TL;DR
This page gives you a step-by-step way to learn how to read a scientific paper in high school, in the order researchers actually use. You don’t read a paper from top to bottom: skim the title, abstract, and figures first, then go back for the introduction, methods, results, and discussion over two or three passes. Below you’ll find the section-by-section method, the three-pass method, plain help with statistics, a note-taking template and a checklist for how to tell if a paper is any good. If you’d like a PhD mentor to read papers alongside you, apply to Horizon.
What are the parts of a scientific paper?
Most science papers follow the same order, often called IMRaD: Introduction, Methods, Results, and Discussion, with a title, abstract, and references around it. Once you know what each part is for, a long paper stops looking like a wall of text. Scan the headings first, since some fields bend the format. Math and computer science papers may add proofs or algorithms, and humanities papers often use topic headings instead. The same parts show up in your own paper, so learning them now helps you write later.
| Part of the paper | What it tells you | The one question to ask yourself |
| Title | The topic, and sometimes the main finding | What is this paper about? |
| Abstract | A summary of the whole paper | What did they ask, and what did they find? |
| Introduction | Background and why the study was needed | What gap are they trying to fill? |
| Methods | Who or what was studied, and how | Could someone repeat this study? |
| Results | What happened, with data | What do the numbers and figures show? |
| Discussion and conclusion | What the authors think it means, and the limits | Do the claims match the data? |
| References | The earlier work the authors built on | Which papers should I read next? |
| Supplementary materials (optional) | Extra data and details that didn’t fit in the paper | Do I need this for my question? |
Source: Horizon Academic Research Program
How to read a scientific paper in high school: the section-by-section method
The reading order is not the printed order. You’ll start at the front, jump to the figures, and only then read the body text. If your first paper feels confusing, that’s normal, and each step below gives you one clear next move.
To show what each step looks like, we’ll follow a made-up study. This is an illustration, not a real paper: researchers randomly assign 120 high school students to sleep either 6 hours or 9 hours a night for one week, then give everyone a 20-word memory test.
Step 1: Read the title and abstract
The abstract is a short summary of the paper, usually one paragraph near the top. In our illustration, it might say that students who slept 9 hours remembered more words than students who slept 6. Before you move on, write the main question in one sentence, such as “Does more sleep help teens remember words?” Don’t trust the abstract alone, since it’s the authors’ best-case summary.
Step 2: Look at the figures and tables before the text
Read every figure title, axis label (the words along the sides of a graph), and legend (the key that explains colors or symbols). Ask what’s being compared and which way the results go. In our illustration, a bar chart might show two bars, one for each sleep group, with the 9-hour bar taller. Error bars, the small lines on top of the bars, show how spread out or uncertain the data is. Give the figures five minutes before you read a full paragraph of text.
Step 3: Read the introduction for the “why”
The introduction explains why the study was needed. Look for the gap the authors say they’re filling and their hypothesis, which is a testable prediction. Skim the background paragraphs, then slow down at the last paragraph of the introduction, where the aim is usually stated. In our illustration, it might say the team tested whether 9 hours of sleep instead of 6 improves word recall in teens. The last paragraph of the introduction is where you’ll usually find the aim.
Step 4: Read the methods for “who, what and how many”
Find out who or what was studied and how many (the sample size, written as n). Then look for what was changed, what was measured, and whether there was a control group, which is a comparison group that doesn’t get the change. In our illustration, n is 120, with 60 students per sleep group, and students were randomly assigned. On a first read, it’s fine to skip lab details like brand names and exact amounts.
Step 5: Read the results and match them to the figures
Results report what happened, not what it means. Match each claim in the text to a figure or table. In our illustration, the 9-hour group might average 14.2 words out of 20 and the 6-hour group 11.8. The statistics section below explains p-values and other numbers you’ll meet here.
Step 6: Read the discussion with a critical eye
The discussion covers what the authors think the results mean, how they connect to other studies, and the limitations paragraph, where they admit what the study couldn’t do. Check whether the claims stay within what the data showed. In our illustration, saying “9 hours of sleep helped these students remember words after one week” fits the data. Saying “sleep makes all students smarter” goes too far.
Step 7: Mine the references
Use the reference list to find the two or three papers this one is built on. Tools like Google Scholar’s “Cited by” link also show newer papers that cite this one. Following the references is the fastest way to build a reading list. See our guides to using Google Scholar and Semantic Scholar for high school researchers.
Read it more than once: the three-pass method
Computer scientist S. Keshav describes a routine in his short paper “How to Read a Paper” (2007), listed in Resources. The idea is to read a paper up to three times, going deeper only when it earns the time. Pass one is a quick skim of the title, abstract, introduction, headings, conclusions, and references, and takes about five to ten minutes. Pass two is a careful read that skips heavy proofs and closely examines figures, and it takes up to an hour. Pass three is a deep read where you try to rebuild the work in your head, and Keshav says it can take about four or five hours for beginners.
For you, pass one decides whether a paper is useful for your project. Pass two is for papers you’ll cite. Pass three is only for the one or two papers your project is built on.
| Pass | Goal | What you read | About how long |
| One | Decide if the paper is useful | Title, abstract, introduction, headings, conclusions, references | Five to ten minutes |
| Two | Understand it well enough to cite | Everything, skipping heavy proofs, with a close look at figures | Up to an hour |
| Three | Rebuild the work in your head | Everything, in depth | About four or five hours for beginners |
Source: Keshav, “How to Read a Paper,” 2007
How to handle jargon without getting stuck
Not knowing a term is expected, and it doesn’t mean you’re not ready for research. Try these tactics:
- Keep a running word list, and write each term as you meet it.
- Look a term up only if it blocks the main idea. Otherwise, keep reading.
- Check a review article or textbook chapter for the basics of the topic.
- Use the paper’s own introduction, which often defines key terms.
- Reread a hard paragraph after you finish the section. It often makes more sense with context.
If you want a starting list, our research terms glossary covers common research vocabulary.
How to read the statistics: p-values, confidence intervals and sample size
You don’t need to redo the math. You need to know what the numbers are telling you and when to be careful. The table below covers the terms you’ll see most.
| Term | What it means in plain words | What to check |
| Sample size (n) | How many people, animals, cells, or samples were studied | Is it large enough for the claim, and how many dropped out? |
| Mean and standard deviation | The average, and how spread out the data is | Is the spread small or large compared with the difference between groups? |
| p-value | If there were really no effect, how surprising would results like these be? A small p-value (often below 0.05) means the result would be unlikely if nothing were going on. | It does not tell you the chance that the hypothesis is true, and it does not tell you how big or important the effect is. |
| Statistically significant | The p-value fell below the cutoff the authors chose | It doesn’t mean “important.” Look at the effect size too. |
| Confidence interval (CI) | A range of values that are reasonable for the true effect, given the data. Narrow ranges mean more precise estimates. | If the range for a difference includes zero, the study can’t rule out “no difference.” |
| Effect size | How big the difference or relationship actually is | Is it big enough to matter? |
| Correlation | Two things move together | Did the study design, such as a randomized experiment, allow the authors to say one caused the other? |
Source: Horizon Academic Research Program
Here’s how it looks in our made-up study. The 9-hour group might remember about 2.4 more words on average, with a confidence interval of 1.1 to 3.7 words. That range stays above zero, so the study can rule out “no difference” for these students. That still doesn’t mean the effect holds for every teen. The American Statistical Association’s 2016 statement on p-values, listed in Resources, is a short, readable guide.
A note-taking template for reading a paper
Fill out one template per paper, in your own words. Write the summary without looking at the abstract. Copy this table into a document and reuse it for every paper.
| Field | What to write |
| Full citation | Authors, year, journal, DOI |
| Type of paper | Primary research, review, or preprint |
| Research question | One sentence, in your words |
| Hypothesis | What the authors predicted |
| Methods | Who or what, n, design, control group |
| Main result | One sentence, with the key number |
| The most important figure | Which one, and what it shows |
| Authors’ conclusion | What they say it means |
| Limitations | The authors’ and yours |
| Words to look up | Terms that blocked you |
| How this connects to my project | Why you’re keeping this paper |
| Papers to read next | From the reference list or “Cited by” |
Source: Horizon Academic Research Program
This template covers one paper at a time. When you have several sets of notes, our guide on how to write a literature review shows how to combine them.
Review articles vs. primary research papers: which should you read first?
Start with a recent review article to learn the field, then read the primary papers it cites for the evidence. A planned Horizon post covers original research vs. a literature review as project types, so here we’ll stick to reading.
| Feature | Primary research paper | Review article |
| What it reports | The authors’ own new experiment or study | A summary of many earlier studies |
| Typical sections | Introduction, methods, results, discussion | Topic headings, summary tables, long discussion |
| How to spot it | Has its own methods and results sections | “Review” in the title or article type label, no methods for its own experiments, very long reference list |
| Best used for | Evidence you can cite | Learning the field and finding papers |
| How to read it | Use the seven steps | Read headings, figures and summary tables |
Source: Horizon Academic Research Program
How long should it take to read a research paper?
Your first few papers will take much longer than an expert’s, and that’s normal. Use the Keshav timings from the three-pass section as your reference point: five to ten minutes for pass one, up to an hour for pass two, and four or five hours for a beginner’s pass three.
A few habits help. Split a hard paper across two or three sittings. Set a goal for each session, such as “figures and abstract today.” Read fewer papers carefully instead of skimming many. For planning your whole project, see how long it takes to write a research paper.
How to tell if a scientific paper is any good
Ask these questions after pass two.
- Was it peer reviewed, or is it a preprint? Peer review means other experts checked the work before publication. Preprint servers like bioRxiv, medRxiv, and arXiv post work before that check, and medRxiv itself warns that preprints should not be used to guide clinical practice or health decisions.
- Where was it published? For open access journals, check whether the journal is listed in the Directory of Open Access Journals (DOAJ).
- Is the sample big enough, and was there a control group?
- Do the results and figures support the conclusion, or do the claims go further?
- Do the authors name their limitations?
- Who funded it, and do the authors declare any conflicts of interest?
- Has it been retracted or corrected? Check the publisher page and the Retraction Watch Database.
- Have other studies found the same thing? One study is a data point, not proof.
Student journals use review too. See our list of research journals for high school students.
Common mistakes when reading your first papers
- Reading word by word from the first line
- Skipping the figures
- Trusting the abstract alone
- Treating one study as proof
- Mixing up correlation and causation
- Ignoring the limitations paragraph
- Not taking notes, then forgetting everything a week later
- Quitting at the first unfamiliar word
These get easier to avoid if you follow the seven steps in order.
Frequently asked questions
Do the first pass only: title, abstract, figures, headings, and conclusion. Then decide whether the paper is worth a second pass.
They are the title, abstract, introduction, methods, results, discussion, and references. The parts table near the top of this page shows what each one tells you.
No. Start with a review article or textbook chapter on the topic, and keep a word list as you go.
AI can help you check your understanding, but you still need to read the paper yourself and verify anything it tells you against the text. Rules vary: the College Board’s 2026-27 AP Capstone policy allows AI for confirming your understanding of a complex text, but some student journals, such as the Journal of Emerging Investigators, don’t allow AI to summarize or analyze published studies for a submitted paper. For more, see Is Using ChatGPT Plagiarism?
Write four to five sentences: the question, the method, the main result with its key number, one limitation, and why it matters to your project.
Try Google Scholar, Semantic Scholar, PubMed Central, and arXiv. See our guides to Google Scholar, Semantic Scholar, and Top Research Databases for High School Students.
Practice with papers in one narrow topic so the vocabulary repeats, and read with a mentor or teacher who can answer questions. Horizon pairs you 1:1 with a PhD scholar or professor for several months of guided research, and reading the key papers in your topic together is part of that. See what a research mentor actually does and Horizon’s programs.
Start with one paper this week
Pick one paper on a topic you care about. Worried about reading papers on your own? Horizon pairs you 1:1 with a PhD scholar or professor (from universities like Oxford, Stanford, and Columbia) for several months of guided research, ending in a roughly 20-page paper. Apply to Horizon or join an info session. Strong reading skills show up in the quality of a research paper, which is what admissions readers notice.
Resources
- Ten simple rules for reading a scientific paper, Carey, Steiner and Petri, PLOS Computational Biology, 2020
- How to Read a Paper, S. Keshav, ACM SIGCOMM Computer Communication Review, 2007
- How to read a scientific paper, Science (Careers)
- The ASA Statement on p-Values, Wasserstein and Lazar, The American Statistician, 2016
- Retraction Watch Database
- About medRxiv, preprint caution
More from Horizon
- 8 tips on how to do biology research in high school
- Can you do real research online in high school?
- Research programs for complete beginners
- How to choose a research topic
- How to use Google Scholar as a high school researcher
- Semantic Scholar for high school researchers
Image source: Google Scholar