Ten tips for revising your first draft (paper writing for PhD students, pt 15)

Previously in this series on how to write a scientific paper: Zen and the art of revising and, if you didn’t read this yet, how (and why) to write a crappy first draft

“Don’t use big words. They mean so little.” -Oscar Wilde

When editing and revising your paper’s first draft, my suggestion is to do two passes: first, a pass that focuses on the broader issues of structure and content, and then a second pass that focuses on the nitty-gritty, sentence-level details. In this post, I will present ten tips for revising your draft that can be used as a checklist for the first pass; this list contains issues that I frequently come across when working with students and revising papers.

If you have read this series this far, you won’t be surprised to see that most of the issues have to do with clarity and focus.

  1. Check that the abstract follows the hourglass structure: broad context, narrower context, the research question, your result, implications of your result on your (sub)field, its broader implications. Also, do make sure that your abstract is as jargon-free as possible: only use words that most readers can understand.
  2. Check that your paper is focused. Choose the point of the paper and its key conclusion before you begin writing, stick to your choice, and write the paper so that the reader gets the point already in the abstract and in the introduction. Leave out results that are not required for supporting the key conclusion, or safely tuck them away in the supplementary information document. When editing, if you feel that your paper loses its focus at some point, take a step back and do a major rewrite.
  3. Check that there is a clear question and a clear answer. A good paper states and then solves a problem; your results are meaningful only if they solve a meaningful problem. Remember that your paper is neither an account of your work nor a lab diary; it should be a story of an important problem and its solution. Emphasise the problem, both in the Introduction where it should really stand out, and in the Results section and the Discussion. Make it clear to the reader how each result contributes to solving the problem, and what the implications of solving the problem are.
  4. Check that the figures tell your story. If you just glance through the figures and skim their captions, do you get the point of the paper and its take-home message? If not, go back and revise—after all, skimming is what most of your readers do.
  5. Check that the reader can replicate your results. Verify that your Methods section (and the supplementary sections if any) contains everything that the reader needs to know. Also, check that you provide links to your code and your data if it can be released without violating anyone’s privacy.
  6. Check that you end the paper with something worth remembering. This means something concrete. “More research is needed” is a platitude and a vague one at that; better, go for something like “because of the results of this paper, we are now in a position to tackle problem X with method Y, bringing us closer to the ultimate goal of Z”. This is far more concrete and memorable. Endings have power; do not waste this power.
  7. Check that you provide enough background information: your reader does not know what you know. Assuming that your reader knows much more than you and therefore omitting background information is a very common problem with students. A typical example would be a Methods section that directly launches into what you have done without first telling why. Although it is evident to you that to get from A to B you need to do X, this is probably far less obvious to the reader. If you only tell the reader that you did X, she is confused. Why did you do X? Never assume that the reader knows your motivation, or the details of every method you used, or why your research question is important. Tell her.
    Many students seem to think that they know little while everyone else knows a lot—therefore they shouldn’t explain things that everyone probably already knows. It is only later in their careers when they realise that no-one really knows that much! Besides, there will be readers from adjacent (sub)fields and readers who are just learning the tricks of the trade. Use a colleague who works on something slightly different than you as a test reader—ask her which parts of the text are hard to follow, and revise accordingly.
  8. Make sure that you take the reader’s hand and lead her through the text with signposts. Or, in other words, check that your writing is not confusing. Writing is, in part, psychology, and it aims to modify your reader’s state of mind and to influence what your reader thinks. Feel empathy for your readers and try to get inside their heads, assuming that they know nothing or very little. Your empathy should be reflected at the level of sentences and paragraphs: present familiar things first before moving to new concepts, use leading sentences, glue your sentences together with expressions that guide the reader. Use subheadings. Gently lead the reader from result to result, from paragraph to paragraph, and from sentence to sentence. Never leave it to the reader to connect the dots— always connect them for her. Err on the side of caution: papers where things have been over-explained are rare (if they exist at all), but papers that are all too difficult to follow are frustratingly common.
  9. Check that you are consistent with nomenclature and notation. Because you have been immersed all too long in the world of your paper, this problem may be hard to spot for you—using an outside reader as a guinea pig is recommended. Problems with notation are easier to detect; problems with naming things are more difficult. Often, while doing research and while conceptualising the paper, there is a number of concepts floating around, and the very same things can have many names in your thinking. Writers of fiction are allowed to use synonyms for variation, but science should be precise: in the final version of your paper, everything should be called by one name only. While it may be evident to you that the thing you call the weight matrix is the same as the thing that was called the correlation matrix in the previous paragraph, your reader quickly gets confused. Never refer to the same thing with multiple terms.
  10. If you feel that it is impossible to get some part of your text just right, this is often a sign, a message from you to you. When you are stuck with a paragraph that just won’t yield, stop trying to force it. Instead, ask yourself: why is this so difficult? Search your feelings. What would make the paragraph easy to write, what are you missing? Often, you will notice that you are not faced with a writing problem at all—rather, you miss some important piece of understanding. Perhaps your result is not clear after all, or you have not thought enough about some tricky issue and that is why you cannot express it in words. So take a time out, and look for understanding first; the words will come more easily when you have found it.

New to this series? It begins here: Why can writing a paper be such a pain?

Zen and the art of revising (paper writing for PhD students pt 14)

[Previously in this series on how to write a scientific paper: how (and why) to write a crappy first draft

How to write a scientific paper, book cover

“If you feel the urge of ‘very’ coming on, just write the word ‘damn’ in the place of ‘very.’ The editor will strike out the word, ‘damn,’ and you will have a good sentence.”—William Allen White

If you have followed the advice in the last chapter, you should now be the proud owner of a crappy first draft of your scientific paper—a draft that serves as raw material, a draft that is for your eyes only, a draft that was written quickly and without too much care.

Now it is time for you to put on another hat and play a different role. It is time to look at your draft critically and to examine each and every sentence and paragraph ruthlessly so that you can cut out everything that doesn’t carry its own weight.

Before that, however, it might be a good idea to take some distance unless you are in a big hurry because a fresh pair of eyes can better spot what needs to be done.

How to revise your paper’s first draft? The process of editing and revising a scientific paper is iterative and it can take many rounds: my most-cited paper was at version number 27 or so when it was finally submitted. This may sound a bit excessive, but hey, it worked! You don’t always need to go to that length, though–just be sure to do several rounds of revisions, first alone and then with the help of your co-authors and/or your supervisor.

Just like with writing the draft, I recommend using a top-down approach when revising—begin with addressing broader issues before homing in on the details. First, read the draft quickly, without getting stuck on sentences, words, or other nitty-gritty details. Then, go through your findings: is the story logical, clear, and exciting? Does the abstract do its job and entice the reader? Is it clear what problem the paper solves? Is it clear what the solution is? Are concepts introduced in the right order? Is the paper balanced, or are there sections that are too long or sections lacking in detail?

Is it clear that your results are backed by solid evidence? Are the figures of a high quality and free of common errors such as microscopic label fonts? Does the paper begin with a proper lede – a sequence of sentences that frame the topic of the paper and entice the reader to read the rest of the story? Does the paper end on a high note?

The answers to the above questions may result in a need to “remix” the paper: to shuffle its contents around, to reorder things, and to completely rewrite some sections. This is normal: if you feel the need, just do it. Then, repeat the top-level analysis of your draft: answer the above questions again, and see if you can think of ways to improve the text further. If the answer is yes, do it. Repeat this loop until you are happy with the outcome and satisfied with the overall structure and flow of your paper. At this stage, you may even feel like returning to your research, say, to look for new results that back up your conclusion even more strongly. If so and if there is time, great, just do it, but please do remember to stop at some point because there will always be something new just around the corner. Leave some of that for the next paper.

When the overall structure is there, you should focus on the level of paragraphs and sentences. Use the same rules as for writing the paragraphs. For each paragraph, check that its topic is made clear in the first sentence or two. Check that the paragraph doesn’t stray away from the topic. If it does, cut it into two, or revise it. For each sentence, check that its meaning is clear, that it connects with the previous sentence, and that the rules outlined in the section on sentences below are fulfilled. Split sentences that are too long. Check the grammar. Use a spell checker.

Check your notation and nomenclature, and straighten them out if necessary. Do you always use the same word for describing a concept, or do you use several names for things? Is your notation consistent, do you always use the same symbols? Do you explain every symbol used in every equation?

Check your figures. Are your axis labels large enough to be seen without a magnifying class? (I repeat, this is the most common mistake in figures produced by PhD students, for reasons unknown to me: fonts whose size is measured in micrometers). Are your axis labels clear, and is the notation consistent with your body text? Are the colour schemes you use clear and informative, and most importantly, consistent across figures? Do the figure captions explain what should be learned from the figures, instead of only describing what is being plotted?

Then, finally, when all else seems in place, do a shortening edit, with the target of removing extra clutter and superfluous words. Make every sentence shorter that can be made shorter. Remove all adjectives, unless really necessary. Remove all repetition. Remove words that exaggerate things, because you sound more confident without them. Remove every instance of the word “very”, because you never need it. Remove the words “in order” from “in order to”.

When you are ready to show your improved draft to others, you can apply a technique that my research group has borrowed from the software industry: Extreme Editing.

In the software industry, extreme programming is one of the fashionable agile techniques, and part of this technique involves programming in pairs. So edit in pairs! Or, if there are more coauthors, involve as many of them as possible. Force your PhD supervisor to reserve several hours of uninterrupted quality time; you can argue that this co-editing session takes less time than several rounds of traditional red-pencil-comments.

This is how extreme editing works: go to a meeting room with a large enough screen and open the draft on the screen. Then, go through your text together, paragraph by paragraph and sentence by sentence. Be critical of each word and each sentence; look for sentences that are unclear and that can be misunderstood. Try to find ways of reducing clutter and shortening sentences. Cut out fat wherever needed. In my group, we jokingly keep a tally of points scored for every removed word. The winner is the one who has most ruthlessly killed the largest number of words that just tagged along, doing no real service to the text.

In the following two posts, I will present some more tips on how to revise your draft, first on the level of meaning and structure, and then on the level of sentences.

How (and why) to write a crappy first draft (paper writing for PhD students pt 13)

For the previous episode in the series on how to write a scientific paper, see here.

“To write is human, to edit is divine” -Stephen King

The best and most productive writers do not write perfect first drafts. The best and most productive writers write crappy first drafts and they do this as quickly as possible. They then edit, revise, and polish their crappy first drafts until those are no longer crappy (and no longer drafts). Or until the deadline makes them stop, whichever comes first.

This is what you should do with your scientific paper too: write the first draft quickly, and then edit, revise, polish, rinse, and repeat, until you are satisfied with the outcome. Or until the deadline comes.

If you have followed the system outlined in this blog, you are now at the point where you are ready to write your very own crappy first draft. You have a story, you have a structure, and you have notes for each section and each paragraph. If you have read the previous chapter, you have some idea of how to organize the building blocks of paragraphs and sentences (recap: the first sentences/words tell what the paragraph/sentence is about; stick to this and keep it simple; put weighty stuff at the end). This is all you need to know for now; I’ll provide plenty of tips for editing later.

So at the time being, put all rules aside, and aim to produce to a complete first draft quickly. Embrace the words of Stephen King quoted above and forget perfection when it comes to the first draft—let it be human, let it be imperfect. Let it be crappy! Why? Because producing and then polishing a crappy first draft is much, much faster than agonizing over every word and sentence and making only perfect choices that take forever to make. When all that time is spent on editing and revising instead, the outcome is much better.

Now that you have to finally produce some text, this is where the pain of writing typically hits you. Coming up with plans and storylines can be fun; writing rarely is. Writing is hard work. Writing the first draft is particularly hard work because not being self-conscious of your words is hard, and because not letting your inner critic stop you in mid-sentence is hard. These demons are difficult to wrestle but wrestled they must be, otherwise, there is no progress and the pages remain blank.

How to ease this pain, especially if you are a novice and it feels overwhelming? How to write all that text that needs to be written before you have a paper? There are some techniques that may help you.

First, make the first draft your own little (crappy) secret. It is not for your supervisor’s or co-authors’ eyes—it is for no-one else’s eyes, it is only for you, and it serves as raw material for editing only. When your supervisor asks you for the first draft, you should give her your second draft instead—by all means, call it the first draft! Keeping your first draft private should make you less self-conscious, at least in theory: no-one else will ever see it.

Second, aim at producing more text than you need. Just let the words come! At this stage it’s OK to have sentences that are too long, it’s OK to repeat yourself, it’s OK to explain the same thing over and over again with different words. In particular, if you are writing, say, one of those 4-page letters with a restricted word count, do not worry about the length at all. Just write. Cutting text is easier than producing it, and the editing phase easily reduces the length of your text by 10-30%. In my experience, the more, the better the final product.

Third, to be productive, schedule writing time and stick to it. Never wait for inspiration to strike, because it rarely strikes those who just sit there waiting. The Muses dislike idleness; they tend to show up when you are already engaged in work. Just sit down, put your phone on silent, remove all clutter from your screen, shut down your Internet access, and do it. Write. A good target is something like 30-45 minutes of uninterrupted writing, followed by a break. For a really good day’s work, four to five of such sessions are already enough. Just keep on doing this daily until you find yourself at the end of your first draft.

Fourth, if you get stuck, try changing the way you write. Take a pen and a notepad and walk away from the computer. Sit down somewhere, get a cup of decent coffee, and sketch your sentences on paper. Try to write as if you would be making lecture notes or just jotting down ideas. When unstuck, go back to your computer and use the material in your notes to continue. Or, instead of a notepad, try dictation, or go for a walk and play out imagined conversations in your head where you explain whatever it is that you are supposed to be writing to someone.

If you are very self-conscious and find it hard to make progress because of that nasty voice in the back of your head, you might want to try something along the lines of the Morning Pages technique. This technique provides desensitization by stream-of-consciousness writing: every morning you take a pen, a journal, and write longhand three whole pages, filling them with anything that comes to your mind. This may feel rather difficult at first; just keep on doing it. Morning Pages were introduced by Julia Cameron in her book The Artist’s Way as a tool for artists to connect with their creativity and overcome whatever fears hold them back. If you’d like to use this technique to help you write your paper, you can fill those three pages with thoughts on your research. See where this leads you.

If nothing else helps and it feels impossible to make progress, stop for a while and think about why this would be. What would need to change for the words to emerge from wherever it is that words come from? Usually, if I find myself in this situation, the answer is that the problem lies not with words or with writing but with thinking: there is something that I don’t yet understand, some pieces that don’t yet fit. Then, the solution is to stop writing (this part of the text, at least) and to solve the underlying problem instead. So take a time-out, and look for understanding first; the words will come more easily when you have found it.

Next: how to edit the first draft of your scientific paper

Paragraphs, sentences, and a toy model of your reader’s mind (Paper writing for PhD students pt. 12)

coverdraft2

[For the previous post in the series, see here]

“Words are to sentences what atoms are to molecules: the basic building blocks that control structure and function. If we extend that analogy, paragraphs become cells: the fundamental unit of life. A cell gains life from its structure, a structure that creates internal cohesion and external connection, allowing it to function as part of a larger organism.” —Joshua Schimel, Writing Science

Finally, after all this planning and outlining, it is time to start filling in the blanks. It is time to write words that form sentences that form paragraphs that form sections that form your story.

Because of the top-down approach that has brought you here, coming up with words and sentences should be easier than having to start with a blank page. You should already have an outline of the paper as well as notes for each paragraph—now you only need to turn those notes into full sentences!

From the point of view of the reader, the best sentences are those that are easy to understand, that make the story flow, that tell the right things in the right order, so that the reader can always expect what comes next. From the point of view of the writer, this is achieved if the writer feels empathy for the reader. A good writer tries to look through the reader’s eyes, taking the reader’s hand and guiding her through the text, making the reader’s job as easy as possible.

To guide the reader through the text, the writer has to gently manipulate the short-term memory of the reader. It has been argued that one’s short-term memory can only hold seven things—such as digits—at a given time. When it comes to concepts that are more complex than digits, even seven sounds like an awful lot to me. There is only so much one can hold in his head.

When a piece of text feels too hard to grasp, this is often not only because the ideas therein are difficult. Instead, there may be a problem with the sentences that should deliver those ideas. This has to do with the order in which the concepts that the sentences contain are placed into the short-term memory of the reader. Bad writing randomly jumps from one thought to another. This creates a traffic jam of thoughts and ideas without giving a clue as to how they relate to one another.

What, then, is the right order of things? How to choose what comes first in a paragraph or in a sentence? To answer this question, we have to understand how the reader processes information—how the reader reads. To this end, let’s construct a toy model of the reader—think of the reader as an automaton of sorts, with three types of memories that you can manipulate. Each type of memory plays a role in interpreting and understanding the paragraphs and sentences that the reader encounters.

The first, long-term memory contains the key concepts that are required for understanding the paper. Using the film script analogy, these are, if you like, the world of your story and its inhabitants. This memory is initialized in the Introduction, and populated with more characters and events as the reader reads. Whenever new concepts become established in the paper, they are added to this memory.

The second, intermediate-term memory contains concepts essential for connecting the dots currently in front of the reader—for making sense of things in the current paragraph and in the current sentence. In particular, this memory contains the topic of the current paragraph (or the reader’s interpretation of it). Not all concepts remain in this memory for long: many are flushed out at the end of each paragraph. If you know time series analysis, you can think of this memory as a sliding window of sorts.

The third, short-term memory holds the concepts that are required for making sense of the current sentence, as in “what is the subject of the sentence, what is the verb, what is the object”, or, “what does what to what”. In addition to grammar, the short-term memory is essential for relating the concepts encountered in the sentence to one another. As the reader reads, the words and concepts of a sentence are placed in this memory in their order of appearance. This memory is flushed at the end of each sentence.

How do these memories work? When the reader reads, she consumes the words from the left to the right and interprets them with the help of all the memories. The long-term and intermediate-term memory help to extract the meaning in the sentences. They also and set the reader’s expectations, “bias the reader” if you will, and determine how words and concepts are interpreted. The short-term memory is used to parse the current sentence and to connect its words to one another.

If all goes well and the reader understands the words in front of her, the concepts that those words hold are first put into the short-term memory, to be flushed out at the end of the sentence. The higher-level concepts formed by the words in the short-term memory are placed in the intermediate-term memory, to be used in deciphering the rest of the paragraph. At the end of each paragraph, this intermediate-term memory is given a vigorous shaking so that those concepts that are unnecessary baggage fall out, never to be seen again. But those that stay and prove to be useful may transcend to the first, long-term memory.

But if the reader encounters something that is inexplicable—something that doesn’t match with anything in any memory stack—an error occurs, and the reader is lost. Or, worse, if the reader uses a wrong concept to parse the current sentence, she can become entirely derailed.

Because of the way this parsing automaton operates, the first words of a sentence are tremendously important, as is the first sentence of a paragraph.

The first words of a sentence initialize the short-term memory, determining how the rest of the sentence is understood—they define the topic of the sentence. Then, when the reader reads on, the words that she encounters are seen interpreted through the lens of the topic. As more and more words accumulate in the memory, the reader is trying to tie them together and make sense of the sentence. The more words there are, the more difficult this gets (think combinatorial explosion). In particular, if the sentence is constructed such that its meaning hinges on the last word(s), it becomes taxing to read.

So, from the point of view of our short-term memory model, the perfect sentence begins with words that clearly define its topic, and is not too long. Any superfluous words like “very” wherever it appears or the “in order” in “in order to” just make the reader’s job more difficult by increasing the number of possible combinations. Also, it helps if the order of words allows one to parse the sentence up to the current word, so do keep the subject and verb close. And one more thing: place familiar words and concepts at the beginning of the sentence, and use them to explain any words or concepts that might be unfamiliar to the reader that appear later on in the sentence.

When it comes to paragraphs, the first sentence of a paragraph establishes its topic, and this topic is stored the intermediate-term memory for interpreting the rest of the of the paragraph. So always make sure that the first sentence of each paragraph is well-chosen and clearly tells the reader what the paragraph is about. This sentence sets the expectations of the reader and determines how the reader will attempt to interpret everything that follows. Always stick to the topic in the rest of the paragraph—an unexpected, unconnected sentence breaks the flow and leaves the reader baffled. But a sentence that makes sense in light of the topic will be interpreted properly, setting expectations for the next sentence as well.

One way to make the topic of the paragraph clear is to begin with a phrase that acts as a signpost that clearly tells the reader where the paragraph is going. Examples include “To measure how X depends on Y, we constructed an elaborate apparatus…” and “In conclusion, in this paper, we have shown that…” When the reader knows what to expect, interpreting the rest of the paragraph is easier.

But not only beginnings are important—endings matter, too, just like for stories. Because the end of a paragraph and the end of a sentence signal a break, the reader has more time to think about the last few words. Therefore, the last words carry extra weight in the mind of the reader, almost unconsciously.  Place important things in these stress positions, things that you want to emphasize.

Next: how to write your first draft.

Discussion: how to wrap up your paper (Paper writing for PhD students pt 11)

[For the previous post in this series on how to write a scientific paper, see here]

Time to wrap up the paper! You have presented your research question, its greater role in the universe, and your findings. Now let’s close the circle and discuss your answers and the questions that still remain, or the questions that you have just discovered. This is what the discussion section of a research paper is for.

First, before we get started, a note for those poor souls who work in fields where you are not supposed to discuss your results in the Results section at all (see the previous Chapter). In this case, the Results section contains just data, and therefore the Discussion section usually has two parts, Discussion and General Discussion. The first part, Discussion, is about interpreting each of your results, which I have already covered in the previous chapter. The second part, General Discussion, is what we will mainly talk about in the following, and I’ll just refer to it with Discussion from now on. Sorry for the confusion (not my fault).

And while we are at it, what’s the difference between writing a Discussion or a Conclusion? The Conclusion is typically fairly short—it is almost like an echo of the abstract. It is a summary of what you have done, and it adds no new ideas or thoughts. In contrast, the Discussion section can introduce new points and insights (but of course no new results). It can connect the dots in new ways, and show how your findings relate to the broader issue at hand. It can ask new questions.

How to write the discussion section of a research paper? In the Discussion, you should remind the reader of your research question and provide a synthesis of your results. You should present the answer to your research question and show how it contributes to solving the broader problem that you have painted in the abstract and the introduction. You should suggest further research based on your results. You should also contextualise your results (and your research problem) within the literature—this is a good opportunity to cite more papers that help to see where your results fit in and what they mean. It is quite common to cite in the Discussion section papers that were not cited in the Introduction: the Introduction and the Discussion are usually the sections of the paper that come with most citations.

There are several ways how to choose the structure of the Discussion section. One commonly used way is to begin by reminding the reader of the broader knowledge gap and the specific research question of the paper. You can then proceed through the results one by one, perhaps grouping them to make certain points and summing up all evidence before arriving at the final conclusion. Or, alternatively, you can begin writing the Discussion by recapping your question and your answer, the final conclusion of the paper. This is the inverted pyramid style, where you begin with the key points—the question and the answer—and then add in details and different points of view. These often appear in decreasing order of importance; however, even when using this approach, you should always have a strong ending (see below).

Whatever the structure you choose, a good Discussion section presents both answers and new questions. These new questions may be related, for example, to explaining your results—say, you have observed an effect that current theories cannot explain—or generalising them to other settings.

While it is common to discuss the limitations of your work in the Discussion section (leading to questions left for future research), it is in my view best not to mention limitations here for the first time. Let the reader know about them earlier. As I wrote in the chapter on Methods, all technical limitations should be discussed immediately when your methods are explained. You should also be open about what you can and what you cannot conclude from your results in the Results section (or the related bits of Discussion). Even though it is common to overgeneralise one’s results because everyone else does that too and the journals and the media like to have their headlines, please don’t. Be an honest scientist. It will pay off in the long run. You can also turn any generalisability issues into questions: instead of claiming that your observation is generally valid, ask if it can be replicated under different conditions or in different sets of data. You never know, one of your readers might pick up the question and do this.

The last paragraph of the Discussion section is very important. It contains the last words of your paper, your final words before back matter such as Acknowledgements—do not waste them. First, if the reader has been with you this far, reward her at the end. Think of a great pop song—even though the listener has heard the chorus already, the producer will always reward her with extra layers of intensity when the last chorus arrives. Do the same: at this point, the reader already knows the point of your paper, so add some icing on top, a new angle, or a new implication of your results. Second, endings are power positions. The last words of your paper will be remembered by your readers (or at least by those of them who made it this far and by those skimmers that directly jumped to the end). Never waste a power position, never waste an ending! Always end on a high note.

There are some all too common ways of ending with a whimper instead of a bang. One way is to focus on the limitations of your work in the final paragraph in a negative light, instead of turning them into a question (see above); in general, it’s better to discuss limitations much earlier than in the last paragraph. Another common problem is ending the paper with a vague statement like “further research is needed”. Further research is always needed, this just makes it sound like you didn’t do enough even when you did! Besides undermining the importance of your work, vague statements are never memorable. Say something that you would want the reader to remember.

One good way of ending the Discussion section—and the paper—is to write a short paragraph that recaps your conclusion and the significance of your work. You can even signpost this to the reader and begin the paragraph with the words “in conclusion, we have shown that…” This paragraph can resemble the bottom part of the hourglass of the abstract; it can also be written in the style of a news lede. What did you found out and why? What does it mean? What are its broader implications? How has your work concretely contributed to the big picture? Where have you come from where you started? How has the world now changed because of your results? This paragraph closes the circle and resonates with your Introduction.

Next: paragraphs, sentences, and a toy model of your reader’s mind.

PS If you liked this post, there’s now an ebook based on this series, available from most digital stores

 

How to write up your results (Paper writing for PhD students pt 10)

For the previous episode in this series on how to write a scientific paper, see here.

Time to start writing up the results! Before you go on, have a look at the earlier post on figures, as it tells how to develop the storyline for this section.

How to write the Results section of a scientific paper? If you have followed the approach in the post on figures, you have now in practice chosen your order of presentation. You have categorized your results (and figures) into Setup, Confrontation, Resolution, and Epilogue. Or, if the movie script analogy is starting to annoy you, into categories that serve a similar purpose.

Results of the first category pave way for what follows. They introduce the reader to your data; they make your final conclusions credible by showing what is in your data and letting the reader gauge whether it looks OK. Results of the second category show that there is an open scientific problem, that there is something surprising that needs to be sorted out. Results in the third category present the Resolution, the key finding that solves the problem (or opens doors to even more problems). Results of the last category, Epilogue, are there to show what follows from the main result and why the main result is important. They would not necessarily work as stand-alone results.

This will also be the order in which your results will be presented.

Please note that the above arc is not necessarily historically accurate (it almost never is): the aim is not to present your scientific findings in the order they came to be, but to present a compelling argument and to provide a bit of entertainment on the way in the form of a good story. And remember what we discussed already in Part I: the paper should only include those results that serve the story and play a role in one of these categories. The rest should be left aside for future papers (or for the Supplementary Information document).

This is the overall arc of the Results section, or, if you are writing a letter-format paper, the arc of the bulk of the paper, sandwiched between the introduction and the conclusion. Make it sure that the reader can easily follow the arc. She should always know where she is and where she should head next.

One way of making sure that the reader is always on the map is to use informative subsection headings. This may require you to divide the four categories into further subsections. A great way of developing subsection headings is to compress each result into a single short sentence and use this sentence as the heading. This way each result gets its own subsection where it can be explained in detail. Note that here, “result” does not necessarily mean a single plot or figure, but rather a conclusion that may be based on several pieces of evidence.

If the results section is organized like this, the reader can get a quick overview of the whole section by just scanning the subsection headings; remember that most of the readers just skim. These skimmers include the editor who decides whether to desk-reject you paper or to send it out to referees.

The above technique is an example of so-called signposting, where the whole paper is made more accessible by covering it with signposts that tell the reader where she is. Clear section headings help, and so do clear figure captions, whose first sentence should tell what the figure is about. Clearly formulated key phrases are also very useful. As an example, it is good to have in the Introduction a sentence that begins with “in this paper, we show”. It is also good to begin each paragraph with a topic sentence that tells what the paragraph is about.

For the results section, the most important signposts are the section headers, the first sentences of figure captions, and something we haven’t discussed yet: the first sentences of all results subsections. The first sentence of each subsection should provide motivation and background: why was the analysis done that led to this result? Why are we discussing this result? This sentence should begin with “to understand why X, we measured Y…” or similar. Even if the motivation has been mentioned in the introduction, the reader should be reminded of it, unless the paper is a very short letter and the introduction is just a few paragraphs away.

So what else is there in a results subsection? Well, results, of course – but there are several layers here. First, the lowest layer contains your “pure” results. Those are, in a way, just data: you have measured X, here’s what you got. You have computed Y, here is a table. The second layer contains direct and unambiguous interpretations of these data: the distribution of X measured under condition A clearly has a lower mean than when measured under condition B. Y grows faster as a function of time than Z. And so on. While such statements may already contain the main conclusion, a third layer of interpretation is usually required – that of giving meaning to the findings, of asking (or telling) what the results mean, of presenting new hypotheses. How do the results bring you closer to answering the broad problem that your paper addresses?

The above layers form a logical order of presentation for each of your results subsection. Begin the subsection with motivation and background. Then briefly tell the reader what you have done, either referring the reader to the Methods section or, if you are writing mixed results and methods, presenting your methods here. Then explain the results that you have obtained and talk about how your interpret them. Make it clear to the reader what layer you are talking about: what is indisputable fact, what is interpretation, and what is speculation. Use signposts for this—saying “these results can be interpreted as follows” and continuing with interpretation helps the reader to understand there might be other ways of interpreting the results. Do not exaggerate or overgeneralise: if there are limitations that haven’t been addressed already (say, in the Methods section), be open about them.

There are traditions in some disciplines where the Results section is strictly about results, and in the fundamentalist interpretation, this would mean only including the first layer (pure data) and perhaps some of the second layer (“these data have a lower mean than those data”). Then, the meaning and interpretation of the results would only be discussed in the Discussion section and not even mentioned in the results. To me, this is, well, insane. It must have been invented by the same evil people who run journals that send papers to referees where figures are separated from the text and the captions are separated from the figures. Why, oh why, should one make the reader’s life so difficult that she has to jump back and forth between Results and Discussion? I cannot think of any other explanation than purely evil intent. But if you work in one of those disciplines and have to publish in journals that demand this, then well, I guess you have to obey the rules. Or to look like you obey the roles. But don’t do it willingly: always try to sneak in at least one sentence that explains your findings. Rejoice if you get it through the editor and the referees!

Before we move on to discussing the Discussion section, one more trick. This has to do with understanding how the mind of the reader works; we’ll talk more about that later. It is very common to begin a paragraph in the results section with “In Figure X, we see that…” Now, next time you read a paper, try to be conscious of your own response to this. What do you do? Do you directly jump to Figure X to have a look, and then try to get back to the middle of the sentence that you were just reading? No? Can… you… resist the impulse? Noooo! OK, but you still feel the impulse, don’t you? It is impossible not to feel it! And the impulse makes it harder for you to follow the sentence to its conclusion. So always refer the reader to the figure last, not first: finish the sentence with “as we see in Figure X” or similar. This way, when the reader arrives at your mental hyperlink, she has already read your sentence and knows what to look for in Figure X.

Next, we’ll discuss the Discussion section.

[Finally: the book. Many have asked me to write a book based on these posts. I’m working on it in an on-and-off way whenever there is time. Having three small kids and being the vice head of a large department doesn’t exactly help. I’ll be happy if the book is out in 2018. My plan is to self-publish, first on Kindle Store and then as an on-demand print version (it is nowadays easy to sell printed-on-demand books through Amazon and other online stores at very reasonable prices; the technology is there). I am at the time being not even considering traditional publishers because i) they would slap a high price tag on the book, limiting the number of readers, and ii) they would then take 85%-90% of said high price without doing much else. I don’t see this as reasonable because, with modern tools, publishers can simply be circumvented. Sorry to say, but traditional publishers: for books like this, you are no longer needed… PS Does anyone know a reasonably-priced proofreader?]

[Update: it’s out as of 1st of November 2018, see below!]

 

How to write a scientific paper, book cover

 

Paper Writing for PhD Students, pt 9: Methods

computer with editors open on-screen

Previously in this series on how to write a scientific paper: the lede – how to hook the reader

It doesn’t matter how beautiful your theory is. If it disagrees with experiment, it’s wrong. In that simple statement is the key to science.” -Richard Feynman

How to write the methods section? While much of this series has been about writing an exciting story, we now need to put excitement aside for a while. I’ve earlier claimed that scientific papers are not only containers of information. Their Methods sections, however, are. Their role is entirely utilitarian. So before we discuss form, let’s discuss function.

A Methods section of a scientific paper serves two purposes. First, it should let other researchers gauge whether your conclusions are justified and backed up by evidence—it should let other researchers assess how credible your data are, and how credible your analysis is. Second, it should allow other researchers to replicate your study and repeat whatever it is that you have done.

Unfortunately, as any experienced researcher knows, these goals are not always met. More often than not, the authors of a scientific paper do not explain the procedures that they have used in enough detail, even if there is a Supporting Information document with an unrestricted page count. It happens all too often that when the reader attempts to understand in detail how the authors have arrived at their results, she has to give up because that information is simply not there or it is too patchy.

Not being able to understand a paper’s methods or to replicate its pipeline leads to many problems. First, this contributes to the replicability crisis and therefore erodes the very foundation of science, the scientific principle itself: only those results that can be replicated by others can be taken as facts. Second, selling your discovery to the scientific community will be hard if your fellow scientists cannot trust your findings because they do not understand how they were obtained. Third, if your pipeline—from data collection to analysis—contains new methods or ideas, those will not be adopted by anyone unless they are clearly explained (or even wrapped up and served on a plate, say, as a software package). This leads to many lost citations and your work not being discovered. If you release data, someone can also use it for things that you didn’t think of, and if you release software, there will always be someone who needs it.

So please do take replicability and reuse seriously. Explain what you have done in as much detail as possible. Release your raw data. Release your intermediate results. Release your code. Reveal everything. Hide nothing. Be a good scientist. Don’t be an evil scientist.

If you release everything that there is to release, you will probably need to use external repositories. Some journals, however, do allow submitting supplementary data and code files, to be published together with the article. If you are thinking of hosting the data and code yourself, consider that we are talking about the scientific record here: your paper, your data, and your code should, in theory, be available forever. And forever is a mighty long time, as the late artist known as Prince once put it. It certainly is longer than the lifetime of the URL that points to your www homepage on your university’s server, or of the server daemon that runs on the Linux machine in your bedroom closet. So no DIY here, please—always use long-term data and code repositories, like Zenodo. While even those might not last forever, they’ll last longer than any self-hosted repository. Note that even GitHub is not futureproofed: it is run by a commercial company that can become extinct just like any other company.

Let’s return to the paper itself, and move from function to form. First, where to describe materials, data, and methods? This, of course, depends on the journal, and there are many options. The top-tier journal style (think PNAS, Nature, etc) is to have Materials and Methods as a separate section at the end of the article, as an appendix of sorts. In these journals, methods are only briefly described in the main text and the reader is referred to the Materials and Methods appendix for details. While writing a paper this way may at first feel difficult, this structure does make sense: the short letter format is all about the story, and technical details that would get in the way of the story are pushed aside. This may make writing feel harder because one cannot hide behind technical details: there has to be a story. However, beware of the dark side: referring the reader to the Materials and Methods section where only superficial details are given and where the reader is further referred to the Supplementary Information that adds detail but still lacks essential information, or where the limitations of the chosen methods are hidden in a subordinate clause on page 28. This structure makes it dangerously easy to sweep something under the rug. Which is why it often happens.

So if you are writing for one of these journals, do resist the dark side: do not hide problems in the SI. Other than that, just strive for clarity in the Materials and Methods. Typically this section comprises independent subsections for different items, so there is not much storytelling involved. In the main text, when talking about methods, describe their purpose, not their details: “we measure the similarity of X and Y with the help of (insert name of fancy similarity measure), see Materials and Methods for details.”

Then there is the style common to biomedical journals, where Methods are described in all their detail straight after the Introduction. This makes it easier to describe everything properly and more difficult to hide problems, which is good. The downside is that being hit by several pages of painstakingly detailed method descriptions is something of a turn-off: the story suffers. While this cannot be entirely avoided, it helps if you remember to provide context: begin each subsection by reminding the reader why this data set was collected, why this experiment was done, or why you are going to next describe some mathematical methods. Often, this is not more difficult than simply saying that, e.g., “to measure the similarity of X with Y, we need some well-behaved distance measure for probability distributions that…” and then describing the chosen measure.

The third way that is common, say, to the journals of the American Physical Society, is to happily mix methods with results, explaining how things were done and what the outcome was without making a distinction between the two. In this case, things like experimental setups or data collection procedures may still be explained separately, but typically all mathematical and statistical methods are described together with the results. In my view, this makes writing a smoothly flowing story easier than the biomedical style. It is easier to motivate the methods by saying that “next, we’ll investigate X, and to do that, we need to do Y, and look, here’s the result”. In the biomedical style, this connection is harder to make because the methods and results are separated, so one has to focus on making sure that the reader understands why the methods have been chosen and why the reader should understand their details.

Before concluding, let us return back to being good versus being evil, and talk about discussing the limitations of your methods. All methods have limitations, as every scientist knows, and it is best to lay these out in the open. In my view, the Methods section is the best place for doing this: while even minor limitations of methods are often discussed in the Discussion section, it feels more natural if they are addressed when the methods are introduced. Strangely, this even feels more honest. First, at least to me, it feels a bit like having been cheated if I have read a long paper, and only in the last paragraph, it is mentioned that by the way, we’re not sure that things work the way we just told they would. Second, it is easier for the writer to explain the limitations together with the methods. Third, it is also easier for the reader to understand the limitations and their implications if the details of the methods are fresh in her memory.

When addressing limitations, you should tread carefully: being honest is different from making it sound like your study is flawed. Joshua Schimel’s “Writing Science” introduces a great principle: say but, yes instead of yes, but. Instead of saying that your quite clear results would be much more detailed if your experimental setup would have a higher resolution (or similar), say that even though the resolution of your experimental setup is limited, your results are quite clear. The latter has a much more positive ring to it, although both sentences have the same information content. So don’t make it sound like there is something wrong with your work—if there is, fix it first, before writing your paper.

Coming up next in this series on how to write a scientific paper: how to write the results section.