10  Scientific Writing and Presentations

When it comes to making an impact, clear communication is as important as good analysis. If your work can’t be understood, it can’t be used — by other researchers, by policymakers, or by the communities we serve.

10.1 Writing Philosophy

Write early and write often. Don’t wait until your analysis is “done” to start writing. Writing is thinking.1 Putting your ideas into sentences forces you to identify gaps in your logic, clarify your assumptions, and figure out what your results actually mean. Write as soon as you start — in the beginning this may just be a record of your methods. Then when you have some initial results, these may be high-level summaries of your results. The key is that the frequent interaction between coding, analyzing, and writing will help you think through the manuscript.

Write for your reader, not for yourself. Assume your reader is smart but does not know your specific problem. Avoid jargon if possible. Define terms the first time you use them. Make your argument’s structure explicit — don’t make the reader piece it together.

Revise, revise, revise. Good writing is rewriting. First drafts are supposed to be rough. I can help you edit a bad first draft into a good manuscript but I can’t help you edit nothing. The goal of a first draft is to get your ideas onto the page; the goal of subsequent drafts is to make those ideas clear, concise, and compelling. Plan for multiple rounds of revision, and take feedback from coauthors and labmates seriously.

10.2 Structuring a Scientific Paper

Most of our papers follow the IMRaD structure: Introduction, Methods, Results, and Discussion. This is the standard in epidemiology and public health, and for good reason — it mirrors the logic of the scientific process. That said, not all journals follow IMRaD, so deciding your submission pathway early will help you identify the appropriate format.

Introduction: Start with the big picture (why does this matter?), narrow to the specific gap in knowledge (what don’t we know?), and end with your research question or objective (what did we do about it?). The introduction should be concise — usually 3-4 paragraphs. Don’t write a literature review.

Methods: Describe what you did in enough detail that someone could reproduce your analysis. This includes data sources, study design, statistical methods, software and package versions, and any sensitivity analyses. Be precise. If necessary, use a methods appendix to be even more detailed. The EQUATOR Network has reporting guidelines for specific study designs (STROBE for observational studies, PRISMA for systematic reviews, etc.), use them.

Results: Present your findings clearly and objectively. Lead with the most important results. Use tables and figures effectively — they should be able to stand alone with their titles and footnotes. Save interpretation for the Discussion — just present the findings here.

Discussion: Summarize your main findings (briefly), interpret them in context, discuss limitations honestly, and explain what it all means. Don’t overstate your conclusions. If your study is observational, don’t imply causation unless your design supports it.2 End with concrete next steps or policy implications where appropriate.

10.3 Writing Process in the Lab

Here’s how a typical paper moves through the lab:

  1. Outline: Before you write, create a structured outline with the key points for each section. Share it with me for feedback before you start drafting. This saves enormous time compared to writing a full draft that needs restructuring.

  2. Key results: I like to write Science-style results in the beginning. Something like “Here is Key Result 1. We find X, Y, and Z. See this related figure.” I do that for 3-5 key results. Then, I can figure out how the key results relate to each other (should they be merged or separated), which ones are most salient and should be put up front, which ones repeat what is already known, etc. Understanding your key results will help you organize the rest of your paper.

  3. First draft: Write the full draft in whatever document editor you want (Quarto, Word, LaTeX, Overleaf, Google Docs, md). Don’t agonize over word choice yet — focus on getting the logic and structure right. Make sure the key results are supported.

  4. Internal review: Share the draft with me and at least one labmate for feedback. We use tracked changes or GitHub pull requests for manuscript review, depending on the project.

  5. Revision: Incorporate feedback, tighten the prose, check that all numbers match between text, tables, and figures. At this stage, you may need to convert to Microsoft Word or whatever our collaborators prefer. We make it as easy as possible for our collaborators to provide useful feedback and to minimize the time burden on them.

  6. Coauthor review: Share with all coauthors. Allow at least two weeks for review (more for senior coauthors with packed schedules). Our full publication workflow is in the Doing Science chapter, and there’s a pre-submission checklist in Checklists (forthcoming).

10.4 Writing Resources

If you want to improve your scientific writing (and you should — it’s a lifelong process), here are some resources I recommend:

  • Hume Center for Writing and Speaking at Stanford — free individual consultations. Use them, especially for your first paper or grant.
  • Schimel, J. (2012). Writing Science: How to Write Papers That Get Cited and Proposals That Get Funded. Oxford University Press.
  • Sword, H. (2012). Stylish Academic Writing. Harvard University Press.
  • Ten simple rules for structuring papers (Mensh & Kording, 2017, PLOS Computational Biology)

For more suggested readings on scientific writing, presentations, and posters, see the Further Reading chapter.

10.5 Giving Talks

Presentations are how we share our work with the broader scientific community, get feedback, and build our reputations. A good talk is not a paper read aloud3 — it’s a story told to a room full of people. Dan has great slides on how to give a good talk — follow his advice.

Structure: The best talks follow a narrative arc. Start with a compelling question or problem (why should the audience care?). Walk through your approach and key results. End with what it means and what comes next. A common mistake is spending too much time on methods and too little on the “so what.”

Slides: Keep them clean. One idea per slide. Use large, readable fonts. Figures should be legible from the back of the room. Avoid dense text — your slides support your narrative, they don’t replace it. If you find yourself reading your slides, you have too much text.

Practice: Practice your talk out loud at least twice before giving it.4 Time yourself. Practice in front of the lab — we do dry runs for all major talks, and the feedback is invaluable.5

Handling questions: Questions mean people are paying attention — that’s a good thing. Listen to the full question before answering. Repeat back the question in your own words. It’s perfectly fine to say “I’m not sure, but here’s how I’d think about it.” It’s also fine to say “I don’t know.”

10.6 Poster Presentations

Posters are a different medium than talks. They’re conversation starters, not self-contained documents.

  • Keep text minimal. Your poster should be readable from 4-5 feet away. Use large fonts, clear headings, and lots of white space.
  • Lead with your findings. Put your main result front and center. People will scan your poster in seconds before deciding whether to stop and talk.
  • Prepare a 2-minute pitch. You should be able to walk someone through your poster in about two minutes, then pivot to deeper discussion based on their questions.6
  • Figures matter most. The figures are the most important part of a poster. If someone only looks at the figures, they should get the main story.

10.7 Science Communication for Broader Audiences

Our research on health inequities, drug overdoses, and structural determinants of health is inherently relevant to public discourse. When we communicate our work beyond academic journals — to journalists, policymakers, or the public — we should do so clearly and responsibly.

  • Translate, don’t dumb down. Respect your audience’s intelligence while removing jargon and technical assumptions.
  • Be precise about uncertainty. Don’t overstate findings. If the result is suggestive rather than definitive, say so.
  • Stick to what you know. Don’t venture into areas outside your expertise. It’s fine to say “that’s outside my area of research.”
  • Preempt misinterpretation. Think about how your findings could be taken out of context and address that proactively.
  • See the Communication chapter for more on public scholarship and media interactions.

  1. This is also why I’m cautious about outsourcing writing to AI tools — see the Ethics chapter for our AI policy. AI has its uses, but it’s not clear to me that scientific writing is one of them.↩︎

  2. This is the section where you get to write ‘future research is needed’ — the only time that sentence is acceptable in a paper.↩︎

  3. Though, to be clear, in some fields a “talk” really is literally reading a paper out loud.↩︎

  4. I still practice every talk in a timed setting at least 10 times before giving it.↩︎

  5. If your dry run goes perfectly, you probably weren’t pushing yourself hard enough. If it goes terribly, congratulations — now you know what to fix.↩︎

  6. Nobody has ever complained that a poster pitch was too short.↩︎