16 September 2026 · 7 min read
The conclusion is written last, at the point where everyone is tired, and it reads that way in most submitted theses. It is also, along with the abstract, the part an examiner reads first when deciding what kind of document they are holding. The gap between those two facts is free marks sitting on the table.
A summary repeats what each chapter said. A conclusion answers the question you asked in chapter one, in order, with the evidence behind each answer stated in a line. If your research had four objectives, an examiner should be able to find four answers, and should not have to hunt.
That last rule is the one most often broken. A new statistic or a new author appearing in the final chapter tells an examiner the thesis was assembled rather than argued, and it invites exactly the question you do not want at viva: why is this here and not in chapter four?
Nepali theses are full of recommendations that no reader could act on — “the government should formulate appropriate policies”, “awareness should be increased”. They cost nothing to write and earn nothing. A usable recommendation names who should do what, and follows from a finding in your own data rather than from general opinion.
Recommendations for further research are not a formality either. “Future studies may use a larger sample” is what everyone writes. What examiners notice is a student who names the specific question their own design could not answer and explains what design would answer it.
State the real constraints — sample confined to one district, self-reported data, a cross-sectional design that cannot establish causation, a response rate that skewed towards one group. Then say what each limitation does and does not undermine. A limitation acknowledged and bounded strengthens a thesis; the same limitation discovered by an examiner in the viva weakens it badly.
Do not over-confess either. A page of self-criticism reads as a student arguing against their own work, and there is no need to volunteer that the whole study was underpowered if it was not. Two or three genuine limitations, each with a sentence on its consequence, is the right weight.
Once the conclusion exists, the abstract almost writes itself: problem, method, key findings with the numbers, conclusion, one implication. Write it last, keep it inside your department's word limit, and make sure every number in it matches the results chapter. Mismatched figures between abstract and results are caught more often than students expect, because both are read closely.
Read chapter one and the final chapter back to back, with nothing in between. The objectives stated at the start and the answers given at the end should line up in number, in order and in wording. Theses drift over months of writing — an objective gets rephrased in chapter three and never updated in chapter one — and this single read catches it.
Then check the conclusion against a similarity report. It is a short chapter that reuses your own earlier wording heavily, and text carried over from your approved proposal is the one thing likely to flag here. Self-matching is treated seriously by some departments, so it is worth seeing before submission rather than after.
Related:
If your last chapter currently summarises rather than concludes, send it with your objectives list and we will show you the difference on your own text.