Last updated 9 min read
Key takeaways
A research paper is not a story with a twist at the end; it is a reference document with a predictable structure. Most empirical papers in the sciences and social sciences follow some version of IMRaD — Introduction, Methods, Results and Discussion — with an abstract in front and references behind. The guide to how research papers are structured explains why. Reading straight through gives every part equal attention, when the parts deserve very different attention depending on your purpose.
Your reason for reading changes the route. If you are screening papers for a literature review, you may only need the abstract and figures. If you have to present the paper in a seminar, you need the methods in detail. If you are building your own study on it, you need to know its weaknesses better than its authors admit. Decide which of these you are doing before you open the PDF.
Knowing what each section is for tells you which one answers the question you have right now.
Humanities articles rarely use these headings, but the same jobs get done in a different order: an introduction that stakes out a position against other critics, a body that builds the argument from evidence, and a conclusion that states the larger claim. Look for the sentence where the author says what they argue; it usually sits near the end of the introduction.
| Section | What it tells you | Read it to answer |
|---|---|---|
| Abstract | The whole paper in one paragraph: question, method, main result | Is this relevant to me? |
| Introduction | Why the question matters and what earlier work left open | What gap does this paper claim to fill? |
| Methods | What was done, to whom or what, and how it was measured | Could this design actually answer the question? |
| Results | What was found, usually in figures, tables and statistics | What happened, and how large was the effect? |
| Discussion | What the authors think the results mean, plus limitations | Do the conclusions go further than the data? |
| References | The work the paper builds on or argues against | What should I read next? |
The computer scientist S. Keshav described a three-pass approach to reading papers that many researchers use in some form. Adapted for students, it runs like this:
Figures are where many papers state their results most plainly, so learn to read them before the text that interprets them. For any chart, read the axis labels and units first, then the caption, then the legend. Ask what is being compared, how many observations sit behind each point or bar, and what the error bars represent — standard deviation, standard error and confidence intervals are different things, and the caption or methods should say which is shown.
You do not need to follow every calculation to read statistics critically. Look for three things: the size of the effect (how big is the difference, in real units?), the uncertainty around it (how wide is the interval?) and the comparison (compared with what control or baseline?). A small p-value says the data would be unlikely if there were truly no effect, given the study's assumptions; it does not say the effect is large or important.
Critical reading means testing the paper's claims against its own evidence. Keep these questions beside you on the second and third passes, and note an answer to each for any paper you plan to cite as key evidence. For a broader checklist that also covers websites and reports, see the guide on evaluating a source.
The first few papers in a new field are slow for everyone, mostly because of vocabulary and unstated background. Start with a recent review article before reading individual studies; it explains the terms and tells you which papers matter. Keep a running glossary of terms and abbreviations, with a plain-language definition beside each, and check it before you look anything up a second time.
Do not get stuck on one paragraph during the first or second pass. Mark it, move on, and come back with the context of the rest of the paper, which often explains it. If the methods rely on a technique you have never met, read one textbook explanation of it; you will see it again in the next paper. Speed in a field comes from accumulated background, not from reading faster.
The test of a good reading is whether you can use the paper a month later without opening it. End each serious read with a short note in your own words — not a set of highlights, which tell future-you what looked important but not why. One reliable template: citation; question; method and sample; main finding with its size; your main doubt; and one line on how it relates to your project.
When the notes pile up, the guide to writing a literature review shows how to group them by theme rather than by paper, so they become the body of a review.
When a methods section stops making sense, the worst option is to skip it. In Cavua's Work space you can upload the paper as a PDF and ask about it directly — “what did the control group do?” or “what do the error bars in this figure show?” — with answers that point to the page, so you check them against the text yourself.
Upload several papers and you can ask across them at once, which helps when you are comparing designs for a review. Use “Quiz me” after the second pass to see whether you can recall the design and findings without looking, and talk a hard paper through with the tutor by voice or video call, within your plan. The guide to studying a PDF with an AI tutor shows this way of working in more detail.
What an AI research assistant can do for students, what it can't, how to test one before you rely on it, and what to keep out of the files you upload to it.
APA citation in the 7th edition: in-text rules for one, two and three or more authors, plus reference formats for articles, books, chapters and webpages.
Learn how to avoid plagiarism with clear rules for quoting, paraphrasing and citing, examples in APA and MLA, and a checklist to run before you submit.
Where to find credible sources for research: library databases, scholarly search engines, official data and citation trails, plus quick checks before citing.
How to use AI for research without being misled: where AI tools help, the five ways they mislead, a verification workflow and how to disclose AI use honestly.
How to write a literature review step by step: set the scope, search and read with a system, group sources by theme, and structure a review that argues.
It depends on the depth you need. A first pass to decide relevance takes five to ten minutes. A second pass that captures the argument and evidence can take up to an hour for a typical paper. A full critical read of a paper central to your project can take several hours, especially in an unfamiliar field. Most papers you find only ever need the first pass.
Yes. The abstract is the fastest way to learn the question, method and main result, and to decide whether the paper deserves more time. Just do not stop there for a paper you will cite: abstracts compress and sometimes state findings more strongly than the results support. Read the relevant results and limitations before relying on any claim the abstract makes.
Not every calculation, but you should grasp three things: how large the effect is in real terms, how uncertain the estimate is, and what it was compared with. Those let you judge whether a finding matters. When a test or term is new, look it up once and add it to a glossary, since the same methods recur across papers in a field.
A common order is title, abstract, introduction, figures and tables, conclusion or discussion, and then the methods and results in detail if the paper earns it. The aim is to understand the claim before judging the evidence. Many readers jump to the figures right after the abstract, because they often show the results more directly than the text.
Write a short summary in your own words immediately after reading, including the method, the main finding and your biggest doubt. Then test yourself later by trying to recall those points before rereading your note. Connecting each paper to your own question, and to the papers it agrees or disagrees with, also makes it far easier to remember than an isolated set of highlights.
Yes, as long as you skip deliberately. When screening, you can skip the methods and results text entirely. For a paper you cite as key evidence, do not skip the methods, because that is where problems hide. Never cite a claim from a part of the paper you have not read, especially when the abstract states it more strongly than the results do.
Every Cavua course comes with a 24/7 tutor you can video call, speak to or chat with — on the material you are actually studying.
