Writing the discussion section

The purpose of the discussion section is straightforward enough — to describe the significance of your results and what they mean to your field of study. However, organizing this section can be a challenge because the structure is much more loose than in the methods, results or even the introduction.

Structuring the discussion

A good place to start is with a one-paragraph review of your major findings that ends with a one- or two-sentence statement about their overall importance. You can then expand from there by answering the following questions:

• How do your results fit into the larger context of what is already known about your subject? Do your findings agree or contrast with the findings of others?
• What are the theoretical and/or practical implications of your study? How are your findings an advance over what others have shown?
• What are the limitations of your study? Could these limitations have biased your results in any way?
• What future directions does the research suggest?

Note: Keep in mind the inverted pyramid style. That is, it’s generally a good idea to lead with the most important points you want to make and follow with points of lesser and lesser importance.

Common pitfalls to avoid

Overstating the significance of the work. Many articles on scientific writing will caution you against this, and with good reason. However, an even bigger pitfall may be the one directly below.

Understating the significance of the work. While it’s true you should take care not to oversell things, it’s also true that young scientists tend to downplay the importance of their research. So, do think about the larger meaning of what you’ve done and then don’t be afraid to say it. As William Wells (2004) writes, there probably was a bigger idea behind your work than the possibility of protein X and Y binding to one other, so “make sure you convey that reason and that excitement.” In other words, readers should not come away from your paper thinking, “so what?”

Being vague. Related to the point above, if you make general statements such as, “We believe our model can help experimental biologists understand their own systems better,” or “Our findings are valuable for the future design of bacteria that can do X and Y” be sure to explain what you mean. What specific insights will your model offer biologists? What exactly will your models contribute to the future design of microorganisms? If you’re hazy about this, reviewers are likely to ask you to provide some explanation and evidence.

Failing to address the question or problem posed in the introduction. “The Introduction and Discussion should function as a pair,” say Day and Gastel (2006). “Be sure the Discussion answers what the Introduction asked.”

References

• Cetin, SA and DJ Hackam (2005) An approach to writing a scientific manuscript. J. Surgical Research 128: 165-167.
• Day, RA and B Gastel (2006) How to Write and Publish a Scientific Paper. Greenwood Press, Westport, CT.
• Ness, R. (2007) Writing science: The story’s the thing. Science Careers: http://sciencecareers.sciencemag.org/career_magazine/previous_issues/articles/2007_04_06/caredit.a0700047
• Wells, W (2004) Me write pretty one day: How to write a good scientific paper. J. Cell Biol. 165(6): 757-758.

Writing the results section

The purpose of the results section is to describe your findings as concisely and clearly as possible. In other words, be careful not to rehash your methods or discuss the meaning of your data in this section.

Organizing the results section

It often makes sense to lead with the findings that most directly address the question or problem you presented in the introduction, followed by results that are still relevant to your overall story, but secondary. One way to envision this type of organization is as an inverted pyramid (pdf), in which you start broadly with your most important findings (e.g., your model’s predictions) and then taper toward less significant results (e.g., validation data).

On the other hand, you might decide that readers need to know how your model works before they can grasp the predictions. In this case, you flip the pyramid right-side up again, i.e., start narrow with the model’s specifics and finish broadly with your major findings.

Whatever structure you choose, make sure each section and paragraph flows logically from one to the next. So, for example, although you may have completed your experiments in a certain chronological order, ask yourself if this is the best order for supporting the main message of your paper and getting readers to follow the story.

Common pitfalls to avoid

Including anything but your findings. To quote the catchphrase of the 1950s TV crime drama, Dragnet: “Just the facts, Ma’am.”

Including too many findings. Remember that the most effective scientific papers usually focus on a single story, message or bottom line. Thus, try to limit your results section as much as possible to those data that directly support the main point of your paper. If you find yourself trying to squeeze in many more, you might consider whether these additional findings should go into another paper.

• Repeating what is shown in the figures and tables. Your text should summarize what the figures and tables show, not go through them data point by data point.
• Losing the connection between the question/problem you posed in your introduction and the answer/solution: your results. In the best-written papers, this link is crystal clear.
• Failing to guide the reader. See the discussion below.

Guiding the reader

With the exception, perhaps, of Materials and Methods, the results section is the most detailed one in the scientific paper. So, while it’s true that this section should primarily present your findings, it’s also true that readers need transitions, summaries and other guideposts to make their way successfully through all your data. Below are three strategies for keeping the reader oriented.

• While you should avoid rehashing your methods in detail, do describe your overall approach briefly at the start of the results section, and at the beginning of each subsection, if needed.
• Offer readers a one-to-two sentence summary of your overall findings for each set of experiments or analyses, before launching into all the specifics.
• Describe briefly the logic behind performing experiments or analyses. For example: “Because A resulted in B, which is in the cascade of C (citation), we decided to see whether A was connected to C; therefore we subjected D to E.”*

Statements like these are not fluff; to fully understand your study, readers need a periodic reminder of where they are in the forest as they move through the trees. Otherwise, they can become hopelessly lost in all the details.

References

• Cetin, SA and DJ Hackam (2005) An approach to writing a scientific manuscript. J. Surgical Research 128: 165-167.
• Day, RA and B Gastel (2006) How to Write and Publish a Scientific Paper. Greenwood Press, Westport, CT.
• *Neil, US (2007) How to write a scientific masterpiece. The Journal of Clinical Investigation 117(12): 3599-3602
• Ness, R. (2007) Writing science: The story’s the thing. Science Careers: http://sciencecareers.sciencemag.org/career_magazine/previous_issues/articles/2007_04_06/caredit.a0700047

Tips for writing the introduction

Here are a few things to keep in mind when writing the introduction to a scientific paper. I developed this material for a writing workshop I recently held with my students.

Main purpose of the introduction
To present the problem (an “intricate unsettled question,” a “source of perplexity”) you’ve addressed and why it’s important. If readers don’t grasp the problem, they aren’t going to care about the solution, i.e., your research.

Components of the introduction

• Background that places your research in a broader context and tells why it’s significant (what is known)
• Description of the knowledge gap your study fills (what is not known)
• Statement of your problem/question/hypothesis
• Description of your approach and why you chose it
• Brief summary of your major findings
• Statement of the major implication of your work, i.e., the take-home message

A simpler approach to the introduction – as an answer to three questions (Cetin and Hackam, 2005)

• What do we know about this topic?
• What don’t we know?
• What are we now showing?

Structuring the introduction

One way to visualize the structure of the introduction is as two funnels connected by their narrow tips.

Common pitfalls

• Failing to state clearly the problem and its significance. In a study by Bordage (2001), “insufficient problem statement” was one of the top reasons reviewers gave for rejecting manuscripts.
• Treating your introduction like a literature review. The background you include should be just enough for readers to understand what your research problem is and why you chose it. Anything more could cause readers to lose interest.
• Not reviewing the literature carefully enough. Omitting a key paper could lead to embarrassment – and negative reviews.
• Underestimating the importance of the introduction. Although your findings and conclusions are the meat of your paper, the introduction is where you set the stage for them. Do this well and you’ll hook people into reading further. Do this poorly and even the coolest, most significant results may go unread.

References

• Bordage, G (2001) Reasons reviewers reject and accept manuscripts: The strengths and weaknesses in medical education reports. Academic Medicine 76(9): 889-893
• Cetin, SA and DJ Hackam (2005) An approach to writing a scientific manuscript. J. Surgical Research 128: 165-167.
• Day, RA and B Gastrel (2006) How to Write and Publish a Scientific Paper. Greenwood Press, Westport, CT.
• Wells, W (2004) Me write pretty one day: How to write a good scientific paper. J. Cell Biol. 165(6): 757-758.

The bottom line in scientific writing

One of the best short pieces I’ve found so far on scientific writing is, “Me write pretty one day: How to write a good scientific paper,” by William Wells.*

The author gives great tips for writing each section of the paper, but what I especially appreciate is his emphasis on defining the “bottom line.” Says Wells:

The first step with any manuscript is to define your bottom line. Be realistic about how much the average reader will take away from an article. Non-experts will retain at most a single message. Make sure you have one, and then repeat it over and over again—at the end of the Abstract, in the Introduction, in the Results, and in the Discussion.

Even if you agree with this, though, defining your main message can be tough. In recognition of this, Wells helpfully offers advice for discovering it:

To uncover your bottom line, ask some questions: What was the mystery that you wanted to answer at the start? Have you answered it? What first got you excited about this area of research? With any luck, it was more than the idea that proteins X and Y might bind to each other—there was probably a bigger idea that motivated and intrigued you. Make sure you convey that reason and that excitement.

Wells’s insistence on a single message is echoed in a 2006 article** from EMBO reports. In it, the authors state:

The primary function of a scientific paper is to transmit a message—to convince the reader and the community that this is important research. It is therefore a good strategy to first think about the message before sitting down to write.

My only quibble with these authors is that they appear to view the development of a message as something that happens strictly through thinking ahead of time — that is, before a single word is typed. What I’m trying to get my students to see is that it’s okay not to know the message (or the complete message) before sitting down at the keyboard, that writing can be a tool for discovering it. Or if they start with one main point, but a different one emerges as they write, that’s okay, too. In other words, nothing’s wrong with their process when things evolve or they realize their original thinking was fuzzy — that’s just the process!

*W. Wells. 2004. Journal of Cell Biology 165(6):757-758
**A.S. Bredan and F. van Roy. 2006. EMBO Reports 7(9):846-849

Making your abstract flow

A good scientific abstract not only needs to contain all the right pieces (background, problem statement, main findings, etc.), but it should also flow smoothly and seamlessly from one sentence and idea to the next. The value of this should not be underestimated, especially when you reflect on how quickly and impatiently most readers will be scanning your abstract.

So, how can flow be achieved? In this slide show (pdf), I offer some thoughts for creating it, along with some examples that illustrate these points.

How to write a good abstract

Your abstract isn’t just a mini-version of your paper, with sentences chopped from your introduction, methods, results and discussion, and pasted together. Instead, it needs to be greater than the sum of those parts, an argument that compels people to read your paper, attend your talk, or visit your poster. Easier said than done, right? Here are a few tips to guide you in writing an informative and compelling abstract.

State the problem. If readers don’t grasp the problem, they’re not going to care about the solution (i.e., your research) or get the significance of what you’ve found. So, the beginning of the abstract should contain the research problem, along with just enough background for readers to understand why it is an “intricate unsettled question,” a “source of perplexity.” Your findings should then be presented as a solution to the problem.

Have a main point. Although the style of writing is different, the abstract is sort of like a brief news item on your research, and news stories always have a main point. So, rather than trying to squeeze in as many findings as possible, write a few bullet points or “sound bites” about your most important data, and then shape the abstract around them. This can be a good thing to do even before you start writing your paper, as it can help you find your overall focus.

Target a broad audience. In my opinion, the abstract should be aimed at a wider audience than the paper itself, because you never know who’s going to pull up your abstract in online and database searches. Thus, it should contain few, if any, jargon terms or acronyms, and include adequate background information for scientists outside your field (this becomes more important, of course, when you’re publishing in widely read journals, such as PNAS). Having a main point will also help capture the attention of a broader audience.

Say what you found, not what you did. Statements about methods (we did this) can almost always be rephrased as statements about findings (we found this). It’s always more interesting to hear about results than methods, plus you’ll save on words.

Be explicit about the significance of the research. If you want to compel people to read your paper, don’t make them guess what your data mean. I think good abstracts always include a statement at the end about the significance of the work, the more specific the better.

Eliminate writing errors. When people are skimming text very quickly – as they are, of course, with abstracts – they tend to be even less patient than usual with writing errors and clunky, hard-to-decipher prose. So, make every effort to use good grammar, proper sentence structure and so on. Your abstract should carry readers along like a gently winding path. Making them hack through thickets of prose, on the other hand, will discourage them from taking the longer journey (i.e., reading your paper).

To summarize. To write an informative and interesting abstract: 1) State the problem; 2) Present your key findings (i.e., the main point), answering as you do how they address the problem; 3) State the overall significance of the research; 4) Provide background as needed, and make your writing as clear and accessible as possible.

To build good titles, break down a few

If you’re feeling unsure about what makes an effective title, one useful strategy is to find ones you like, analyze them and then try to emulate them.

To get you going, here are a few recent titles that I like:

• Quantitative analysis of mechanisms that govern red blood cell age structure and dynamics during anemia
• The role of anorexia in resistance and tolerance to infections in Drosophila
• Integrative analysis of transcriptomic and proteomic data from Desulfovibrio vulgaris: A non-linear model to predict abundance of undetected proteins

What do these titles have in common? They use simple (but not simplistic) words, avoid acronyms completely, and keep jargon to a minimum. They are grammatically correct and well structured, making for easy reading. They are also specific and unambiguous. For example, the first title names the research subject (red blood cells), what was studied (cell age structure and dynamics) and the larger context (anemia). And in the third: Desulfovibrio was the research organism, transcriptomic and proteomic data were studied, and the problem was finding undetected proteins. After reading these titles, I feel like I have a good grasp of the papers’ contents.

What these titles don’t do is preview the main result. Here are three that do:

• Rapid response of a marine mammal species to Holocene climate and habitat change
• The circadian clock in Arabidopsis roots is a simplified slave version of the clock in shoots
• Circadian clock genes contribute to the regulation of hair follicle cycling

Titles like these that make a statement don’t necessarily work for every paper, but they can be very effective at getting people’s attention. And that’s an important function of a title, of course: to get your paper noticed. So, in addition to being clear and concise, you want to make your title as broadly appealing as possible, without overstating things.

For example, “Circadian clock genes contribute to the regulation of hair follicle cycling” is a good, descriptive title. However, upon reading the paper, I learned that hair follicle cycling occurs on much longer time scales than the approximately 24-hour cycles normally regulated by the circadian clock. In other words, say the authors, “While circadian clock mechanisms have been implicated in a variety of diurnal (daily) biological processes, our findings indicate that circadian clock genes may be utilized to modulate the progression of non-diurnal cyclic processes.”

Thus, the authors might have tried to incorporate the study’s larger significance into the title, as in “Circadian clock genes contribute to the regulation of a non-diurnal biological cycle in mice.” While my title may not be the best alternative (for instance, it assumes people will know the circadian clock normally regulates diurnal cycles), it does offer a way to prevent people from reading “hair follicle cycling” and thinking, “so what?”

A few quick tips for writing titles

If there’s one part of a paper that hundreds or even thousands of people will read, it’s the title. And if your title is good, they may read further. So, what makes a good one? A good title tells readers exactly what your paper is about in the fewest possible words. But achieving this isn’t easy, so let’s look at some guidelines and examples.

Avoid abbreviations
Consider this title from a recent Bioinformatics paper: “Complex discovery from weighted PPI networks.” While many scientists in the authors’ field likely know what PPI stands for, remember that this title could be read by thousands of others. That’s why a better, more informative title avoids the abbreviation: Complex discovery from weighted protein-protein interaction networks.

Be specific
Often, authors achieve brevity at the expense of clarity and specificity. For example, in “Complex discovery from weighted PPI networks,” what does the word “complex” mean? I read it first as an adjective, that is, as “complicated” or “multifaceted.” But a quick look at the paper reveals that the authors are referring to protein complexes. Thus, a less ambiguous title would be: Protein complex discovery from weighted protein-protein interaction networks.

I also learned from reading their introduction that many methods exist for predicting protein complexes from PPI networks. So, it’s possible the word “weighted” isn’t sufficient to differentiate the authors’ new algorithm. Thinking about how to do this in a small number of words might be another way to make this title more useful to the reader.

Here’s another example: Neural basis of cold-seeking behavior in endotoxin shock. Not only does this sound a tad strange (i.e., like endotoxic shock is exhibiting the behavior), but it also doesn’t say anything about the model that was used (rodents? monkeys? people?). To me, “cold-seeking behavior in rats with endotoxin shock,” is much more informative — and it only adds two words.

Use good syntax (word order)
This rule can be easy to break when you’re trying to be brief, so take care. For example: “Comparative toxicity of fumigants and a phosphine synergist using a novel containment chamber for the safe generation of concentrated phosphine gas.” In this structure, it sounds like the phosphine synergist is using the chamber to generate gas. The problem is easily fixed, though, by rearranging the sentence and substituting a different word for “using”: Comparative toxicity of fumigants and a phosphine synergist during the safe generation of concentrated phosphine gas in a novel containment chamber.

How about this title: The environmental dependence of inbreeding depression in a wild bird population. While it contains no outright errors, the order of the words makes the meaning hard to grasp. A better title might be “Inbreeding depression in a wild bird population varies with environment” or something similar, so long as the confusing phrase “environmental dependence of inbreeding depression” is removed.

What’s the problem? Part 2

I’m reading a book right now called Engaging Ideas, by Seattle University English professor John Bean, and the following passage from chapter 1 caught my eye:

“Academic writing begins with the posing of a problem. The writer’s thesis statement is a tentative response to that problem, a ‘solution’ that must be supported with the kinds of reasons and evidence that are valued in the discipline.”

Bean makes this statement in support of his thesis: that incorporating writing into the undergraduate curriculum serves to promote critical thinking and problem solving in students. But it’s relevant to scientific writing, too, as I tried to convey (perhaps not as eloquently) in two recent posts: “What’s the problem?” and “Writing the introduction.”

So, once more for good measure: If you are ever wondering how to structure the introduction to a manuscript or what the introduction should include, just remember that the overall goal is to introduce your research problem, including its importance, and your proposed solution to your readers. (I’ve emphasized the reader here as a reminder to think about what background your audience will need to understand the problem). You will then go on to support your solution with the data and reasoning you present in the results and discussion sections.

For me, thinking about the introduction in this way creates a structure on which to hang all the pieces that this section is supposed to include: background, definitions, rationale for the study, research question, etc. In other words, it keeps me grounded in what the introduction is all about. I hope it does the same for you.

Writing the introduction

The results and discussion sections are undeniably critical pieces of a scientific paper. But don’t overlook the importance of the introduction. As I alluded to in my last post (What’s the problem?), the introduction is where you unveil your research problem and try to interest people in its solution. If you don’t hook the reader at this stage, chances are your results and discussion sections will never be read.

This isn’t tabloid journalism, however, where writers must resort to glitz and sensationalism to grab readers’ attention. Luckily, all that scientists really need to get engaged is a solid understanding of your research question and its significance. To help guide you in providing this, I’ve listed a set of questions to keep in mind. Consider them guidelines, rather than questions you need to answer explicitly or in the order listed. If you take a look at this set of sample introductions from recent papers, you’ll see that authors address these questions in different ways.

Suggested questions to cover in the introduction

1. What is the overall topic or problem you address in your paper? What background and definitions do readers need in order to understand this problem?
2. What is already known about this topic in the literature? What remains to be discovered?
3. What specific question or problem do you tackle in the paper? Or, what is your specific purpose?
4. What methods did you use to answer your question and why? (The “why” can be omitted if it’s not really needed)
5. Why is your system (genes, organism, proteins) a good model for addressing your question?
6. What are your main findings, and what are their implications or benefits? Or, how do they fill the gap in knowledge that you described earlier?