How to write a Discussion Section.

A Discussion section might be the most difficult for me to write. You want to be clear that you know what is going on in your method and the implications of the assumptions in your methods and analysis choices on the results, but sometimes these questions are too self-deprecating and can be used against you by reviewers to extend the scope of your work beyond where this paper should go.

In general, I follow the following three paragraph format for a Discussion:

• Summarize the main contributions of the paper, the results (three sentences, all referencing specific “these suggest” statements in your Results), the conclusions in one paragraph.

• Bring up some (obvious) limitations of the assumptions. Describe why you have bounded the impact of these assumptions in your results.

• Bring up some (obvious) future steps for this approach, and motivate them in terms of the new domain-specific results from these future steps.

I say “obvious” only because you have been thinking about and discussing this method with everyone you meet for the past year or so, so you may think these limitations and future steps are obvious at this point. Don’t dig deep here – these obvious ones will be most helpful for readers getting to know this work. These can often be at the level of questions you have received at seminars where you discuss this work. If you dig too deeply to find these limitations, you will lose the readers and appear as if you overlooked the obvious limitations of your approach.

That said, my favorite Discussions have every limitation and assumption that got me wondering during my reading laid out clearly. If you can guess all the reasons your readers might doubt your work, you earn their trust. This is an important part of the process. Some Discussions are fast and write themselves; others you really need to be careful about integrating all your feedback thusfar and knowledge to put together these points.