Portal de Periódicos da CAPES

Writing a Research Paper

2020; Elsevier BV; Volume: 106; Issue: 4 Linguagem: Inglês

10.1016/j.ijrobp.2019.12.005

1879-355X

Theodore S. Lawrence,

Academic Writing and Publishing

Although the papers I have written with residents and faculty are not always accepted, most reviewers begin their comments with “This is a clearly written paper.” I think this must be why several people have asked me if I have a formula for writing research papers. The answer is yes, I do, so I decided to write this brief essay to describe it. I do this because I have learned the hard way that good data can be sunk by bad writing and have seen the pain in some colleagues’ faces when they have a good project but can only stare at a blank computer screen, not knowing how to start. I hope these comments may be of help to the novice, and, perhaps, even contain one or two light bulbs for more experienced authors. Writing a research paper begins 6 months to a year (and sometimes longer) before writing the paper itself by establishing the concept of the paper. I will not discuss in detail in this short essay the topic of how to pick a good project, but a few comments are in order. Every project, whether it is a laboratory study, a prospective clinical trial, an analysis of a national database, or a single-institution retrospective review, should test a hypothesis of interest with appropriate statistical analysis. I am rarely engaged by an article entitled “Stage II A Seminoma: Treatment Results and a Review of the Literature” or something to that effect. These articles are of a standard form: “This is what we did. This is what we found. You should do it, too.” Instead, as you begin to conceive of a project, you should always be asking yourself “How can I make a unique and valuable contribution?” Or think like a reviewer and ask yourself “Why should I care about this?” I imagine that it is late in the evening and someone is going to read a few papers before going to bed. They read my abstract. Do they say, “That is interesting; I never thought of that”? If not, is the study really worthwhile? Even a single-institution retrospective review, when thoughtfully carried out, can generate an interesting hypothesis that is the seed of a more definitive study. Let’s say we have chosen a great project and are in the process of carrying out the study. I always ask my students and colleagues, “What are the 6 figures in our paper?” You can even sketch out your hypothesized figures to focus your efforts. Every experiment or datum in your chart extraction is aimed at supporting one of those figures. Otherwise, you are diffusing your efforts. Scientific inquiry produces surprises (that is often the best part), so your 6 figures may change, but you always have them in front of you. The alternative is to do a series of experiments or pull random data from the chart and then stare at them and ask, “What do they mean?” There is a good side to this: Sometimes you can achieve a serendipitous insight. More times than not, however, when you finally see the pattern, you must fill in multiple experiments or go back to the charts in an attempt to support the hypothesis you drew from your random walk. The random walk may be more fun than the determined march toward your 6 figures, but in this era of constrained resources, I cannot recommend it. Now you have analyzed your 6 figures, and you are ready to write. Remember that writing a research paper is not like writing a mystery. By the third paragraph of the introduction, we should know that the murderer was Mr Green, with the lead pipe, in the conservatory (for those of you who have ever played Clue). A paper tells a story in which there are no surprise endings. In a perfect paper, the reader would anticipate each paragraph just as you have written it. To this end, although all of the data must be correct, you are not required to describe every experiment that led you in what turned out to be the wrong direction earlier in the course of the study. You are allowed poetic license in describing the chronology. I will often write in the results: “After we found x, we decided to do y.” That may not be precisely true. What may have actually happened is “after we found x, we were confused. We tested hypotheses a, b, and c, and we were even more puzzled. Finally, we decided to try y, and it all made sense.” I remember reading some 50-page articles in the Journal of General Physiology when I was writing my PhD thesis that were written chronologically, and I recall thinking how refreshing it was to see how these great scientists took two steps forward and one step back. Modern journals, however, permit you 3000 to 3600 words, and you have to cut some corners to achieve clarity. Here is the formula I use, in the order in which I write the sections.1.Lock your 6 figures. Unfortunately, when you try to lock your figures, sometimes it turns out that you have a hole in your logic. Best to get this set first. A few comments about figures:a.The figures should have readable fonts, and all axes should have units of measurement.b.Try to keep to a maximum of 4 lines per graph.c.Lines should be labeled on the graph. You should want to make a slide from it.d.If data are continuous, use line graphs. Save bar graphs for ordinal data, as they are less informative.e.It has become popular to have 6 or more panels in a figure, given the relentless desire of reviewers to see more and more data while journals want to limit the number of figures. I do not like this because it produces clutter and decreases clarity, but I do not like Ann Arbor weather either, and I still live here. I suppose we are stuck with it.2.Title page. I always do this as the first real writing. Once you have a title page, the paper feels official. A glow of “I am ready to communicate my hard-won data and hypothesis to the world” should suffuse you. You cannot turn back now!3.Methods. I write this next because it is the easiest to do. It is calming to make progress on the paper without contemplating the potential weakness in your data.4.Introduction. Some people write results next, because, like the methods, it should not be too controversial if you have locked your figures. I have found that I like to write the introduction next because, as you will see, I use the last paragraph of the introduction to organize the results. The introduction is organized as follows:a.Paragraph 1: Big picture. However, I would avoid first sentences such asi.“There are 52,000 cases of pancreas cancer each year”ii.“Pancreas cancer is a terrible disease and new treatments are urgently needed”Anyone who is reading your paper will know this. Why waste words boring your readers with what they already know?b.Paragraph 2: Focused picture (in some more complicated subjects, you might have 2A and 2B to discuss 2 areas needed to set up paragraph 3)c.Paragraph 3: “Therefore, we decided to study…” This offers a high-level outline of the article with the often-used expression “When we found x we did y.” I do not recommend a full summary here, although it is a recognized style. You have the abstract and the first paragraph of the discussion to summarize your article.5.Results. The results should follow a logical order. Each paragraph has a structure (as outlined here), so there is an orderly and logical progression of the story. Subheads are very useful and should appear every 2 to 3 paragraphs. You should be able to achieve a high-level summary of the results by just reading the subheads. Some additional points:a.I will occasionally have a small bit of discussion in the results if it is a small point that is not worth a paragraph in the discussion section. For example: “…as had been previously reported (ref).” You are simply strengthening a finding that you are building upon to make a more important point.b.I try not to have too much methodological detail in the results. This is what the methods section and the figure legends are for. Results should flow.c.For laboratory studies, I tend to start with in vitro and move to in vivo. For clinical studies, I tend to begin with patient characteristics, then efficacy, then toxicity.6.Discussiona.Paragraph 1: “In this study, we have found…” List your main conclusions. The last sentence should be the take-home point of paper. You should be able to read smoothly from the end of the introduction to the beginning of the discussion.b.Paragraph 2A: Compare your primary finding to the literature.c.Paragraph 2B: Compare your secondary finding to the literature, etc.d.Paragraph 3: “This study has some limitations.” Be careful here. If a limitation seems important and addressable, you will likely be asked to address it in the revision. Therefore, all the limitations should be either not important (ie, sentence pairs of limitation and antidote to limitation) or so large it would take an entire paper to address them.e.Paragraph 4: Looking into the future. Hopefully, you already have some data that will lead to your next paper. Ideally, this paragraph resembles the last paragraph of the introduction of your next paper.7.Figure legends. These could have been written earlier along with the results, but somehow I always put them off to the end. The title of the figure should state the result.8.Abstract: After the paper is written, I write the abstract. The abstract is a little paper: It has a purpose, methods, results, and conclusion, but each is only a couple of sentences. As is the case in the discussion, you should be able to read smoothly from the purpose to the conclusion. It is remarkable to me how often the purpose and the conclusion of abstracts are not connected in both papers and abstracts submitted to national meetings. I will conclude with some general comments about writing. Most of these comments come directly from Strunk and White, Elements of Style, which is required reading for every scientific writer. I still read it once a year and am amazed every time by its clarity and charm.1.Every paragraph has a structure. A topic sentence states what the paragraph is about. This is followed by a series of statements unified by this concept by content or chronology. A concluding sentence bridges to the next paragraph. It makes the paper much easier to follow.2.Active voice makes a paper more fun to read.3.Avoid needless words. My “hit list” includes “the fact that,” “it has been shown that,” “clearly,” “importantly,” and “consistently.” None of these words contain information. If the statement is not “important” or “clear,” stating that it is does not make it so. Another example is “a viable option.” I try putting the opposite of the word in its place: “a nonviable option.” If it is not viable, it is not much of an option, so “an option” should suffice.4.Avoid idiosyncratic abbreviations. I find it irritating to read the conclusion of an abstract that sounds like “This trial did not show noninferiority of FST versus SCM with the chosen Nlm.” (This is a real example from the Journal of Clinical Oncology.) Pick up any journal and you will find even worse. I hope this is helpful. May your next paper be accepted without revisions!