I was puzzled recently by someone’s comment that a paper I was reading didn’t have any conclusions—until I realised that the speaker meant that it didn’t have a section titled ‘Conclusions’. I explained, too dogmatically, that in the classic IMRD structure (Introduction-Methods-Results-Discussion), an important role of the Discussion section is to draw out and state conclusions. At the same time, the observation made me reflect on how articles are structured.
The use of an Introduction section and a Methods section is still almost universal (although some important journals like PNAS prefer to relegate Methods to a small-type section at the end). By contrast, the presentation of results, discussion, and conclusions is quite variable. Here is my take on some of the pros and cons of both pure IMRD (separate Results and Discussion sections, no Conclusions section) and two common departures from it: the combined Results and Discussion section and the separate Conclusions section:
| System | Pros | Cons |
|---|---|---|
| (A) Separate Results and Discussion sections. No Conclusions section. | Promotes clarity of distinction between results and their interpretation. If done well, makes for an elegant and persuasive transition from results to conclusions. | The Results section may seem very short and dry. Not everyone has the writing skills needed to ensure that the discussion argues smoothly and persuasively from results to conclusions. |
| (B) Combined Results and Discussion section. | Easier to write than (A); removes risk of having a very short Results section. | Risks blurring the distinction between results and their interpretation, and removes a stimulus to clear thinking. Unless (C) is also used, the paper is reduced to three sections. |
| (C) Separate Conclusions section. | Conclusions are clear. Easy to write. | The Discussion section is more difficult to write if the conclusions are removed from it; it has no climax to build towards and it becomes inconclusive (literally). |
How to weigh these pros and cons? The sensible approach is to ask how they affect the communication of the results of research. I suggest that:
- A short and ‘dry’ Results section should not be a problem (if anything, quite the contrary). For example, a Results section may consist only of a single sentence such as ‘There was no significant difference between treatments (p=0.35) (Table 1); mean values were similar to each other (Table 2)’. Such brevity does not imply at all that the results are unimportant.
- In science, the distinction between fact and opinion or interpretation is important. A separate Results section leaves results bare and blinking in the light, and it enables readers to quickly appreciate what the results are—as opposed to what the authors’ opinion of them is, which belongs in the Discussion.
- A well-written discussion, incorporating conclusions, is the most elegant way to finish a paper. Nevertheless, it is difficult to mount a serious argument against the use of a separate Conclusions section, particularly for those who struggle to write a well-structured discussion. Some readers, too, will appreciate a separate Conclusions section. Nevertheless, writers should take care: the separate Conclusions section may aid clarity, but in many cases it duplicates (under the guise of reiteration) what’s already in the Results and Discussion. If you have a separate Conclusions section, it’s best to avoid stating your conclusions in the Discussion.
Take-home message: Use a separate Conclusions section if permitted by the journal and if it aids clarity, but avoid combined Results and Discussion sections—or at least be careful to distinguish facts and opinion.