Your analysis is finished, your tables are formatted, and the document has stopped moving. Chapter 4 took three weeks; Chapter 5 has taken three weeks to produce two paragraphs. This is the most common stall in the whole thesis, and the reason is structural rather than intellectual: every other chapter has a template you can follow, and the discussion appears not to. It does. Below is the shape, the paragraph pattern that fills it, and the specific handling for the two situations that cause the paralysis — a finding that contradicts the literature, and a limitations section you are afraid to write.
The cost of this stall is not abstract. Weeks lost here fall at the end of a candidature, against a fellowship end date and a re-registration deadline, which is why a chapter that is conceptually easy becomes the one that costs a semester.
Why the chapter feels impossible
Because you have been forbidden to interpret for four chapters and are now required to do nothing else. Everything you have written so far has been disciplined by an external standard: the literature review reports what others found, the methodology reports what you did, the results report what happened. The discussion asks what it means — and there is no output window that produces that.
The unlock is recognising that the discussion is not free-form prose. It answers a fixed set of questions in a fixed order, and once you see the order, most of the chapter is assembly.
The five-part structure
- Restatement of purpose and headline findings. One short section: what the study set out to do and, in two or three sentences, what it found. No new numbers, no citations.
- Discussion of each finding. The body of the chapter, organised by objective or hypothesis in the same order as Chapter 1 and Chapter 4. One subsection per finding.
- Implications. Separated into theoretical implications, what your findings mean for the models and frameworks in your field, and practical implications, what a practitioner or policymaker should do differently. Keep the two apart; merging them is how vague recommendations get written.
- Limitations. Honest, specific and consequential.
- Directions for future research. Derived from the limitations and from the questions your findings opened, never a generic list.
If your department requires a combined “Results and Discussion” chapter, the same five parts apply, interleaved after each result. Your institute’s format manual decides; check it before you restructure anything.
The four-move paragraph pattern
This is the engine of the chapter. For each finding, write four moves in order. Do them for every result and the body of the chapter is written.
- Move 1 — Restate the finding in words, not numbers. “Perceived supervisory support was the strongest predictor of research self-efficacy.” Not the beta, not the p-value; those live in Chapter 4 and repeating them here is the most common padding in the chapter.
- Move 2 — Locate it against the literature. Does it agree with prior work, extend it, or contradict it? Name the studies and be specific about the relationship: “consistent with”, “extends to a context where”, “in contrast to”. This is where your literature review earns its keep, and where a properly built synthesis matrix from our guide to writing a literature review for a PhD thesis in India saves you days, because the comparison you need was already tabulated.
- Move 3 — Explain the mechanism. Why might this be so? Offer a reasoned account, grounded in theory or in your context, and label it as interpretation rather than finding. This is the move that most drafts skip entirely, and its absence is what makes a discussion read as a summary.
- Move 4 — Qualify. What does this finding not establish? One sentence bounding the claim. It costs you nothing and it removes the examiner’s most obvious line of attack.
Four sentences per move, four moves, six findings: that is a substantial chapter, and none of it required inspiration.
When your finding contradicts the literature
Scholars freeze here, or worse, soften the result until it appears to agree. A contradiction is not a problem with your thesis — handled properly it is the most interesting thing in it.
Work through four explanations in order, and say which you find most plausible. Context: the prior studies were conducted in a different setting, and yours is the first in this one — the most common and most legitimate explanation for an Indian study departing from a literature dominated by other contexts. Measurement: you operationalised the construct differently, or used a different instrument. Sample: a different population, a different sector, a different level of seniority. Method: a different design, a different analysis, or a difference in statistical power.
Then state what you conclude. “The divergence is most plausibly attributable to context, since the construct as measured here reflects an institutional arrangement absent from the settings of the earlier studies” is a strong, defensible sentence. What you must not write is “this finding differs from prior studies, which may be due to various factors” — a sentence that announces you did not think about it.
Non-significant results are findings too
A hypothesis that was not supported gets the same four moves as one that was. What it does not get is the phrase “there was no effect”: a non-significant result means you did not detect an effect, which is a different claim, and the distinction matters most when your study was modestly powered. If your sample was tight — the calculation behind that judgement is set out in our guide to calculating sample size for a thesis — say so here, because an underpowered null result is a limitation with a name rather than an absence of a result.
Limitations that strengthen rather than apologise
Two failure modes dominate. The generic list — “small sample, limited generalisability, time constraints” — which says nothing and protects nothing. And the confessional, in which the scholar apologises for the study they conducted.
A good limitation has three parts: what the constraint is, what it means for your claims, and what it does not undermine. “The cross-sectional design means the relationships reported here are associations rather than causal effects; the direction of influence between support and self-efficacy cannot be established from these data. The descriptive findings on prevalence are unaffected.” That is bounded, specific and finished.
Cover the categories that actually apply: design, sampling and access, measurement, and any assumption departure you had to work around. If, for instance, you switched to a non-parametric test after an assumption failure — the routes are set out in our recovery guide for when data fails the normality test — the consequent shift in what you tested belongs here in one sentence, not buried in Chapter 4.
The three sentences that must not appear
“Further research is needed” with nothing after it. “The results were as expected” — if they were entirely expected, the reader wonders what the study contributed. And any causal verb attached to correlational data: caused, led to, resulted in. Cross-sectional survey work supports “was associated with”, and an examiner who finds a causal verb in your discussion will spend ten minutes of your viva on it.
What the examiners will test here
The discussion is the chapter examiners read most closely, because it is where your judgement is visible and where a thesis either makes a contribution or merely reports data. Expect to be asked what your single most important finding is, why one specific result diverged from the literature, and what you would do differently. All three are answerable if you have done the four moves honestly. The dynamics of that session, and why the questions come as specifics rather than generalities, are described in our guide to what happens in a PhD viva voce in India.
Getting the chapter out of your head and onto the page
The discussion is almost always already formed in your mind — you have been explaining these findings to your guide and your lab mates for months. What is missing is the structure that turns those explanations into a chapter, and the discipline of keeping every comparison attached to the paper it came from while the argument grows.
Tesify builds your chapters from your own synthesis and keeps each citation bound to a real source as the document develops, so the argument you already have becomes a draft instead of a blank page. Every interpretation, and the responsibility for it, remains yours.
Start writing your discussion chapter in Tesify
Frequently asked questions
What is the difference between the results and discussion chapters?
Results reports what you found; discussion explains what it means, how it relates to prior work and what follows from it. Numbers belong in results and interpretation belongs in discussion.
How long should the discussion chapter be?
Length norms are set by your institute’s format manual, but as a working guide it is usually comparable to the results chapter. Completeness matters more: every finding discussed, every objective answered.
Should I repeat the statistics in the discussion?
No. Restate findings in words. Repeating beta values and p-values is padding and it makes the chapter read as a second results section.
Can I cite new literature in the discussion?
Yes, and you often must — particularly when explaining an unexpected finding that your original review had no reason to cover. If a source becomes central, consider adding it to the literature review too.
What do I do if my results contradict prior studies?
Work through context, measurement, sample and method as candidate explanations, then state which you find most plausible and why. A well-explained contradiction is often the most valuable part of a thesis.
How do I discuss a hypothesis that was not supported?
Exactly as you would a supported one, using the same four moves. Say you did not detect an effect rather than that there was no effect, and address power if your sample was modest.
What should go in the limitations section?
Design, sampling and access, measurement, and any assumption departures — each with what it means for your claims and what it does not undermine. Avoid generic lists and avoid apologising.
Are limitations a weakness in the eyes of examiners?
The opposite. An unstated limitation that the examiner finds is a weakness; a stated one demonstrates that you understand the boundaries of your own evidence.
How do I write future research directions?
Derive them from your limitations and from the questions your findings opened, and make each one specific enough that a reader could actually design the study. “Further research is needed” is not a direction.
Should implications be theoretical or practical?
Both, in separate subsections. Theoretical implications address the models in your field; practical implications tell a named audience what to do differently.
Can I use first person in the discussion chapter?
Conventions vary by discipline and by department, and your format manual or supervisor decides. Whichever you choose, be consistent across the whole thesis.
Where should I start if the page is completely blank?
Take your least complicated finding and write the four moves for it in four short paragraphs, without editing. The chapter has a first section, and the pattern is now visible for the rest.