How to Write the Discussion

[See also: How to write a research project. How to write the background, theory, method and analysis sections. How to write the introduction and conclusion. How to review the literature and how to structure a research paper. How to finish on time and how to reference properly. Part of the Craft of Research series. Full program here.]

Video from March 23, 2023

I sometimes say that your analysis should artfully disappoint your reader’s (theoretical) expectations of your (empirical) object. It’s the sort of statement I don’t want my audience to understand too quickly. After all, the “disappointment” in question isn’t the familiar bad feeling we get when things don’t go our way. The reader wants to be disappointed.

Your reader wanted to learn something from your paper, not merely have their existing beliefs affirmed. It may be more accurate (or less prone to misunderstanding) to say that the analysis should challenge your reader’s expectations, it should give your theory a little “pushback”, it should offer some “material” resistance to your idealism. However you want to think of it, there has to be some added value from actually looking at the world, actually investigating the empirical phenomena.

This is especially true if you’re a student, working with what is usually a partial understanding of a given theory. You will never fully understand a theory before you have used it to analyze some data. But since you used your theory to frame your research question, and to inform your methodological decisions, it can sometimes feel like your theoretical insights always come too late.

That’s what the discussion section is for. It lets you make explicit what you have learned about your theory from applying your methods and doing your analysis. First you tell us (in your theory section) what you learned from reading books and articles and then you tell us (in your discussion) what you learned from looking at your data. The difference between the analysis and your discussion is that the analysis is about the data, while the discussion is about the wider consequences.

Because learning always has consequences. Just as you had to change your mind about various things along the way, your reader’s mind has been transformed by reading your analysis. In principle, the reader’s mind was not really changed by the background section, which merely informed your reader about things they had no specific knowledge of in advance, nor by your theory section, which just reminded your reader of what the literature they had already read says. It’s your analysis that has challenged your reader, forcing them to change their mind, and in your discussion section you should make those changes explicit.

Think about this in terms of the implications of your results, either for theory or for practice. Your results will generally show that there is some tension between what the theory expected of the people you studied and how they actually behaved in practice. This tension has to be resolved somehow; something has to give. The interesting question is which way you’re going to make this happen. Will you demand that the theory adapt to the practical facts, or that the practice adapt to the theory’s demands?

You can propose changes to the theory that incorporate the experience of your subjects. Next time you study a similar kind of practice, then, you will expect to find (or will at least not be “disappointed” to find) the same kinds of behavior. You are explaining to the reader how they should look at the world going forward, what they should notice, and what they should not be blinded by.

You are adjusting their perspective on the world, setting it in a new light. The focus of this adjustment (the “knobs” you are “tuning”, if you will) are the concepts that the theory organizes into a system. You might propose to add a concept or remove one. Or you might propose some “tweaking”: a slight adjustment in the scope or application of the concept. Whatever you do, remember that you are rendering future observations easier or harder to make; you are giving the reader new or better tools to form judgments about their own data.

Alternatively, you can propose changes to the practices you have studied. This might involve sketching new policies that could govern those practices in ways that bring them more into line with the expectations of the theory. You are here assuming that, even though the study didn’t find what your reader expected, it is the practice, not the theory, that is to blame. There are lots of papers like this. They will derive “best practices” or “codes of conduct” from the results of the study; or they might propose changes in government regulations. Some will be explicitly political, stating a clear policy objective.

Keep in mind that what you propose might be controversial, so there’s a lot of “rhetoric” in a discussion section. It’s not about how things are but how they ought to be. Accordingly, the practical implications will often turn on the readers’ emotions, just as theoretical implications hinge on their concepts.

In any case, try to keep your discussion section focused on a limited set of key points. These will be the key sentences of a finite number of paragraphs. Each will propose a change in how the reader sees or does things. You are trying to bring about a shift in the reader’s understanding and/or engagement with the world that you share. Give a paragraph to each proposed change. Argue as best you can for why your results demand this change. Throughout, make a deliberate effort to appeal to the reasons and the passions of your readers–their conceptual and emotional apparatus. At this point in your paper, after all, they should be ready for a change.

PS: For reference, here is what I say about writing the discussion in a lecture on “How to Structure a Research Paper” lecture, I held back in 2019. My views haven’t changed much.

Here’s the 2022 lecture: