Thesis Writing Rules: A Complete Checklist

Most of the mistakes found in theses do not come from a lack of talent; they come from a lack of order. A chapter started too late, a source quoted half-read, raw data never kept in order, a citation style that changes halfway through — none of these are failures of thinking, only failures of organisation, and disorder can be treated with a list and a sequence.

This article is a working checklist: from before you write to the day of the defence. Each section sets out one clear rule, and at the end there is a list you can keep beside your desk. If you are just starting out, writing a thesis with artificial intelligence: a complete and honest guide explains the bigger picture and where tool-assisted help ends.

Before you write: proposal, problem, ethics

A thesis begins before the first sentence is written. The proposal has to make clear what the research problem is, why it matters, and which question it is going to answer. If you cannot state your problem in one clear sentence, none of the chapters will take a proper shape either, because each chapter is really part of the answer to that one sentence.

At the same stage, settle the ethical side of the research. If your work involves human participants or personal data, informed consent, how data will be stored, and how respondents will be anonymised all have to be decided before data collection, not after. Changing those decisions halfway through is both difficult and costly for the credibility of your work.

A third point at this stage concerns the expectations of your degree level: what is expected at bachelor level, in a master's thesis and in a doctoral dissertation is not the same, and that expectation is the standard your work will be judged by. See thesis writing at different degree levels for those differences, and agree the scope of your work with your supervisor before you start.

Structure: chapters and how they relate to IMRAD

The conventional thesis structure is broadly stable across disciplines: introduction and general framing, literature review, research method, findings, and discussion with conclusion. What turns those parts into a single argument is the logical chain between them: the problem built in the introduction becomes a testable question in the method, receives its answer in the findings, and gains meaning in the discussion.

The same logic in its compressed form is the structure of a scientific paper, known as IMRAD: introduction, methods, results and discussion. The relation between the two is straightforward: a thesis executes the same logic with more detail and a wider review of the literature. If you understand the IMRAD logic, chapter structure stops feeling like administrative paperwork. We set out that structure step by step in writing a scientific paper and the IMRAD structure.

One structural rule: before writing a chapter, draft the list of its subheadings; that prevents repetition and shows you clearly which parts of the work are still empty.

Sources and citation: where honesty is measured

Choose your citation style at the start and do not change it; a style that shifts mid-text is among the most common signs of disorder. Every idea, every dataset and every quoted sentence needs an in-text citation, and the reference list at the end must match the text: no source left uncited, no citation without a source.

The hard rule of this section is simple: never cite a source you have not read. This is exactly where working with AI tools becomes dangerous, because a tool can easily produce a reference that does not exist. Fabricating a source is always a violation and no practice permits it; even if your text is otherwise flawless, one invented reference can damage the whole work. So find and read every reference yourself before the text is final.

To be safe, take your own notes on each source from the beginning: such a note is an honest summary of the source and, while you write, a guard against accidentally copying the other author's phrasing.

Research ethics and data

Data honesty is the red line of research work. Every number in the text has to come from a real measurement or observation, and the path you took to that number has to be retraceable. Keep the raw data, make dated versions, and record every change you applied to it — removing incomplete records, transforming a variable, choosing a test. That record is stronger at a defence than any spoken explanation.

The second rule is to write conclusions in proportion to the data rather than to the ambition of the project, and to state the limitations of the work yourself, plainly. If your data shows a weak relationship, writing about a definite effect is not only a scientific error; it also leaves the work exposed to reviewers' questions. Claiming more than the evidence supports is a small fabrication whose size becomes obvious in review.

Writing and language: keep the text your own

Consistency of tone matters across the whole text. If some chapters read as formal and others as conversational, the reader notices the disorder of the language before reaching the content. So before you submit, take one full editing pass: check spacing and Persian punctuation conventions, standardise terminology, and cut the filler sentences.

The main rule of this section, though, is about ownership: every sentence that stays in the thesis should be a sentence you understand and can defend. That standard is stricter than any detection system and it is also what carries you through the defence. If you used a supporting tool for part of the text, state its role in a suitable place in the preface or an appendix.

Before submission and defence

In the final days, shift your routine from writing to reviewing. Registering with Irandoc and running the similarity check before the defence is a mandatory procedure of graduate study, and the "Hamandjoo" system shows you the matching passages, the degree of similarity and the source of each match. Read that report carefully and mark the passages that need rewriting or a proper citation.

Prepare the final version in the format your university requires, and read the whole thing once on paper; some errors are visible only there. Prepare for the main defence questions too: why this topic, what your own contribution is, and what the limitations of the work are.

The practical checklist

This list is not a substitute for advice; it is a list of things that genuinely happen along the road to a thesis. Read it once before you start, and go through it point by point once more before you submit.

• State the research problem in one clear sentence and repeat that sentence in the introduction and the conclusion.
• Write the research question so that it can actually be answered with your method and available data.
• Agree the scope and expectations of the work with your supervisor in your first meeting.
• Decide the ethical requirements and informed consent before collecting data.
• Choose the citation style at the start and keep it unchanged.
• Take your own reading notes for every source.
• Never leave a citation to a source you have not read.
• Draft the subheadings of each chapter before writing it.
• Keep findings separate from discussion so data and interpretation do not blur.
• Keep raw data and the path of your analysis in dated versions.
• Give every table and figure a number, a caption and an in-text reference.
• State the limitations of the research plainly in the discussion.
• Write conclusions in proportion to the data, no further.
• Make tone, vocabulary and terminology consistent across the whole text.
• Do one full pass for spacing, punctuation and Persian typographic conventions.
• Cut filler sentences and decorative description.
• State the role of any supporting tool in the preface or an appendix.
• Complete Irandoc registration and the similarity check before the defence.
• Reconcile the reference list with the text so no citation is left unsourced.
• Prepare practised answers to the main defence questions, and keep a backup copy of your text and data.

You do not have to move all of these forward at once; it is enough that each point covers one specific part of your work, so that the final weeks are not spent fixing structure.

Conclusion

None of these points builds a thesis on its own, but together they keep your work defensible. Order in your sources, honesty in your data, consistency in your language and transparency about the tools you used are exactly what speaks for you in the defence room.

If you want to move through these steps with a tool alongside you, start from writing a scientific paper and the IMRAD structure to get the overall shape clear. The rest of the articles in this cluster each open up one part of the same road.