Submit Manuscript

Easy Online Form

Get Newsletter

Sign Up Today

How To Write A Discussion Section For A Scientific Paper


The discussion section of a scientific paper is probably the hardest section to write. Authors must highlight their findings but not overstate their relevance. They must introduce other studies for comparison and give their own results context. In many cases, authors also must highlight the limitations of their work and then give readers a conclusion to take away. The process is complicated and different for every field and even journal by journal.


What to include in a discussion section

Your best bet is to look at your target journal and see what discussion sections from recent papers look like. Do the authors have a lot of room to write, speculate, or even give a kind of literature review? Or does the discussion follow an obvious structure, beginning with the main findings, offering two or three paragraphs of context and comparison with previous studies, and closing with limitations and conclusions?

What you find in these papers will help guide your own decisions about how to structure your discussion. Once you have a good idea of how to map out this section, you’ll need to decide what to put in it.


How to write a discussion section

You want to highlight what you view as your most substantial and important finding or findings at the beginning of your discussion, regardless of what else you write about. The first paragraph of that section should include what you most want readers to take away from your results. That might be the easiest part to write.

Once you have that information summarized, you need to decide which information comes next. If your target journal has an obvious structure for discussions, that format will help you decide. But if discussion sections in your target journal are more free-form, your best option might be to follow the same order of presentation that you used in your methods and results. If you echo the same order, your reader might find it easier to follow this last section of your paper.

Regardless of what you choose to do, you will have to draw in findings from other studies and compare and contrast your results with theirs. When selecting studies for comparison, you should choose those that are most like yours in as many ways as possible, and studies that are as recent as possible.

You do not need to give every detail from these previous studies to present your comparison. Simply state their main finding—e.g., “In Lopez et al. (2015), patients with liver cancer fared better after radiotherapy than patients who did not receive the treatment, with an OS of 26 months vs 10 months.” You do not need to overwhelm your readers with details from these earlier studies. If they are interested, they can consult these reports themselves.

Likewise, you do not need to repeat detail from your study in the discussion section. The discussion section is not the right place to reproduce a lot of data and P values that you’ve already given in your results section.

In addition, most discussions should not include references to tables and figures that already were presented in the results section. It is acceptable, usually, to present a new figure in the discussion only if it describes a hypothesized pathway or other complex idea that you’re introducing based on your findings.


Major pitfalls in writing a discussion section

One major pitfall to avoid in your discussion section is introducing new results. All findings from your study must be mentioned and presented in your results section before you bring them up in your discussion. Even for a minor finding, if you have not introduced it and presented the data in the results, you should not mention it in your discussion.

Another pitfall in writing a discussion section is writing too much. Usually, this section is not the place for a complete literature review. You also do not need to compare and contrast your findings to every other study that ever asked a similar research question. Confine your context to the most recent publications and most relevant work to yours.

Finally, do not overinterpret your results. Reviewers easily detect speculation or interpretations that go too far. Present inferences that are reasonable to make from your data, and err on the conservative side.


How to write the final elements of a discussion section

Most discussions will end with two elements. One is the limitations section, where authors mention potential issues with study design, glitches in the data, and problems they encountered during the work that might affect the results.

For some study designs, such as observational studies, the limitations are a constant: every observational study will, by definition, involve a risk for residual confounding. For other studies, however, you may need to identify factors unique to the work that require readers to exercise caution in interpreting the results. You don’t need to go into every little detail about these issues, but you should still be candid about them. It is OK to offer some explanation of factors that might mitigate these limitations, as long as you do not simply sound defensive.

In some journals, this section also can open with some strengths of the study before discussing its limitations. Check published papers from your target journal for the format previous authors have followed.

The final part of the discussion, sometimes presented as a separate subsection, is the conclusions. Often, this section echoes the final sentences of your abstract and the opening paragraph of your discussion, or a combination of the two. You should summarize briefly what your main findings are and what the main take-homes from those findings might be.


San Francisco Edit specializes in scientific editing in the United States and we work with scientists from all over the world.


Sign Up For Our Newsletter

This field is for validation purposes and should be left unchanged.

Join 90,000+ Scientist Who Get Useful Tips For Writing Better Manuscripts

Don't miss out on future newsletters.
Sign up now.

This field is for validation purposes and should be left unchanged.