Writing a scientific paper in four easy steps
Aug 29, 2013 in writing
Writing is hard. But writing a scientific paper? Much less so. That’s because nearly all scientific papers follow a simple, four-section outline: Introduction, Methods, Results, Discussion. Just put the required information into those four sections and you’re done.
The Introduction sets the stage. It explains the scientific question, why the question is interesting, and what you have done to address the question. Many papers have never-ending Introductions, but frankly I doubt anybody actually wants to read a paper with a long Introduction. In my mind, an introduction requires only three paragraphs: A first paragraph introducing the broad topic and explaining why it is interesting. A second paragraph describing a specific gap in knowledge or unsolved problem. And a third paragraph explaining how your work attempts to close the gap or solve the problem. Done.
The methods need to describe in detail what experiments you have done, what analyses you carried out, and so on. There’s little to be said about Methods, other than that most Methods are actually too short. It’s a lot of work to describe everything you’ve done in exquisite detail, so most people get lazy at some point and leave something out. I’d encourage you to be as detailed as possible with your Methods.
In terms of actually writing and structuring the methods, it’s a good idea to group related methods into subsections. An example could be: Experimental protocol; Sequencing; Data analysis; Modeling.
The Results are the heart of the paper. There’s no research paper without Results. I like to approach Results from the figures and tables. Think about which figures and tables you want to present, then write a paragraph for each. Each paragraph should start with a brief explanation of what you did and why, and then it should present the actual results.
For example, a Results paragraph could read:
To determine whether lions can thrive on a vegetarian diet, we raised 10 captive-bred lion cubs on tofu burgers. As control, we raised a second group of 10 cubs on raw beef. We monitored body weight and size, activity level, fur shininess, and health of each cub for a time period of three months. We found that vegetarian-raised cubs grew slower and had a lower overall weight increase over three months (t-test, P = 0.002). Surprisingly, diet did not affect fur shininess (t-test, P = 0.15).
Notice how the first three sentences describe the question and briefly outline the experiment, while the subsequent two sentences describe the actual results. The next following paragraph in the Results should now raise another question and answer it.
The Discussion will generally need at least 4-5 paragraphs, and will likely grow a few paragraphs after you have addressed the referee reports. While the Discussion is probably the hardest part to write, there’s a simple structure to a typical Discussion that makes things easier.
The first paragraph in the Discussion should summarize the Results. Most readers will read the Abstract, maybe the Introduction, and then the Discussion. Write the Discussion as if it were the first thing your readers saw.
Then you need 2-3 paragraphs placing your results into a broader context. What does your work mean for the major question described in the Introduction? Also, how does your work relate to other work in the field? What specifically are the similarities and differences?
Now you need at least 1-2 paragraphs pointing out some potential drawbacks of your work. Every work has potential drawbacks, or at least limiting assumptions. List specific conditions under which your conclusions might be invalid, or assumptions you made that may not be true. As reviewers read your paper, they will assemble in their mind a list of potential drawbacks they’re going to call you out on. To the extent that you’ve already addressed their concerns in your manuscript, you’ll have an easier time in review.
Finally, and this is somewhat optional, you can write another paragraph that summarizes all your findings once more. Many papers have such a concluding paragraph, but it can feel redundant if you opened the Discussion with a strong summary paragraph, in particular if your Discussion isn’t that long. I’d say use good judgement on whether you need the closing paragraph or not. In case of doubt, write it, and see how the paper flows with or without this paragraph.
Results, Methods, Discussion, Introduction
While the order of the sections in the paper is going to be Introduction, Methods, Results, Discussion (or possibly Introduction, Results, Discussion, Methods), you shouldn’t write the paper in that order. That would be backwards. There’s no paper without Results, therefore you should write Results first, possibly in parallel with Methods. Once you have Results and Methods, you can start writing the Discussion. Only once you have a pretty good draft of Results through Discussion should you start working on the Introduction. If the Introduction is the last major section that is missing, it will write itself, I promise. In particular if you keep it to three paragraphs.
I haven’t talked about the abstract yet. The abstract should be a mix of the first two paragraphs of the Introduction and the first paragraph of the Discussion. Write the abstract at the very end, when everything else is done.
There’s no minimum length to a scientific paper. In general, I’d say scientists are more likely to write papers that are too long than papers that are too short. I would recommend not to worry about insufficient length at all (but do worry about excessive length). If you’ve got all the elements I’ve discussed here, and your paper is 3 pages long, great. Send it out! Also, 3-6 figures and tables total is a good number. If you’re way beyond 10, you need to cut or move stuff to the supplement. A paper that is too long will not be read, regardless of how good it is. Honestly. Just cut it down.