The general discussion is where a PhD thesis has to show that its chapters add up to one piece of research. It is also the part most often drafted last, under time pressure, and the part where claims tend to grow. A good general discussion answers the research questions across the chapters, says what the thesis adds as a whole, and keeps every claim within what the chapters showed.
No one sees their own blind spots, with or without AI help. Esmond points them out in your draft from within your field, along with factual errors and inconsistencies.
Norwegian doctoral guidelines are explicit about a thesis made of several papers. In NTNU's summary of them, the candidate “must compile a separate part of the thesis that not only summarizes but also compares the research questions and conclusions presented in the separate pieces of work.” The closing chapter of a monograph has a similar job: it should leave the reader with an answer the single chapters could not give on their own.
In practice, a general discussion does four things. It restates the overall aim in one or two sentences. It answers each research question with the findings from all chapters that bear on it, including the ones that point in different directions. It states the contribution of the thesis as a whole, set against the literature from the introduction. And it sets out what the answer does not cover, so that the implications that follow stay within reach of the evidence.
A hypothetical example of a commentYour general discussion recommends routine screening in primary care, but the three studies before it were done in specialist clinics. Which finding supports the step to primary care?
A first comment on one part of your own draft, right away and without an account.
Get a free previewThe most common weak form is a general discussion that walks through the chapters in order: Chapter 2 found this, Chapter 3 found that. Every sentence is accurate and the reader learns nothing new. Slides from Örebro University in Sweden, prepared for doctoral students working on the introductory chapter of a compilation thesis, list, among the purposes of the introductory text, summarizing the “results as a whole, not only copy-paste”. The table shows the difference.
| Summary per chapter | Synthesis | |
|---|---|---|
| Organized by | Chapter order | Research question or theme |
| Typical sentence | Study 2 showed that X. | Studies 2 and 4 both show X, but only in settings where Y; Study 3 suggests why. |
| Tensions between chapters | Left unmentioned | Named and explained, or left open with a reason |
| Contribution | Repeated per chapter | Stated once, for the thesis as a whole |
Claims usually grow because the general discussion is drafted from memory, with the chapters closed. Three shifts come back across fields.
Copy every claim from your general discussion into a list. Next to each one, note where in the thesis it is shown: chapter, table or section. Then compare the wording. If the chapter says "was associated with" and the discussion says "leads to", one of them has to change, or the discussion needs an argument for the stronger reading. Do the same with the introduction: the questions the general discussion answers should be the ones the introduction asked, in the same words.
Run the same check for limitations. Each limitation named in a chapter should still be visible where you interpret the findings it limits. A limitations section that lists five caveats in five sentences, none of which comes back in the interpretation, tells the committee the caveats did not change anything.
Chapters that disagree are common in a PhD thesis, especially one built from separate studies with different samples or methods. Leaving the disagreement unmentioned is the weak option: a committee member who read both chapters will raise it anyway. Naming it, and offering an explanation or saying why it cannot be resolved with the present data, turns it into part of the contribution.
Null results belong here too. A study that found no effect is a finding, and the discussion should treat it as one, without reading it as proof that there is no effect.
If some chapters were published years ago, the field may have moved on. The University of Oslo allows updates in the summary or in each article; NTNU leaves published articles unchanged apart from minor formal errors and puts further discussion in the summarizing part or a concluding chapter. The general discussion is a natural place to say what newer studies add to or change in your earlier findings. More on that format on the page about a thesis by publication.
This example is made up. A candidate in nursing science has four studies on a screening questionnaire, all carried out in specialist clinics. The general discussion, drafted in the last two months, opens with a page per study, then recommends that general practices adopt the questionnaire. On rereading, the candidate restructures the discussion around the two research questions, adds a paragraph on why results from specialist clinics may not carry over to general practice, and keeps the recommendation as a question for further study.
Points that follow from the sections above.
Esmond evaluates every assignment before you submit. It never writes it.
Esmond is an AI tool that doesn't write for you. If your course doesn't allow any AI tools, don't use Esmond for that work. If you're unsure, ask your instructor or advisor.
Esmond reads the general discussion from within your field and annotates passages in that chapter where something raises a question: a claim stated more strongly than the evidence the chapter itself reports, a recommendation the chapter does not trace to a finding, a term that shifts meaning within the chapter. Upload the closing chapter and it is read as a part in its own right: whether its claims are carried by what it reports, and whether its terms keep their meaning. Within the part you upload, it points out where a claim goes further than the evidence in that part supports.
It adds no text, and whether the thesis holds together as a doctorate is for your supervisor and committee to judge. Start from PhD thesis feedback for the other parts.
Names differ by field and country. In many theses the general discussion and the conclusion form one closing part. What matters is the function: answering the research questions across the chapters and stating what the thesis adds.
There is no general figure. Your program or faculty may give guidance; check it there. Length follows from the number of research questions and how much the chapters need to be brought together.
Only those that limit the overall answer, and only where they change what you can claim. A list that repeats every chapter's limitations adds length without adding a point.
Yes, where it helps to interpret your findings together or brings an earlier chapter up to date. Keep the main positioning in the introduction.
Read on November 7, 2026.
Get your paper evaluated before you submit