading
When permitted, supply the relevant passage and ask the tool to separate the authors’ claim from its own explanation. Ask it to identify where the passage does not answer your question. Then inspect the source yourself.
Explain this passage at undergraduate level. Identify the claim, the evidence mentioned, and any limitation explicitly stated. Point to the wording supporting each item. Keep your own background explanation separate. If the passage is insufficient, tell me what section to read next.
If the answer cites a page, check it; page numbers and quotations can be wrong. Our NotebookLM/Gemini Notebook vs ChatGPT comparison explains source-based workflows. Neither tool makes it safe to cite a paper you have not checked.
Frequently asked questions
Should I read a research paper from start to finish?
Not always on the first pass. Preview it, identify the relevant claim, then investigate the evidence and context. Read in greater depth when the assignment or your use of the source requires it.
Is reading the abstract enough?
It may be enough to decide whether to read further. It is usually insufficient for evaluating the methods, checking a specific claim or writing a detailed critique.
How long should it take to read a paper?
It depends on your purpose, background knowledge and the paper’s complexity. Screening relevance is different from understanding a new method or preparing a critique.
How do I critique a paper without being an expert?
Start with the question, evidence and limits: what was studied, how the evidence supports the conclusion, and where it might not apply. Record technical uncertainties and ask for help rather than inventing a criticism.
