How to Write a Research Paper: Steps and Rules

You have a topic, your advisor has signed off on it and the due date is on your calendar. What is usually missing is the answer to a question no syllabus covers: what exactly makes your text a piece of research, and who decides that?

You write a research paper by framing one narrow research question, working through the existing literature on it, choosing a method others can follow, reporting your results in full and without selection, and citing every borrowed claim so that a third party can check it. By the end of this guide you will know the sequence, the rules behind it and where those rules are actually written down.

📌 Key points at a glance

• Traceability decides the grade, not the page count.
• The research question comes before the outline, never after.
• Federal rules define misconduct as fabrication, falsification, plagiarism.
• Self-plagiarism is expressly not federal research misconduct.
• Many honor codes still treat resubmitted work as a violation.

What makes a paper scholarly?

A research paper is a text that answers one clearly bounded question using a method it states openly, and that cites every claim so others can retrace the route to the result and check it. What makes a paper scholarly is therefore not the subject but the traceability.

A single sentence shows what that means in practice. “Remote work increases satisfaction” is worthless in a term paper unless it says who found that, where, and by what method. Add the source, the year and a note on the group studied, and the same sentence becomes a scholarly claim. The difference lies in verifiability, not in wording.

Four features appear in almost every department handbook, whether the assignment is a term paper or a master's thesis:

• Traceability: every step is described well enough for others to repeat it.
• Citation: borrowed ideas are marked, including paraphrases and translations.
• Completeness: results that contradict your expectation still go into the text.
• Objectivity: judgments are labeled as judgments and justified.

If you first want to know what scholarly practice means at all and how it differs from everyday knowledge, the article on scholarly practice and its typical sequence sets that out.

How to write a research paper: the seven steps

A research paper comes together in seven steps, and the order is not a suggestion. It is the reason some papers have a clear thread running through them and others read as a sequence of sections. The most common detour is starting with the outline. An outline without a research question is a list of topics.

The worked example throughout this guide is a senior thesis on satisfaction with remote work arrangements at a mid-sized company. It reappears at several points so the steps stay concrete.

1. Narrow the topic. “Remote work” becomes “satisfaction with remote work arrangements at a company with 120 employees”. Check that enough literature exists.
2. Frame the research question. One question you can answer with the time and means available. The distinction between research question, thesis and hypothesis shapes the whole structure.
3. Search the literature. Read broadly first, then closely. Record full citation details right away, or you will hunt for them again three weeks later.
4. Choose the method. Literature review only, or your own data collection? This decision affects your timetable more than any other.
5. Build the outline. Only now. The outline follows the research question, not the other way round.
6. Write and cite. Add each citation as you write, not afterward.
7. Revise and check the formalities. Put introduction and conclusion side by side at the end: whatever the introduction promises, the conclusion must deliver.

💡 Tip

Draft the introduction first and rewrite it last. The draft forces you to put the research question into words. The rewrite makes sure it matches what the paper actually ended up saying.

Term paper, thesis, dissertation: what the terms mean in the US

American terminology differs from British usage in one important respect. In the United States a thesis is the extended piece written for a master's degree and a dissertation is the doctoral work; in the UK the two are reversed. Cornell's Graduate School, for instance, speaks of completing and defending the master's thesis or doctoral dissertation.

Undergraduate work carries its own labels: term paper, research paper, senior thesis, honors thesis and, in some programs, a capstone project. The figures below are guides from common practice, not rules. Your department handbook is what binds you, and it varies between two programs at the same school.

Which parts belong in the finished piece, from title page to appendix, is covered in the article on the structure of a research paper. Whether you should collect data at all is settled by the comparison of empirical and theoretical theses.

Who sets the rules: federal regulation and your honor code

Two separate rulebooks apply in the United States, and confusing them is the most common misunderstanding in this area. Federal regulation governs federally funded research. Your own school's honor code or student conduct code governs your coursework. They overlap, but they do not say the same thing.

The federal rule is 42 CFR Part 93, rewritten in a Final Rule published on 17 September 2024, effective 1 January 2025 and applicable to conduct from 1 January 2026. Section 93.234 defines research misconduct as fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results, and adds that it does not include honest error or differences of opinion.

Self-plagiarism: where federal rules and honor codes part ways

Section 93.227(b) is unusually blunt about a case students ask about constantly. It states that plagiarism does not include self-plagiarism or authorship or credit disputes, and that self-plagiarism and authorship disputes do not meet the definition of research misconduct. That is federal law. Your school may say the opposite: Stanford's Honor Code, for example, treats submitting work more than once as a violation unless you have sought and been granted your professor's approval. Reusing your own earlier paper can therefore be federally unremarkable and still cost you the course.

The three elements a finding requires

Section 93.103 sets a deliberately high bar for a federal finding. There must be a significant departure from accepted practices of the relevant research community, the conduct must have been committed intentionally, knowingly, or recklessly, and the allegation must be proven by a preponderance of the evidence. Section 93.319 makes the boundary explicit: institutions may have standards of conduct different from the standards for research misconduct under this part.

A result that disproves your hypothesis is still a result. Only hiding it is a mistake.

Both texts are public: 42 CFR Part 93 in the eCFR and the Stanford Honor Code as one example of an institutional standard. What applies to your own paper is decided by your school's policy, and by your advisor if in doubt. This article is not legal advice.

What belongs in your methods section

The methods section of an empirical paper has to be precise enough for a stranger to repeat your data collection. That is the only standard that matters, and it is where most papers fall down, not on the statistics. Writing “an online survey was conducted” announces the section without writing it.

In the remote work example that means: who was asked, how many replied, over what period, through which channel, and who was deliberately not asked. The last point is almost always missing and is the most interesting one, because it sets the limits of what you can claim.

The six details every methods section needs

• Population: the group your claim is meant to cover.
• Sample and sampling method, including any exclusions.
• Number of cases, given as n, and the response rate.
• Collection period with a start and end date.
• Instrument, meaning the questionnaire or interview guide.
• Analysis method and the software you used.

The full questionnaire belongs in the appendix, not in the body. How to build one without ruining your own analysis is set out in the guide to creating a questionnaire. For the question of how many responses you actually need, see the article on sampling.

⚠️ Careful

Research involving human subjects usually needs review before you collect anything. At most US institutions that means an Institutional Review Board, and approval is granted before the first response arrives, not afterward. Ask your advisor which category your study falls into. What to do about the data itself is covered in the article on data protection in surveys.

Common mistakes when writing a research paper

The four mistakes below turn up in feedback again and again, and none of them is about effort. They appear in places where the paper looks perfectly fine on the surface.

They share one pattern: somewhere the link breaks between what was promised and what the text delivers. Once you know the four places, you can find them in your own work in under an hour.

The research question is not answered in the conclusion

Many graders read the introduction and the conclusion first, and next to each other, because that is where the thread shows. If the introduction announces three questions and the conclusion answers two, it is obvious immediately. The mistake usually happens because the introduction was written early and never revisited. Put both sections side by side at the end and cut whatever the paper did not deliver.

The body summarizes instead of arguing

A body that summarizes one source after another is a collection of material. It becomes scholarly only when it puts positions in relation to one another: where two studies contradict each other, why that is, and what follows for your question. A simple test: delete every sentence that only reports what someone else wrote. What remains is your contribution. If little remains, the argument is missing.

Results are interpreted before they are reported

In empirical work, reporting and interpretation stay separate. First comes what the data show, plainly and without judgment. Then comes what that means and where the limits lie. Mixing the two makes it impossible for the reader to tell your reading of the data from the data themselves. Graders tend to check that separation particularly closely.

Reused work is submitted without asking

Reusing a section from your own earlier paper feels harmless, and under federal regulation it is: section 93.227(b) states plainly that self-plagiarism does not meet the definition of research misconduct. Honor codes are stricter. Stanford requires the professor's prior approval before work is submitted more than once, and many schools follow the same logic. Ask before you reuse, and cite your earlier paper as you would any other source.

Conclusion

Writing a research paper is less a matter of talent than of sequence. Settle the research question before the outline, state your method openly and leave the inconvenient results in, and you already meet the core of what federal regulation and any honor code ask for. Everything else sits in your department handbook, which outranks any guide.

Where to go next

• Want to know which sections come in what order? Structure of a research paper
• Writing a bachelor's thesis? Guide to the bachelor's thesis
• Collecting your own data? Writing an empirical paper
• Looking for the right tenses and phrasing? Academic writing

Does your paper need data of its own?

With empirio.ai, an online survey tool from Germany, you build the questionnaire, send it out and analyze the answers, with hosting in the EU and free use up to 50 responses a month.

Create a survey for your research paper