Lead with the finding that matters
Not with a summary of everything you found. Open on the result that answers your question, say what it means in terms a reader in your field actually cares about, and only then move outward toward context and implication.
A discussion recapitulating every finding in the order the results section presented them is simply a second results section, and it reads as though the interpretive work never happened at all. One constraint fixes most of this: never begin a paragraph with a number.
Place it against the literature, including where it disagrees
Where your finding matches earlier work, say so and say what yours adds beyond confirmation. Where it does not match, that is the most interesting paragraph available to you, and the job is to offer plausible reasons — a different population, a different measure, a different setting — rather than to explain the disagreement away.
A discussion finding only agreement usually means the literature review was selective, and any reader who knows the field will notice which studies have gone missing. Engaging with the study that contradicts you is one of the clearest signals of a strong candidate.
Limitations are structural, and then you stop overclaiming
Name what the design genuinely cannot support and why, in terms of what it means for your claims. A cross-sectional design cannot establish sequence. A single site limits transferability. A self-report measure carries known and nameable biases.
What weakens a paper is not having limitations but listing generic ones nobody considered, or naming them carefully and then overclaiming in the conclusion regardless. End somewhere actionable: a practice doctorate wants a recommendation somebody could implement, a research degree wants the next question.
The other frequent fault is length in the wrong place. A discussion spending three pages restating the field and half a page on your own contribution has its proportions backwards, and a reader who reaches the end still waiting for your interpretation will mark it accordingly. Everything that is not your reading of your own results is scaffolding for the part that is.
Three steps, then the record keeps itself.
Send the results and the review
Both together, because the discussion is a relationship between them rather than a section that stands alone.
Get it interpreted
Led by the finding that answers your question and placed against the literature, including the work that disagrees with you.
Check every claim
Nothing in the conclusion the design cannot support, which is where committees push hardest and where careful papers still lose marks.
Queries on record.
How do I stop repeating the results?
Forbid numbers from the opening sentence of every paragraph. If a paragraph starts by restating a figure, it is a results paragraph wearing a different hat. Begin instead with what the figure means and bring the number in behind it as support. That single rule fixes most of the problem in one editing pass.
What if my results were not significant?
Report and interpret them as findings, because that is what they are. A well-designed study finding no effect has told the field something real, and treating it apologetically damages the paper far more than the result does. What matters is whether the study was powered to detect an effect worth detecting, and that belongs here explicitly.
How many limitations should I include?
The ones that genuinely bear on your claims, usually three or four, each described in terms of its consequence rather than listed as a formality. Generic limitations applying to any study of that type add nothing at all. The limitation worth writing is the one a reader could not have guessed from your methods section.
Can I recommend a change to practice?
Where the design supports it, and be precise about how far that reaches. A single-site improvement project can recommend something for that setting and for closely similar ones with appropriate caution; it cannot recommend a national standard. Overreaching in the final paragraph after a careful limitations section is a common and entirely avoidable loss.
What will a committee ask about this section?
Usually whether a claim exceeds its evidence, and whether you have engaged with the study that disagrees with you. Both can be answered in advance by writing the section so those points are already addressed, which is a large part of why a mock defense before the real one is worth the afternoon.
Elsewhere in the hall.
Send the assignment and the deadline.
A specialist credentialed in your own field reads what you actually have and replies within two hours with a plan and a price. The reading costs nothing, and if the honest answer is that we are not the right hall for it, you will get that instead.
