How to Write an Academic Paper: A Section-by-Section Guide
Writing an academic paper is not simply arranging everything you know in order. You need a traceable line from the research question through methods and results to interpretation, with every external claim supported by the right source. This guide follows each section from the first outline to journal submission.
1. Define the question and contribution in one sentence each
Before drafting, write separate sentences for the question your study answers, the limitation or gap it addresses, and the contribution the results can support. These sentences become the paper's backbone: the introduction leads to the gap, the methods answer the question, and the discussion returns to the contribution.
A broad topic is not a research question. Specify the measurable relationship, context or population, and comparison where applicable. Treat a gap discovered in a literature scan as a proposition to verify, not an established fact.
2. Finalize the title and abstract last
A useful title communicates subject, design and context without exaggerated novelty claims. Include the terms a researcher would naturally use to find the work. The abstract should state purpose, methods, principal results and conclusion without introducing claims absent from the manuscript.
Retain essential numerical results, limit abbreviations, avoid citations, and be cautious with words such as first, unique or definitive. The abstract is a compressed account of the paper, not a promotional alternative to it.
3. Move the introduction from problem to question
- Establish the problem with reliable and current evidence.
- Synthesize relationships between studies instead of listing summaries.
- Identify a contradiction, methodological limitation or missing context.
- End with the objective, research question and hypothesis where relevant.
- Do not argue for your results before presenting the study.
4. Write methods so the study can be evaluated and reproduced
Describe the design, sample or data source, eligibility criteria, variables, instruments and analysis. Readers need to understand both what you did and why the chosen procedure was appropriate.
Report ethics approval, consent, preregistration, data availability and code access where applicable. Software version and consequential analysis settings can matter as much as the software name.
5. Separate results from interpretation
Present what the data show in Results without duplicating every value across prose, tables and figures. In Discussion, answer the research question, compare findings with prior evidence and offer explanations proportional to the strength of the design.
Do not infer causality from an associational design or treat statistical significance as practical importance. Limitations should define the conditions under which findings can be interpreted, rather than serve as a generic apology list.
6. Audit references together with the claims they support
- Match every in-text citation to the bibliography in both directions.
- Compare title, author, date, journal, pages and DOI with the original record.
- Remember that a real paper does not necessarily support the attached claim.
- Check retractions and substantial post-publication corrections.
- Apply the target journal's citation style consistently.
What should happen after the draft?
Verify the bibliography, compare claims with cited evidence and confirm the target journal's identity. Open the pre-submission workflow →