On this page
A teacher mentions a study, a news article links to a journal, or you start a project and need real sources. You open the paper and meet a wall of terms, graphs and statistics. Most students try to read it like a textbook, from the first line to the last, and give up halfway. This post explains how researchers actually read papers, walks through a real example, and shows where to find papers without paying.
How is a research paper organised?
Most science papers follow the same shape, called IMRAD: Introduction, Methods, Results and Discussion. Sollaci and Pereira (2004) traced it in four major medical journals and found it began to be used in the 1940s, reached 80% of articles in the 1970s, and was the only pattern used for original papers in the 1980s. Before the introduction there is an abstract, a short summary of the whole paper.
Each part answers one question:
Introduction: what was already known, and what question this study asks.
Methods: what the researchers did, with whom, and how they measured it.
Results: what they found, usually in tables and graphs.
Discussion: what they think it means, and its limits.
What is the three-pass method?
The computer scientist S. Keshav described a method in How to read a paper (2007) that many students now use.
First pass, about five to ten minutes. Read the title, abstract and introduction carefully. Read the section headings but nothing else. Read the conclusions. Glance at the references to see which ones you recognise. You should now know what kind of paper it is and what it claims.
Second pass, up to an hour. Read with more care, but skip proofs and fine details. Look closely at the figures: are the axes labelled, are there error bars? Mark references you may want to read later. After this pass you should be able to summarise the paper, with its evidence, to someone else.
Third pass, only when you need it. Try to rebuild the study in your head, making the same assumptions as the authors. Keshav notes this can take beginners four or five hours. Most school and college students rarely need it.
The method gives you permission to stop. Many papers only need the first pass.
Which parts do beginners find hardest?
Hubbard and Dunbar (2017) surveyed 260 undergraduates and researchers in biological sciences at a UK university. Inexperienced readers found the methods and results sections the most difficult, and undervalued the results section and critical reading of data. Skill in reading results developed slowly, even across a research career.
So if graphs and statistics feel hard, you are not behind. That is the normal place to struggle, and it is also where the evidence actually lives.
A worked example: a real study on exam nerves
Take the abstract of Jamieson and colleagues (2010), a study on test anxiety. A first pass gives you this:
Question: does thinking of physical arousal as helpful improve exam performance?
Who: students preparing for the GRE, a graduate admissions test in the United States.
Method: some were told arousal improves performance; others were not. Saliva samples measured a marker of nervous-system activity.
Result: the "arousal helps" group scored higher on the maths section in the lab, and again on the real GRE one to three months later.
Now ask the questions a careful reader asks:
How many people? The abstract does not say, so you would check the methods.
Who were they? Graduate test-takers in the US, not Indian school students. Would the result carry over?
What exactly was compared? A message about arousal against no message. Was the control group treated the same in every other way?
What was measured? Maths scores. The verbal section is not mentioned in the abstract, which is worth checking.
You have not read the full paper yet, but you already know what it claims and what you would check next. Our post on treating exam nerves as energy shows how this study and two others fit together.
How can you tell if a paper is trustworthy?
Check the journal and whether it is peer reviewed.
Look for the sample size and who the participants were.
See whether the study was randomised and had a control group.
Search for later studies that tried to repeat it. One study is a starting point, not a final answer.
Be wary of a news headline that says more than the abstract does.
AI chatbots can help you understand terms, but they sometimes invent references. Our post on AI and made-up sources shows how to check.
Where can you find papers for free?
Many papers are free to read. PubMed Central holds free full texts in biology and medicine, and arXiv holds preprints in physics, mathematics and computer science. arXiv itself does not peer review them, so check whether a preprint was later published in a journal. Authors often post accepted manuscripts on their university's repository, so searching the title plus "pdf" often finds a legal free copy.
Reading research also becomes easier when the background concepts are clear. A course that teaches the underlying science helps: Fermi's courses are university-level courses taught by working scientists, with the AI tutor available on every lecture, so you can ask about a term while you learn it.
A practice routine
Courses that teach students to read papers in a structured way can shift their confidence. A study of the C.R.E.A.T.E. method by Hoskins, Lopatto and Stevens (2011) found significant gains in undergraduates' confidence in reading and analysing research articles within one semester. You can build a lighter version on your own:
Pick one paper a week on a topic you are studying.
Do the first pass and write a five-line summary: question, who, method, result, one doubt.
Once a month, do a second pass on one paper and redraw its main figure by hand.
This week, find one paper in PubMed Central or arXiv on a topic from your syllabus and write its five-line summary.
