How to structure a discussion
A discussion mirrors the introduction: it starts narrow, with your finding, and widens to the field.
Results or discussion?
The most common structural problem is material in the wrong section. A simple rule: results say what you found, discussion says what it means.
| Content | Results | Discussion |
|---|---|---|
| Numbers, tables, test statistics | Yes | Only the key figure, briefly |
| Whether a hypothesis was supported | Yes | Yes, as the opening |
| Comparison with other studies | No | Yes |
| Explanations for the findings | No | Yes |
| Limitations | No | Yes |
| New data or analyses | Yes | Never |
A paragraph order that works
Main finding. How it compares with earlier work. Possible explanations, including ones you can't rule out. Limitations and strengths. Implications for practice or theory. What should be studied next.
Some journals want a separate conclusion; others fold it into the last paragraph. Check the format.
Matching claims to design
What you can say depends on what you did. A randomized experiment supports causal language. A cross-sectional survey supports association. A qualitative study supports claims about how and why, within the group studied.
Reviewers read verbs closely. "Leads to", "results in" and "improves" are causal claims, whether or not you meant them that way.
Unexpected and null results
Report them and discuss them. A result that contradicts your hypothesis is often the most interesting thing in the paper. Consider whether the study had enough power to detect the effect before concluding there is none, and say so either way.