26 March 2011

How to write a scientific paper: a discourse on discussions

It is customary for scientific papers to have a "discussion" section. This is different than a review of the relevant literature and should not be confused with it. The literature review, often excessively long and destined to be skipped in its entirety by the readers, is usually placed as an introduction to the paper. The discussion section, on the other hand, comes after the "results" section.

What is the purpose of the discussion section?

The relevant dictionary definition of discussion is: "A formal discourse on a topic; an exposition." Whereas discourse is defined as "A formal, lengthy discussion of a subject, either written or spoken." So, the discussion section of a paper literally discusses the findings reported elsewhere in the paper. Let me expound that a bit more.

1. The discussion section of a manuscript should try to explain the results within a theoretical framework accepted by the general scientific community.

If the results do not fit into an accepted theory, then there are 3 possibilities: 1. the results do indeed fit into an accepted theory, but the author doesn't know how to do it; 2. the results are wrong, because either the experiments or the observations were not done properly and therefore, they should not be published; 3. the accepted theory or theories are wrong, which may mean that the results are truly revolutionary. The last case is a rare happening. But when the author is convinced of it, the discussion section of the manuscript should boldly declare the existing theories kaput and, if possible, devise a better theory that indeed explains the results.

2. When applicable, the discussion section of a manuscript should explain, or provide clues towards a future explanation of, phenomena that were unexplainable before the results presented in the manuscript were obtained. Therefore, these results also suggest how an initially small increase in the size of the foot could have initiated the evolution of slugs from snails.

3. Note that some scientific explanations are actually predictions. I think my results will also explain in the future this presently inexplicable observation.

Therefore, when applicable, the discussion section of a manuscript should make a justifiable prediction or two towards the clarification of present unknowns. In lieu of a prediction, the discussion could pave the way for future research. One of the shortest papers I have written (about a new record of the snail Vertigo pusilla in Turkey) ended with a 3-sentence "discussion" that fit my finding within the previous records and gave a direction for future surveys.

The southern border of the distribution range of V. pusilla is not known (Pokryszko, 1990). The present record and those in Schütt (2001) establish the definite presence of the species in a relatively narrow region restricted to northwest Turkey. It remains to be determined if the species is present further south and east in the country.
I am a big supporter of short scientific papers. All scientists should strive to trim off the needlessly long introductions, methodologies, discussions, conclusions from our papers. Even the results sections of manuscripts should have no more than one or two essential experiments or observations. If you need to do 20 different experiments or observations to prove one point, then you are probably missing one crucial test that would be enough.

More can be written on this subject, but I will leave further discussions to future posts.

I was prompted to write this post by the poorly written discussion section, in addition to poorly written everything else, of a manuscript I have been peer-reviewing. I will probably include some of the ideas presented here in my anonymous review report. The author, if he/she ever does an Internet search on the writing of scientific manuscripts, may eventually figure out that it was I who rejected the manuscript for publication.

No comments: