Used by 1,000+ researchers and students
A research paper example, section by section
Most examples online are templates with the sentences taken out. This one is a published, open-access paper read in order, so you can see what each section actually carries — and then write yours with the sources and statistics in one place.
A published paper, read in order
Below is one real open-access paper, with a link to the original. Each section is named with what it is doing, not with what a template says it should do. Read it once in order, then look at your own draft and ask which of these jobs your sections are not doing.
Abstract
States the question, the method, the main result and what it means — in that order, in one paragraph.
We found significantly reduced hopelessness and trait anxiety in the training group compared to the control group at post-assessment (one day post intervention).
Introduction
Narrows from the field to one gap, and ends with the question the paper will answer.
Therefore, our main research questions were to investigate the training’s effects on the primary outcome self-efficacy and on the secondary outcomes anxiety, stress, hopelessness, hope, and positive and negative affect (pre-post changes).
Methods
Enough detail for someone else to repeat it. Not a tutorial on the technique.
The study was a randomized controlled trial (Fig 1). Ethical approval was obtained from the University of Zurich Ethics Committee of the Faculty of Arts (No. 20.4.24).
Results
What was found, with the statistics. No interpretation here — that is the next section's job.
As hypothesized, we observed a significant effect of the self-efficacy training on changes in hopelessness, b = −0.06, SE = 0.03, p = .02, β = 0.24 and trait anxiety, b = −0.18, SE = 0.06, p = .003, β = -0.30
No significant effects were detected for the other outcome variables, including self-efficacy.
Discussion
What the result means, where it disagrees with earlier work, and what it cannot show.
First, we did not implement longer-term follow-up measures, so we cannot elaborate on the long-term effects of the self-efficacy training.
References
Only what is cited in the text, in the journal's style. This journal numbers them in the order they are first cited: 89 in this paper.
Self-efficacy refers to an individual’s belief in their ability to exercise control over their environment, thoughts, emotions, and behavior and has been associated with positive mental health outcomes and adaptation to adverse events [1, 2].
Notice the Results: next to the two outcomes that improved, the paper says in the same plain voice that the others, self-efficacy among them, showed no significant effect. A results section that leaves out what didn't work is the first thing a reviewer asks about.
From sources to a submitted manuscript
The parts of a paper that usually need four separate tools.
- Search
- Read
- Analyze
- Draft
- Check
- Export
The literature you will cite
OpenAlex, Crossref and DOAJ run on every search — plus PubMed and arXiv when your topic is in their fields. Free PDFs come first from the indexes' own open-access records, with Unpaywall second.
Statistics reported the way journals want them
Upload your table and the tests are computed and written in APA format. The numbers come from computation, not from the model — the model writes the interpretation and you check it against the output.
Sections drafted from your library
Writing tops up from the indexes unless you turn External sources off. A citation that resolves to nothing in your library does not enter the text; when a search result is cited, that record is added. Changing a written passage takes a rewrite, and the instruction there is careful academic prose, never a performance of casualness — no slang, no jokes, no manufactured typos — with every change shown as a tracked edit you accept or reject.
A manuscript file, not a text dump
Word, PDF and LaTeX keep the page setup, numbered headings and captions, and Word keeps real footnotes. Markdown carries none of it, which is fine for a draft and wrong for a submission.
What the journal will ask you to declare
Publishers now ask about AI use at submission, and the question is narrower than people expect: they want to know whether a model wrote text, not whether you used a spell checker. Most ban AI as an author outright, because an author has to be able to take responsibility.
Grongy does not read the journal's policy. What it keeps is the record you answer from: every passage the AI adds is logged at paragraph level, and the disclosure draft is built from that log for you to edit.
Check your target journal's instructions for authors before you write the statement — the required wording differs between publishers, and a statement in the wrong place gets the manuscript returned.

Your data, analyzed where you cite
Upload an Excel or CSV file and pick the comparison. The table is read in your browser and the analysis runs there, with no code to write.
t-tests · ANOVA · correlation · regression · chi-square · nonparametric tests · qualitative coding
The output comes in APA format. When you ask for an interpretation, the first 40 rows of the table and the computed statistics go to the AI. The statistics themselves are not generated — they are computed, and the model writes only the reading of them.
Turning a thesis chapter into a paper
A chapter and an article are not the same document. The chapter proves you understand the field; the article makes one claim and defends it in a fraction of the words. Turning one into the other is structural work.
What usually has to change:
- One claim, not three — the chapter's secondary findings become future work
- The literature review shrinks to what the claim needs
- The method loses the teaching tone: no explaining what a t-test is
- Tables and figures are rebuilt to the journal's limit, not the chapter's comfort
- The abstract is written last and to the journal's word count
- Your own chapter becomes a source you cite, not text you reuse silently
The last line is the one that gets missed. Reusing your own text without citing it is still something your publisher will flag.
Checks before you submit
Two things are worth running before a manuscript leaves your hands, and neither of them is a similarity score.
The citation audit compares each sentence against the source it cites, using the source's own indexed pages and abstract — never an AI summary — and shows the evidence as a verbatim quote. The submission check runs a fixed list with no AI: an abstract outside 80 to 500 words, a missing keywords line, figures and tables never mentioned by number, citations with no source behind them, placeholders left in the text.
Source integrity reports fifteen kinds of finding on the records themselves — retracted papers, unregistered DOIs, records that do not match the index — with no AI and no usage credit.
What the journal expects in the file
Instructions for authors differ in detail and agree on the frame. The parts below are the ones that come back as a desk rejection when they are missing.
- An abstract inside the word limit, with the keywords the journal asks for
- Numbered sections, captions above tables and below figures
- A reference list in the journal's style — the editor carries more than 10,000
- Figures at the resolution the journal states, not at screen resolution
- A data availability statement if the journal requires one
- An AI use statement where the publisher asks for it
Grongy exports with your page setup, heading numbers and captions intact; the rest of the journal's instructions are still yours to confirm.
Start free
Paper structures, data analysis, the submission check and version history are on the free plan; building the reference list in your chosen style and the share link are on Plus and Pro.
Free
$0/mo
Start for free- Find academic sources and write from them
- A limited review of your thesis before you hand it in
- Upload PDFs and ask your papers questions
- AI help that refills every 5 hours
- Word, PDF, LaTeX and Markdown export
- 20 PDF uploads (10 MB, 150 pages each; 50 MB in total)
Plus
$11.99/mo
Get PlusEverything in Free, plus:
- More AI use
- Vision: slides, designs and dashboards
- A deeper review of argument and structure
- Automatic reference list in any citation style
- Advanced models in chat: Claude, GPT, Gemini, Grok
- 10 GB storage
- Share drafts with your supervisor
Pro
$19.99/mo
Get ProEverything in Plus, plus:
- Even more AI use
- 25 GB storage
- Early access to new AI models
Prices in US dollars, taxes included.
Questions authors ask
Can I publish a paper written with AI?
Most publishers allow AI assistance and require you to declare it, and the major ones do not list AI as an author. The line they draw is responsibility: you have to be able to defend every sentence, which is also why the use record exists.
How long should a research paper be?
The journal decides, and the range is wide — 3,000 words in some fields, 10,000 in others. Write to the instructions for authors, not to a general rule, and check whether the limit includes references.
Is a research paper the same as a lab report?
No. A lab report documents one procedure and its result for a course; a research paper makes a claim to a field and defends it against the literature. Both structures are in the editor, and picking the wrong one shows immediately in the sections.
Can it run the statistics for me?
It computes them. t-tests, ANOVA, correlation, regression, chi-square and nonparametric tests are calculated in your browser and reported in APA format; you choose the test. A model guessing at statistics would be the worst possible use of one.
What happens to my data file?
The analysis runs in your browser. When you ask for an interpretation, the first 40 rows and the computed statistics go to the AI — not the whole file. Which services are involved is written in the privacy policy.
Start with the sections you have
Bring your sources and your data table. The structure, the statistics and the reference list come from the same place, which is the part that usually costs a week.