What should the discussion section of a research paper include?

The principal findings, how they compare with earlier studies, possible explanations, what they mean for practice, policy or theory, the study’s limitations, and unanswered questions. Many journals also want a short closing message. Follow your target journal’s instructions where they differ.

How do I start a discussion section?

With the main finding, in one sentence, tied to the aim you stated in the introduction: “We found that…”. Do not restate the whole Results section; give the one result the rest of the discussion depends on.

What is the difference between the discussion and the conclusion?

The discussion interprets the findings, compares them with other work and weighs their limits. The conclusion is shorter: it states what follows from them. Some journals combine the two. See the Conclusion Paragraph Generator for a standalone conclusion.

How long should a discussion section be?

Often about as long as the introduction, but requirements vary by journal and discipline. A thesis chapter will be longer than a journal article. Enter your target length and the tool divides it among the parts.

Where do limitations go?

Within the discussion, usually just before the closing message or future research. STROBE asks authors to address limitations, including potential bias and imprecision, and to discuss the direction and magnitude of any bias. Name each limitation and say how it affects what you can claim.

Can I say my results prove my hypothesis?

It is safer to say the results support it or are consistent with it. A single study cannot rule out every alternative explanation, and the wording signals that you know it. The generator’s checklist flags a short list of proof words so you can reconsider them.

Discussion Section Generator

Enter your principal findings, the earlier studies they should be compared with, and your explanations, implications, limitations and open questions. The generator arranges them in the order journals teach for a discussion, adds a word budget if you give a length, and shows a sentence frame wherever a part is missing. Honest templates, no AI. Free, no sign-up.

Your study

Working title

Target length (words, optional)

Adds a word target to each part of the discussion.

Aim or question of the study

Type of study

Findings and earlier research

Principal findings — one per line

Earlier studies — one per line, “Study: how yours compares”

What might explain the findings?

Meaning and limits

Implications for practice or policy

Implications for theory

Limitations — one per line

Unanswered questions and future research — one per line

Closing message

Checklist

Words written

Findings

Earlier studies

Limitations

To do: State the principal findings first.
To do: Compare your findings with earlier studies.
To do: Offer an explanation, and a rival one.
To do: Say what the findings mean for practice, policy or theory.
To do: Name the limitations. Readers care about them, and an unmentioned flaw costs trust.
To do: Say what is still unanswered.
To do: Close with a message the reader can carry away.
Done: No words that assert proof.

Your discussion outline

10 parts still to write

Discussion

Discussion section outline

To write: Restate the aim or question of the study in one sentence.

1. Principal findings

Tip: Open with the main result, in a sentence, without repeating the Results section.

To write: State the principal finding in one sentence: “We found that [result], [with its size or direction].”

2. Comparison with earlier research

Tip: Set your results against earlier studies and spend the most time where they differ. Explain differences by sample, method, measure or setting.

To write: For each relevant earlier study: “[Author, year] found [result]; ours is [consistent with / larger than / different from] it.”

To write: Where your results differ from earlier work, say why: a different sample, method, measure or setting.

3. Interpretation

Tip: Offer possible explanations and a rival one. Where the design cannot show cause, write “suggests” or “is consistent with”.

To write: Offer an explanation: “One possible explanation is [mechanism]. An alternative is [rival explanation], which our design [can / cannot] rule out.”

4. Implications

Tip: Say what the findings mean for practice, policy and theory, and go no further than the evidence does.

For practice or policy

To write: Say what practitioners or policymakers might do differently, and how far the evidence supports it.

For theory

To write: Say how the findings support, qualify or challenge a theory or earlier model.

5. Limitations

Tip: Name the limits of design, sample and measures, and say which way each could bias the result and by how much (STROBE asks for direction and magnitude of bias). Say how far the results apply beyond your sample.

To write: List the main limitations of design, sample and measures, with the likely direction of bias: “Because [feature], [result] may be [over/under]estimated.”

6. Unanswered questions and future research

Tip: Turn the limits and open questions into specific studies someone could run.

To write: Name specific studies the work points to: “Future research should test whether [question], using [design].”

7. Closing message

Tip: End with a message the reader can carry away, and the caveat it needs.

To write: Close with the take-home message and its main caveat.

How to Use the Discussion Section Generator

1. State the study. Give the aim and choose quantitative, qualitative or other. The type changes the advice on limitations: bias and generalizability for quantitative work, transferability for qualitative work.
2. List your findings and the earlier studies. One principal finding per line, then one earlier study per line, written as “Author, year: how yours compares”. They become a numbered list and a comparison table.
3. Add your reading of them. Possible explanations, what the findings mean for practice and for theory, the limitations, and the questions still open.
4. Fix what the checklist flags, then copy. The checklist tells you which parts are empty and flags words that assert proof. Replace each remaining frame with your own prose.

The Order of a Discussion

Michael Docherty and Richard Smith argued in a 1999 BMJ editorial that discussions often read as sales pitches, and proposed a fixed order of five parts: the principal findings; the strengths and weaknesses of the study; the strengths and weaknesses in relation to other studies, particularly any differences in results; the meaning of the study, including possible mechanisms and implications for clinicians or policymakers; and unanswered questions and future research. Reporting guidelines such as STROBE ask for the same ground in their discussion items: key results, limitations (including the direction and magnitude of potential bias), a cautious interpretation, and generalizability. The generator follows this order and ends with a closing message.

Part	Job	Planning share
1. Principal findings	State the main result in a sentence, without repeating Results	10%
2. Comparison with earlier research	Place your results against other studies and explain differences	30%
3. Interpretation	Offer explanations and a rival one, cautiously	15%
4. Implications	What it means for practice, policy and theory	15%
5. Limitations	Name them, and say which way each could bias the result	15%
6. Unanswered questions	Turn gaps into studies someone could run	10%
7. Closing message	One takeaway, with its caveat	5%

The shares are this tool’s planning convention, not part of any guideline. Journals set their own headings and limits, and some merge the discussion with the conclusion.

Worked Example: A 1,500-Word Budget

“Try an example” loads an illustrative discussion of a study of peer tutoring and re-enrollment. The effect sizes and earlier studies are bracketed placeholders, since they belong to your own research. If the example’s sections stood at 1,500 words, the tool would target the following, rounded to the nearest 10 words with the rounding balanced to add up to the total.

Part	Target words
1. Principal findings	150
2. Comparison with earlier research	450
3. Interpretation	230
4. Implications	230
5. Limitations	220
6. Unanswered questions	150
7. Closing message	70
Total	1,500

Common Mistakes in a Discussion

Repeating the Results. Restate the main finding once, then move on to what it means. New numbers belong in Results.
Claiming more than the design allows. Observational data can suggest and be consistent with; they rarely prove. The checklist flags words such as “prove” and “undoubtedly”.
Hiding the limitations. Docherty and Smith note that readers care most about weaknesses, because every study has them; an unmentioned flaw can cost the paper its credibility.
Ignoring results that disagree with earlier work. Differences are the interesting part of the comparison. Explain them by sample, method, measure or setting.
Introducing new results. A discussion interprets what was reported. If a finding matters, it belongs in the results.

Frequently Asked Questions

What should the discussion section of a research paper include?: The principal findings, how they compare with earlier studies, possible explanations, what they mean for practice, policy or theory, the study’s limitations, and unanswered questions. Many journals also want a short closing message. Follow your target journal’s instructions where they differ.
How do I start a discussion section?: With the main finding, in one sentence, tied to the aim you stated in the introduction: “We found that…”. Do not restate the whole Results section; give the one result the rest of the discussion depends on.
What is the difference between the discussion and the conclusion?: The discussion interprets the findings, compares them with other work and weighs their limits. The conclusion is shorter: it states what follows from them. Some journals combine the two. See the Conclusion Paragraph Generator for a standalone conclusion.
How long should a discussion section be?: Often about as long as the introduction, but requirements vary by journal and discipline. A thesis chapter will be longer than a journal article. Enter your target length and the tool divides it among the parts.
Where do limitations go?: Within the discussion, usually just before the closing message or future research. STROBE asks authors to address limitations, including potential bias and imprecision, and to discuss the direction and magnitude of any bias. Name each limitation and say how it affects what you can claim.
Can I say my results prove my hypothesis?: It is safer to say the results support it or are consistent with it. A single study cannot rule out every alternative explanation, and the wording signals that you know it. The generator’s checklist flags a short list of proof words so you can reconsider them.

More in Essay & Paper Writing

See all Essay & Paper Writing