A final-year student is handed fifteen papers for a literature review. She starts with the first, reads every line from the abstract onwards, looks up every term, gets stuck in the statistical methods, and reaches the discussion three hours later. Fourteen papers are still waiting. The problem is not her English or her intelligence. She is reading a paper like a textbook chapter, and research papers are not written to be read in order.
Experienced researchers rarely read a paper start to finish on the first go. They read in passes, each with its own purpose, and most papers never get past the first pass. Here is how that works.
The structure, and the question each part answers
Most empirical papers follow the IMRaD structure. Knowing which question each section answers lets you jump to the right place:
| Section | Answers | Read when |
|---|---|---|
| Abstract | What was done and found | Pass 1 |
| Introduction | Why it matters, what gap exists | Pass 1: last paragraph; pass 2: all |
| Methods | How, on whom, measuring what | Pass 2 skim, pass 3 in depth |
| Results | What they saw | Pass 1: figures and tables; pass 2: text |
| Discussion | Meaning, limitations, comparison with other work | Pass 1: first paragraph and limitations |
The last paragraph of the introduction usually states the aim or hypothesis. The first paragraph of the discussion usually restates the main finding. Those two paragraphs plus the abstract give you the paper's skeleton in ten minutes.
Pass one: ten minutes to decide whether to continue
- Read the title and abstract.
- Read the final paragraph of the introduction: what is the research question?
- Look at every figure and table and read their captions. Figures are where the results actually live.
- Read the opening paragraph of the discussion and the paragraph on limitations.
- Write one sentence: “This paper shows … by … in …”.
If you cannot write that sentence, or the paper is not relevant to your question, stop. Most papers in a literature search stop here, and that is perfectly normal.
Read the figures before the text
The results text often interprets figures the way the authors would like them seen. Looking at the figures first lets you form your own view before comparing. For each figure, ask:
- What are the axes and units? Does the y-axis start at zero, or has it been truncated so differences look larger?
- What do the error bars show? Standard deviation, standard error or a confidence interval? They give very different impressions of certainty.
- What is the sample size per group? It is usually in the caption.
- What is the comparison? Without a control or comparison group, claims about effects are shaky.
Pass two: understand what the paper says
For papers that survive pass one and bear directly on your work, read the whole thing, still without needing every technical detail. The goal is the argument: the problem, the approach, the main results and the limitations the authors admit. Mark references that come up repeatedly; they are usually foundational papers you should read too.
Look up unfamiliar terms only when they block the main argument. Looking up every one is the surest way to spend three hours on a single paper.
Pass three: read to critique or to reuse
Pass three is for a small number of papers: those you will rely on heavily, those you are presenting at journal club, or those whose method you want to adopt. Now read the methods closely and ask:
- How was the sample selected? Who was excluded and why?
- What instrument measured the key variables? Has it been validated?
- Does the design support causal conclusions, or only associations?
- Do any analyses look as though they were added after seeing the data?
- Could you replicate the study from the information given?
A one-page note template for every paper
Scattered highlights in PDFs are hard to reuse when you write. Use a fixed template instead:
- Full citation, saved straight into Zotero or your reference manager.
- Research question, one line.
- Design and sample: study type, size, setting.
- Main finding, in your own words, with the single most important number.
- Limitations: those the authors state and those you notice.
- Link to your project: does this support, contradict or extend your argument?
The last field matters most when you write. A strong literature review is built from connections between papers, not from summaries placed end to end.
Reading in a second language
- Papers in one field reuse a fairly small vocabulary. After ten papers, your speed improves noticeably. Keep a personal glossary for your field.
- Machine translation is useful for checking a tricky sentence, but translating a whole paper tends to distort technical terms.
- Read a review article on the topic before diving into primary research. Reviews give you both the map and the vocabulary.
Common mistakes
- Trusting the abstract completely. Abstracts are written to make the paper appealing; the real numbers and caveats are in the body.
- Skipping the limitations. This is where authors are most candid.
- Reading to understand everything instead of to answer your question. Always know what you are reading this paper for.
With your current reading list, do pass one on every paper before reading any in depth. By the end of an afternoon you will have a list of one-sentence summaries, a shortlist for pass two, and a sense of how the papers connect, which reading them one by one never gives you.
Câu hỏi thường gặp
In what order should you read a scientific paper?
On the first pass, read the abstract, the end of the introduction, the figures and tables, and the start of the discussion plus its limitations. Only relevant papers get a full read on the second pass.
How long does it take to read a research paper?
A first pass takes about ten minutes and is enough to decide whether to continue. A full, careful read can take one to several hours, and only a few papers need that depth.
Why read the figures first?
Figures and tables hold the actual results, while the text often frames them as the authors prefer. Looking at figures first lets you form your own judgement before reading the interpretation.
How do I read papers when English is not my first language?
Start with review articles, keep a personal glossary for your field, and only look up words that block the main argument. Vocabulary within a field repeats a lot, so speed improves quickly.
How should I take notes on papers for a literature review?
Use a fixed template: citation, research question, design and sample, main finding, limitations and link to your own project. The link field is what makes writing the review easier.
