This was part of a series of writing tutorials I wrote for my students. This series is compiled on the page A Beginner’s Guide to Writing Scientific Manuscripts.
The Discussion/Conclusions section is where you synthesize the information from multiple experiments into some new insights and make connections with the wider research world. Journals differ significantly in their formats for Discussion/Conclusions and Results. Some journals combine the Results and Conclusions sections so that you can incorporate the broader significance of your results as you write them. With this style, there may also be a very short (1-2 paragraphs) section dedicated to pure ‘Discussion’ where authors can make other connections to their research field. In the absence of this as an official section, authors typically use their final paragraph to reiterate the main points of the work, what they mean for new knowledge in the field and what outstanding questions remain. Other journals will have a dedicated Discussion/Conclusions section that is separate from the Results section. In this section, you have more space to write about the significance of your results. You should include references to other related studies for comparison and contrast to your conclusions. If there are any outstanding questions left by your research or obvious future directions, you can dedicate the very end of your manuscript to writing about that.
In either format, this is the last substantive writing that your audience will read. It is important that you package your work in some coherent way (new model, refuting a hypothesis, etc) so that the reader will understand the same accomplishment. It is important to strike the proper balance when interpreting your data and incorporating it into the larger knowledgebase of your scholarly field.
A weak Discussion/Conclusion section can leave the reader wondering why they started reading the article in the first place. “So, what?” This is not the way you want others to feel about _X_ years-worth of work on your part. Remember that your first readers are reviewers to decide your manuscript’s publication-worthiness. If they can’t figure out why they read this manuscript (other than the fact that they were assigned to by the editor), then they won’t want others to read it either, at least not without major revisions). You may have a particular interpretation of your data, but if it is poorly articulated or not explicitly stated, then your readers will be left to do this on their own. When this happens, your readers will begin to ask questions like, “Why didn’t you do this experiment?” In some cases, this is a perfectly valid point and you should’ve done that experiment. In other cases, their questions and other possible conclusions are not warranted because you should have more clearly explained how your controls or previous literature argue against certain interpretations.
At the other end of the spectrum, over-reaching your data and making grand interpretations that extend beyond what you have actually shown can also get you into trouble. As fellow scholars in your field with finely-tuned BS-meters, your reviewers will call you out on egregious cases of over-interpretation. This will usually come in the form of requests for numerous, difficult and time-consuming experimentation. If you want to make bold claims, you have to prove it. If you are ready to publish, you need to rein in your writing and tone down your claims. Research always continues, so it is reasonable to stand by a decision that enough experimentation has been done to form a publishable manuscript. Just make sure that any planned future experiments or outstanding questions that you raise aren’t critical to the work at hand. If more than one reviewer disagrees with you about your arbitrary stopping point, then you really should get back to the bench and collect more data.
Ideally, the Discussion/Conclusion section should tell the story you intended when you put your figures together. By this time in the process, you’ve finally found your groove for writing and it’s important to check yourself to see if you’ve found the right balance. Re-read the first draft of your Discussion/Conclusions section with fresh eyes alongside the figures with supporting data and look for any potential weaknesses. Should you do additional experiments or include other controls? Should you clarify your claims? Should you tone them down? Is there any other relevant literature that you should cite to facilitate interpretation? In the end, you should strike the proper balance so that your work is significant and interesting within the boundaries of sound data. Your reviewers will inevitably find something they would like you to change, either with writing or experimentation, but you should do your best to avoid baiting them right out of the gate. Even if you are making a conscious effort to objectively review your Discussion/Conclusions section, as an author you are biased and too close to your own data to see some potential flaws. Have another colleague outside the author list look it over prior to submission for publication.
One final note on grammatical formatting…
For this section, you can carefully marry past and present tense. When referring to specific data or experiments (which you will do only briefly; no need to completely repeat everything from the results section), use the past tense. When describing what it means in a bigger context, these things are true, will always be true and you can use the present tense. Grammar and clear writing are essential in all sections of your manuscript, but meticulous wording is critical for this section to avoid confusion about data vs. interpretation and past vs. present work.