How to Write a Thesis Discussion Chapter: Structure, Examples, and Common Mistakes 2026

The discussion chapter is where your thesis transforms from a data report into a scholarly argument. It is the section that most clearly demonstrates your intellectual contribution — your ability to interpret evidence, engage critically with theory, and place your findings in the context of existing knowledge. It is also the section that students most frequently write badly, either by simply repeating their findings or by making interpretive leaps their data cannot support. This guide shows you exactly how to structure a discussion chapter that impresses examiners and makes a genuine academic contribution.

Discussion Chapter Structure

The standard discussion chapter structure follows this sequence:

1. Opening summary (1–2 paragraphs): Brief restatement of the research question and the core findings. This anchors the reader before interpretation begins.
2. Discussion by research question (main body): For each research question, interpret the relevant findings in light of existing literature. This is the largest section.
3. Unexpected or contradictory findings: Address any results that did not match your hypotheses or the existing literature. Offer reasoned explanations.
4. Limitations: Acknowledge the constraints on your study honestly and specifically.
5. Implications: What do your findings mean for theory, practice, or policy?
6. Future research directions: What questions does your study raise that could be addressed by subsequent research?
7. Conclusion of the chapter: Brief synthesis of the chapter’s key interpretive points, leading into the overall conclusion chapter.

How to Relate Your Findings to the Existing Literature

This is the core intellectual task of the discussion chapter. Every significant finding should be connected to at least one published study. There are three possible relationships:

Agreement: Your finding aligns with existing research.
“The significant positive relationship between supervisor contact frequency and thesis completion rate (r = .43, p < .001) is consistent with Wellington’s (2024) systematic review, which found fortnightly supervision to be the single strongest predictor of research degree completion.”

Extension: Your finding adds nuance or extends existing knowledge.
“While García (2024) demonstrated this relationship in undergraduate students, the present study extends this finding to the postgraduate research context, suggesting the effect may persist — or strengthen — at higher levels of academic autonomy.”

Contradiction: Your finding differs from existing research.
“In contrast to Smith and Brown’s (2023) finding of a negative relationship between AI tool use and originality scores, the present study found no significant relationship (r = .09, p = .41). This discrepancy may reflect differences in the measurement of ‘originality’ — Smith and Brown used a single-item self-report measure, while the present study used examiner-rated rubric scores…”

The citation density in your discussion should be comparable to your literature review. A discussion chapter without literature engagement is not a discussion — it is a repetition of your findings chapter. For citation format guidance, see our Harvard referencing guide or the APA citation guide.

Handling Unexpected or Contradictory Findings

Examiners particularly value how a student handles unexpected findings — it shows intellectual honesty and genuine analytical ability. The steps for addressing an unexpected result:

1. State the unexpected finding clearly: “Contrary to the hypothesised direction, participants in the AI tool condition reported higher anxiety (M = 4.1) than control participants (M = 3.8).”
2. Acknowledge it is unexpected: “This finding was not anticipated given the literature suggesting AI tools reduce cognitive load…”
3. Offer at least two alternative explanations: “This may reflect the novelty effect — participants unfamiliar with AI tools experiencing anxiety about using them correctly. Alternatively, it may suggest that AI tools, by making gaps in knowledge more visible, temporarily increase rather than decrease academic anxiety.”
4. Acknowledge the limits of your explanation: “The present study cannot definitively distinguish between these interpretations; further research with experimental manipulation would be required.”

Writing the Limitations Section

Every thesis has limitations. The limitations section is not a place to apologise for your research — it is a place to demonstrate that you understand the boundaries of your claims. Common limitation types:

• Sampling limitations: “The sample was recruited via social media and may overrepresent digitally active students, limiting generalisability to less digitally engaged populations.”
• Measurement limitations: “Self-reported AI use may be subject to social desirability bias, with participants potentially underreporting use they perceive as academically inappropriate.”
• Design limitations: “The cross-sectional design prevents causal inference — it is not possible to determine whether AI use causes higher anxiety or whether higher-anxiety students select AI tools.”
• Context limitations: “The study was conducted entirely at UK Russell Group universities; findings may not generalise to different institutional contexts or national higher education systems.”

The key principle: state the limitation, then explain why it does not entirely undermine your conclusions. “While the sample size (n = 67) is relatively modest, it exceeds the minimum threshold recommended by G*Power analysis for detecting the medium effect size hypothesised a priori (n = 52).”

Implications and Future Research Directions

The implications section addresses “so what?” — why your findings matter beyond academic knowledge. Consider three levels:

• Theoretical implications: Does your study support, challenge, or extend a theoretical framework?
• Practical implications: What should practitioners do differently based on your findings?
• Policy implications: What should institutions, funders, or policymakers consider?

For future research, be specific: “Future studies should employ a longitudinal design to track anxiety over the full thesis-writing period (rather than a single time point), include a pre-test measure of baseline anxiety, and use behavioural measures of AI tool use (usage logs) rather than self-report.”

7 Discussion Chapter Mistakes That Examiners Flag

1. Repeating the findings chapter — summarise briefly, then interpret. Do not redescribe your data.
2. Making claims beyond the data — “this proves that…” Be careful with causal language unless you have experimental data.
3. Ignoring contradictions with the literature — examiners look for engagement with disconfirming evidence.
4. Listing limitations without explaining why they do not undermine your conclusions
5. No engagement with the theoretical framework from your literature review — your theory should appear in discussion, not just literature review.
6. Vague implications — “this has implications for educators” is not enough. What specifically should educators do?
7. No transition to the conclusion chapter — the discussion should lead naturally into your overall conclusions.

For the complete thesis chapter sequence, see the complete thesis writing guide and the preceding chapter in sequence, how to write a findings chapter.

If you are working through your discussion and struggling to connect your findings to the literature coherently, Tesify provides AI-assisted academic writing support that can help you articulate interpretations and theoretical connections clearly and precisely.

Frequently Asked Questions