Tips for Scientific Writing in the Clinical and Cognitive Sciences
Scientists rarely receive formal training in scientific writing. Many of us learn on the job through feedback from mentors and reviewers. I am not an authority on scientific writing, and it is a skill that does not come naturally to me. Nevertheless, I have struggled to improve and have found some simple guidelines that help. The tips below represent my own personal preferences, any of which might be wildly incorrect. Some of these suggestions might make you angry, and that’s okay. I invite you to email me with any suggestions. If I use your suggestion, I will acknowledge you on the list of contributors. Thanks to everyone who has pitched in to make this such a helpful resource!
Contributors include: Jamie Reilly, David Kemmerer, Bonnie Zuckerman, Dan Mirman, Emily Myers, Ingrid Olson, Barbara Malt, Mike Dickey, Gayle DeDe, Laurel Buxbaum, Lila Chrysikou, Maura O’Fallon, Rebecca Alper, Felicidad Garcia, Matt Sayers, Steve Lepore, Elizabeth Heller Murray, Jacqueline Gore, & Anita Bowles
kickstart the process by drafting a title page
Start a new manuscript by roughing out a title page. Include a tentative title along with co-authors and institutional affiliations. This helps me kickstart my writing by overcoming anxiety about getting started. I am reasonably confident that I can type my own name. Once that step is behind me, I now have the shell of a document. I can then glimpse at a title page and imagine the glory of eventually sending it off for peer review. This weird little step works for some odd reason.
use an outline and practice summarizing your story
Effective scientific writing conveys complex ideas using a linear and clear progression. Terrible scientific writing reads like a laundry list (e.g., We did X, then we did Y). A good scientific paper has a beginning, middle, and end. Outlines help to organize this flow and keep you focused. Do not skip the outline stage. It is the blueprint for building a complex structure. Once you feel solid about your outline, practice orally summarizing your story to a friend. Tell the story of your paper in about two minutes. If that story is incoherent, meandering, or unconvincing, then stop and regroup.
methods and results sections first
Methods and results sections tend to be formulaic. Get these out of the way first. Methods describe what you did. Results describe what you found. Do not confuse “what we did” vs. “what it all means”. The interpretation of what it all means belongs in the discussion. An informative methods section is like Goldilocks tasting the porridge. Just the right amount of detail is needed — not too little and not too much. The bar for a method section is whether another researcher can replicate your experiment with the level of information given. Err on the side of too much detail. Make use of supplemental material and consider making your stimuli publicly accessible via data repositories like the Open Science Framework.
rough out the introduction and general discussion: get paint on the canvas
Now that you have a title page, methods, and results it is time for the tough part. Hammer out a rough draft of your introduction and general discussion. This rough draft should align with your outline. Avoid censoring yourself at this stage. You will eventually wordsmith, line edit, and improve paragraph transitions later. Just get your raw creative thoughts down on paper at this stage.
do not plagiarize
The archetype of plagiarism is copy-and-pasting text verbatim from one document into another. This does occasionally happen, but most plagiarism is more subtle. Careless paraphrasing (i.e., changing the original by a few words and failing to cite the source) is probably the most common case I see. Err on the side of caution when you are deciding whether a statement needs a citation. A good guideline is that any assertion you make that is not common knowledge or your own opinion requires a citation. When in doubt, cite. Reference managers like Zotero make the bibliography process much easier. Take this seriously. Citing work can be a PITA, but plagiarism is a form of fraud that can ruin your career. To learn more about what constitutes plagiarism, visit plagiarism.org. Another interesting fact about plagiarism — it is even possible to plagiarize yourself. Thanks to Emma Carey for sending along this list of plagiarism checkers (including some free versions).
show-don’t-tell your reader
Students of Creative Writing are taught to ‘show don’t tell’. This guideline also applies to scientific writing. For example, “It is very important to assess digit span in Alzheimer’s Disease” violates the show-don’t-tell rule. You have explicitly told the reader something they should infer through a persuasive literature review. The show-don’t-tell rule is a stylistic suggestion that does not meet with universal agreement. Some people prefer the more direct approach (e.g., Drs. Emily Myers, Anthony Dick).
do not stigmatize — ‘people first’ vs. ‘identity first’ expression
This sounds obvious, right? You are a good person who believes that others should be treated with respect. Of course, you would not stigmatize anyone intentionally. Many people have strong preferences and compelling reasons for how they wish to be referred. Identity is a nested, multilevel construct (e.g., gender, race, disability, nationality). There is no uniform standard for addressing people across the spectrum of their identities. People-first language shifts disability away from the forefront of a person’s identity. Rather than claiming someone is an ‘aphasic’ or a ‘blind person’, you might better describe them as a ‘Person with Aphasia’ or a ‘Person who is Blind’. That is, you are a person first, and a disability does not define you (read more here). In contrast, others prefer identity-first language (e.g., trans person, African American, Native American). These distinctions and their many exceptions are confusing and often unintuitive. It is crucial to get them right though, and a few minutes of Googling is worth the effort to prevent accidentally stigmatize someone.
do not stigmatize part II - race and ethnicity
The Chicago Manual of Style recommendation for referencing race includes capitalization. For example, ‘He is a Black man.” or “She is White.”
embrace parsimony
Occam’s Razor also applies to scientific writing. Be parsimonious. Do not use six sentences when three words will do. The concepts we seek to convey are often complex. Verbosity and obscure vocabulary do not strengthen your case. Strive to be direct, accessible, and straightforward. If you can cut a word or phrase, do it. Steven King once said, “the road to hell is paved with adverbs”. This is especially true in scientific writing. Cut adverbs. Also cut modifiers (e.g., an interesting phenomenon) and intensifiers (e.g., very very impaired). Do not qualify a finding as interesting or unfortunate. Let the reader infer that the finding is either interesting or unfortunate for herself (i.e., show don’t tell). Some people love Strunk & White for their emphasis on parsimony, but it feels a bit too rigid for me.
mind the citation bias
Is your citation list composed exclusively of old, dead white men? Scan your references. You might be surprised at how your knowledge base is skewed toward a particular viewpoint. For example, much of the early work on intelligence and psychometric measurement (Galton, Pearson, Fischer) was informed by scientists with a vested interest in advancing eugenics. If you skip this step, you could miss diverse viewpoints and inadvertently perpetuate the exclusion of dissenting views and/or lived experience. Here’s a great resource for checking on citation bias in your work.
do not confuse hypotheses, research questions, and predictions
People often conflate hypotheses, predictions, and research questions. A hypothesis is a falsifiable explanation for a phenomenon that cannot be explained by extant theory. A hypothesis is not a question or a prediction. A hypothesis is a tentative directional explanation that awaits empirical support. Here are some examples of each. Hypothesis (specifies mechanism): High iron content in groundwater causes insatiable hunger for Spaghettios. Prediction (mechanism in action): Cities with high iron content in their drinking water supply will show higher Spaghettio consumption. Research Question (how to test the prediction): Do cities with high iron content in their drinking water consume lots of Spaghettios? — KEEP THESE STRAIGHT.
Make your data and code publicly accessible
You have completed a study, and the information is now out in the world. That’s great. However, it is not enough. Be a good scientist. Make your data freely available to others. Annotate your computer code, and make this available too. My lab posts all of our research and as much de-identified data as we can on the Open Science Framework. This serves many purposes. I like to think that our code and stimuli might be useful to someone someday. There is also some thinking that open science will promote better transparency and reliability in cognitive science - not sure this is the case but here’s to hoping.
avoid jargon and ‘homebrew’ acronyms
People often feel self-conscious about using plainspoken language and will compensate by using jargon and complex vocabulary. Jargon does not bolster confidence in your message. Jargon makes life more difficult for your reader. Similarly, many scientists extol the space-saving virtues of acronyms. People will often generate long lists of novel acronyms known only to them. Esoteric acronym usage is extremely taxing on the reader’s working memory. The value gained in saving space is lost by obscuring the message. An over-reliance upon jargon, obscure vocabulary, and acronyms does not make you look like a gifted insider. This makes you look like a nimrod.
absence of a difference is not equivalence
Many of us are trained in the framework of null hypothesis significance testing. The null hypothesis specifies no difference between one or more experimental conditions. People often erroneously equate ‘no difference’ with ‘equivalence’. There is a big problem with this logic. Namely, if you fail to reject the null hypothesis you cannot conclude equivalence. Consider, for example, the American court system. When people are tried for a crime, they can be found either guilty or not guilty. A defendant is NEVER found innocent. That’s how the null hypothesis works. If you have an explicit hypothesis about equivalence between conditions, you must conduct an equivalence test. Options are Bayesian region of practical equivalence (ROPE) testing, TOST tests (test of one-sided significance), and probably many others I’m not aware of.
Know your audience
Scientific papers are not typically didactic. Your aim is not to teach but to persuade or convince. Persuasion should be subtle, and the data should do most of the talking. Craft your argument with your audience in mind. Your reader does not have privileged access to your brain. She might not even know your field. Step outside yourself and think about your story as a theory of mind exercise. Carefully consider the aims and mission of the journal you have in mind. Tailor your work to that audience.
“because This has never been done before” is not a compelling justification
“This study is important because it has never been done before” Many things have never been done before. Most things have not been done before because they are stupid. Never done before is not a compelling rationale for investigating anything.
Use a reference manager. be meticulous about formatting your bibliography
There are great options out there for reference managers. Zotero is free, open-source, and it works well. You can import citation data and manuscripts from PubMed or Google Scholar directly into your library. I annotate each reference using keywords and insert references into my documents as needed. Zotero can reformat your references to virtually any journal format in one click. Ensure that your references adhere to the required format (e.g., APA7) and that all of your references are uniform (e.g., full names of journals, titles lowercase, DOIs present, etc.) and complete. This takes time and effort. Professional tennis players are often seen tweaking their racquet strings, and oboe players are notorious reed fiddlers. The research equivalent is editing and maintaining your reference library.
ask for feedback, give feedback, don’t shoot the messenger
Your writing will improve with practice but not just any old practice. You need expert feedback. Unlearning engrained habits is challenging. This takes dedication, fortitude, and thick skin. Try not to take comments personally. Ask for feedback from people whose writing you admire. Be generous in helping others who want your feedback. Get out of your own head. You might think that your story is clear and compelling only to have an outsider perceive it as a hot mess.
avoid contractions
Scientific writing need not be overly rigid or formal. However, in my humble opinion, contractions (e.g., don’t, isn’t, wouldn’t, and shouldn’t) are just too colloquial.
avoid shitty figures and plots
I’m talking to you Excel. Take time to learn R or Matlab for plotting and Adobe Illustrator for figure layouts. I realize that this recommendation isn’t entirely aligned with scientific writing. However, a poorly executed figure can detract from the quality of your work. You would not submit a paper in Comic Sans 16pt, but Excel plots are roughly the equivalent of Comic Sans. It turns out there is a whole science behind generating effective displays of quantitative and anatomical figures Edward Tufte, the Godfather of this science, argued for radical minimalism in plots and figures. Tufte advocated highlighting the data above all else. He even proposed maximizing the data-to-ink ratio of figures.
Go on a ‘which’ hunt — replace ‘which’ with ‘that’
Scour your paper. Go on a ‘which’ hunt. Kill every ‘which’ you possibly can (for a dissenting view, see Dr. Gayle DeDe).
find a writing style you admire
Some high-profile researchers are terrific writers. Their arguments are so clear and the data are so compelling that their argument just seems self-evident (i.e., of course the world works like that!). They make it all seem simple. Other scientists write like shite and could not possibly care less whether the peasantry can understand them. Crack the nut and figure out what is you like about your favorite scientific writers. You are a scientist, right? Dissect prose, and pay attention to the aspects of style you currently struggle with. This process demands insight into your own strengths and weaknesses and a motivation to improve. You obviously don’t want to copy someone else’s style, but you can learn a great deal from deconstructing the work of others.
Be consistent (courtesy of Dr. Anita Bowles)
Consistency is essential in scientific and technical writing. It is often necessary to repeat specific terms or phrases throughout your manuscript. Explicitly define technical terms before using them in an idiosyncratic or esoteric way. For example, you might be interested in specifying a broad construct such as semantic memory. Do not assume that your reader has access to your mind. Clearly state your operational definition of semantic memory before you jump into theory and/or methods. Be consistent when referencing the particular construct the unique way you have operationalized it.
active > passive voice
The passive voice oozes humility and an external locus of control (e.g., “errors were made”). In contrast, the active voice (e.g., “We ran this study”) somehow feels macho and egocentric. Go with the active voice. Simple active declarative sentences (S-V-O) reduce cognitive demands on your reader and can be written in ways that do not make you look like an asshole.
use whitespace
Have you ever read anything by David Foster Wallace? Infinite Jest had paragraphs that spanned five or more printed pages without a break. You need to read Hunger Games or People magazine next just to catch your breath. It’s like eating at a fancy restaurant and then stopping at Macdonald’s on your way home. Whitespace is the equivalent of a breather. Breathers facilitate emphasis and memory encoding. Use whitespace strategically. Do not feel compelled to fill every micron of space with characters.
Chekhov’s Gun & Pandora’s Box
Anton Chekhov was a renowned Russian author and playwright. Chekhov once stated, “If you say in the first chapter that there is a rifle hanging on the wall, in the second or third chapter it absolutely must go off.” This is now known as Chekhov’s Gun principle. All themes introduced within a story MUST be used, and any unused element should be discarded. Checkhov’s Gun also applies to scientific writing. Do not open a door without being prepared to close that door. Your outline will help you to avoid opening extraneous doors and taking your reader on tangents. Reviewers are critical of unresolved or extraneous points. Be willing to cut text that is not central to your message even if you are fond of that text.
paraphrase; don’t quote
My high school papers consisted of >75% block quotes. Just as hammers are essential for carpenters, block quotes are essential for reaching minimum page requirements in elementary school. In contrast, scientific papers rarely contain direct quotes. Paraphrase instead and make certain you acknowledge the source. You might guess that the best way to acknowledge Jane Smith, Ph.D. as the source might be something like, “Dr. Jane Smith, Ph.D. said”. Wrong again! We typically refer to authors only by their last names (e.g., “Smith (1999) explained…”). Groups of authors are often referenced with et al (e.g., Smith et al. (1999) showed that…) with some noteworthy exceptions. Et al. is a Latin abbreviation for et alia, meaning ‘and others’.
proofread then proofread again
Writing is difficult. Take the time to proofread. Have others proofread your work (and return the favor). Grammatical errors, sentence fragments, spelling errors can all be caught and repaired without too much pain. Grammarly offers a free Chrome plugin that helps. I am in fact using it right now. Sometimes it is difficult to catch errors when you are in your own head for too long. There is a lot to be gained by stepping away from your work and returning a day or two later with fresh eyes. Writing and proofreading in English is especially challenging when English is not your native language. Publishers are beginning to offer more assistance with English language proofreading and editing for scientists whose native language is not English. If this situation applies to you, consider reaching out to the Editor-in-Chief of the journal you are writing for and ask what resources are available.
strike semicolons and exclamation points
Strike semicolons and exclamation marks. When joining two independent clauses with a conjunction, precede that conjunction with a comma.
Mary completed a naming test, and then she ran an MRI study. (yes to comma)
Mary completed a naming test and ran an MRI study. (no to comma, second phrase is subordinate)
oxford commas
Oxford commas are so controversial that there’s even a song about them. An Oxford comma sets off the final item in a sequence. Consider the following example: “We bought apples, oranges, and pears.” The comma after ‘oranges’ is known as an Oxford Comma. Another loose guideline for comma placement involves reading your text aloud. Place commas where you would naturally take a breath or pause.
adhere to statistical reporting guidelines for your field
Scientific research seeks to build upon what has been done before. It is important that you check the standards for your field for how quantitative results should be presented. This might include confidence intervals, effect sizes, or other important descriptive statistics (e.g., variance) that promote meta-analyses and replication studies. Applebaum and colleagues recently published an outstanding guide, outlining these standards for psychology.
use a sensible filenaming convention
I am not talking about how to choose a title for your paper. I am talking literally about what to name the document file on your hard drive. During my Ph.D. program, I had hundreds of partially completed drafts of different papers all named some derivative of ‘Paper.docx’. I stored all these files in one folder called ‘Papers 2004’. If a document was really important. I might have named it ImportantPaper.doc. I had no idea how much I was kneecapping myself until I revisited the ‘Papers2004’ folder during my postdoc in 2006. I could not find anything. We have created a separate webpage with suggestions on how to manage files. Make sure you name your files using a systematic convention. There should be no periods or spaces in your filename except to mark the file extension (e.g., .docx). PLEASE store your files on a cloud drive (e.g., Google Drive) with auto-backup. A good test of data redundancy is if you can chuck your laptop in the Delaware River with no data loss.
frame limitations as opportunities
This might sound like I am trying to sell you a used car. So be it. It is crucial that we acknowledge limitations and weaknesses in our papers. Yet, all too often this section sounds like a groveling apology. Framing is the impression you create with particular word choices. Consider, for example, acknowledging that your study was relatively low-powered. You could say, “Our failure to include a substantial number of participants was a major weakness”. Alternatively, you could positively frame this limitation as, “Future studies will benefit from a larger and more diverse participant pool.” Readers are often quite sensitive to framing effects. Framing is one of those robust cognitive heuristics that behavioral economists love.
never use any of these words in a scientific paper
Never use any of these words in a scientific paper. You might find a beloved word or two in this list. It’s okay to be angry:
plethora, passionate, passionately, importantly, interesting, interestingly, very, hugely, fascinating, unfortunately, amazing, easy, tragic, tragically, terrible, never, dumb, retarded, defect or defective (in reference to a person), astonishing…
never use any of these phrases in a scientific paper or review
We are the first to report
is trending toward significance
is marginally significant
is highly significant
As I previously mentioned
As I already stated
As I above stated
As stated above
dampened my enthusiasm
big problem
we proved our hypothesis (*you can never prove a hypothesis)
we utilized (*just say used)
confined to a wheelchair
afflicted by
words and phrases commonly abused or misused
principle vs. principal: Principal means primary (e.g., principal investigator), whereas a ‘principle’ is an idea.
subjects vs. participants: Rats are subjects. People are participants.
affect vs. effect: Affect (n) means nonverbal cues for emotions or mood (e.g., He has a flat affect.). Affect (v) means influence or change (e.g., He affected the outcome.). Effect (n) is the result of the change (e.g., Poor concentration is an effect of sleep deprivation.)
i.e. vs. e.g.: ‘i.e.,’ is a Latin abbreviation for id est, meaning ‘that is’. Example: I’m tired of all his bullshit (i.e., I’m sick of him). ‘e.g.,’ is a Latin abbreviation for exempli gratia, meaning ‘for example’.
your vs. you’re: ‘Your’ is possessive (e.g., “your car”). ‘You’re’ is a contraction meaning ‘you are’.
penultimate: Penultimate means second to last, you dummy
Acknowledge that you have read the guidelines
If you have been referred to review these guidelines by a particular instructor, please fill out the email form below. Select your instructor from the dropdown menu.
References worth checking out
Appelbaum, M., Cooper, H., Kline, R. B., Mayo-Wilson, E., Nezu, A. M., & Rao, S. M. (20180118). Journal article reporting standards for quantitative research in psychology: The APA Publications and Communications Board task force report. American Psychologist, 73(1), 3. https://doi.org/10/gctpzj
Bassett, D. (2020). Uncovering Racial, Ethnic, and Gender Imbalances in Reference Lists. Conversations on Gender and Inclusive Excellence in Academia, Johns Hopkins University.
Mensh, B., & Kording, K. (2017). Ten simple rules for structuring papers. PLOS Computational Biology, 13(9), e1005619. https://doi.org/10/ckqp
Strunk, W., & White, E. B. (2007). The Elements of Style (4th ed. edition). Penguin Books.