Conclude your science. Don’t summarise it.
A summary that merely repeats previous material is prohibited for the journal Nature and would be be edited out. Other journals are less strict, but perhaps they should follow Nature’s lead and recommend instead that the conclusion offer something new to the reader. This is often easier said than done. Scientists default toward endings that are typically cliche, uncompelling, or just tail off. Let’s look to factual but more expressive forms of writing, such as long-form journalism and narrative non-fiction, for examples of better endings that could be applied to scientific papers and talks.
Journalists arguably have a little more freedom than scientists in how they word the ending of a piece. A memorable quote or a clever joke, perfect fodder for a popular article, would be out of place in a scientific article. Yet there are several forms of conclusion that we could borrow from journalists to provide a more engaging ending. I’ll borrow my examples primarily from The Atlantic, but any decent popular publication can help.
A conclusion, of course, distills and emphasises the main points of an article or talk. The trick is to sound like you’re not repeating anything. This can be achieved in various ways, either with a single sentence or paragraph.
Devise an eloquent, succinct message that epitomises what you want the audience to take away
Discussing the apparent disadvantages of being introverted, Olga Khazan finishes by stating
When you keep your personality on the inside, people think it doesn’t exist. The tyranny of extroversion persists.
I provide the penultimate sentence only for context. Otherwise she encapsulates the entire article in five words.
A scientific example of a succinct takeaway is given by Stephen Heard for the last sentence of his book concerning scientific names:
Ignominy and heroism; obscurity and fame; hostility and love; loss and hope. It’s all in a Latin name.
Introduce something new, but related.
Introducing something new is not incompatible with concluding. As above, the key is for the new material to epitomise or illustrate the main content. Examining workplace theft, Rene Chun finishes by describing an extreme, memorable example that he had not previously mentioned:
… Then let me tell you about Gilberto Escamilla, the employee who stole $1.3 million worth of fajita meat from a juvenile-detention center in Texas…
This isn’t his final sentence, but rather part of a concluding paragraph. You’ll have to read the article to find out what ultimately happened to this workplace thief.
I’ve personally used this method of concluding in several scientific talks. For example, I’ll play an animation of an ocean simulation that is simple and intuitive. Although it is different to anything I shown prior to that point, it incorporates several of the processes that I discuss earlier in the talk. As it plays, I can point out how the different processes interact.
Answer the overarching question
Science isn’t meant to be a mystery novel where the suspense builds up until the final page. But that doesn’t preclude a similar style of ending. Questioning the possible outcomes of the Brexit deal, Rosa Prince ends by taking all the evidence she has just presented and uses this to ultimately answer the question the reader wants to know:
Will Theresa May choose heart over head and allow the United Kingdom to crash out of the European Union? Everything in her history suggests she’s capable of it.
A great example from the scientific literature comes from Carmack et al. (2016). The concluding paragraph is so well written that additional context isn’t necessary:
In closing we ask, “Are the current trends of change observed in the ocean, its ice cover, its contributing watersheds, and its atmospheric heat and moisture fluxes sufficient to push this complex, coupled system into a new stable mode of behavior, with altered gyre configurations, modal jet stream structures, and storm track arrangements?” The answer is, “probably not,” but the risk posed to our existing infrastructures, economies, and life support systems is of sufficient magnitude to warrant very deep consideration.”
What not to do
As scientists, we are interested in the unknown. Perhaps too much so. Too many articles end by noting unanswered questions or directions for future research. Although concrete suggestions for future research (but not vague calls for study in a general area) are an appropriate part of a conclusion, they shouldn’t be the final sentence or paragraph. A published paper encapsulates months or years of hard work answering a question. Yet, almost invariably, the final sentences highlight what we don’t know rather than promoting the new piece of knowledge. We are underselling our work.
Another cliche way to conclude is to recap a study’s limitations. Examples include “We cannot rule out A”, “Distinguishing between B and C remains a challenge”, or “Because of D, we were unable to analyse data between years E and F”.
Scientific talks have their own assortment of poor endings. The most common is the slide containing only bullet points of the main findings. Another is the unnecessary slide, like Any Questions, which results in the talk tailing off, rather than finishing emphatically. Make it nice and clear to the audience that the talk is finished. If you have to say something like “and that’s all for today”, you’re doing something wrong.
One more example
Consider the concluding paragraph from one last Atlantic piece, this one regarding the lack of people learning trades:
Each of these trends is a puzzle piece that can partially, but not fully, explain the men missing from the workforce. The study authors openly acknowledge that there is not yet a complete answer, though they’re attempting to piece together a more complete picture. “This is really quite a complex phenomenon, and even though it started 50 years ago, we still don’t fully understand it,” Binder says. However, one thing is abundantly clear: Each of these trends is amplified by the lack of a college degree.
The journalist promotes the established result. The researchers promote the limitations and unknowns. It is time for this to change.