Recommendation 2
Sara Stoudt (chair), Victor Piercey, Larry Lesser, Dave Hunter
Emphasize effective written and oral communication of results from data, with attention to the scope and limitations of conclusions.
Producing results is only one step in learning from data; students must also communicate their findings clearly and accurately, both orally and in writing. This requires responsiveness to both the genre of communication (i.e., oral or written presentation, journal article, newsletter, conversation with classmates, op-ed, etc.) and the audience background.
Effective communication starts with understanding and conveying the provenance of the data—their source, completeness, validity, relevance, and currency—emphasizing their effect on conclusions. Inferences, visualizations, and analyses should be presented as evidence, not definitive proof, of conclusions. Conclusions drawn from data can yield a better understanding, and [iterations of the process] (https://amstat.quarto.pub/college-gaise/recommendations/recommendation-01.html){target=“_blank”}can continue to improve that understanding. Ethical communicationrequires transparently acknowledging limitations and contextualizing results at an audience-appropriate technical level—keeping the focus on conceptual understanding.
Effective communication requires clarity of purpose). For example, this purpose could be to inform, influence, problem-solve, motivate, or facilitate dialogue. To aid students in formulating the purpose of their statistical communication, we recommend the mnemonic “genre, audience impact statistical effectiveness.” There is a spectrum of audience background knowledge, ranging from general to more specialized, as well as a spectrum of forms of communication, from the informal or conversational—e.g., blogs or social media—to detailed research reports. Below, we suggest a range of sample activities to provide students with experience communicating along both spectrums.
Audiences also differ in ways other than statistical background knowledge. Their goals and power relations also inform decisions such as the level of detail to communicate and the location of the results in the broader story. Audiences can even be heterogeneous themselves. Students should also get practice with these nuances as they complete communication activities that span the spectra of audience and genre.
| genre/audience | 1. Other statisticians | 2. Other people with some math literacy (e.g., classmates) | 3. Other targeted audience (e.g., policymakers, business people, healthcare professionals) |
|---|---|---|---|
| A: Formal report/presentation | A1 | A2 | A3 |
| B: Executive summary/memo | B1 | B2 | B3 |
| C: Op-ed/persuasive speech | C1 | C2 | C3 |
| D: Blog post/edutainment item | D1 | D2 | D3 |
Meta assignment: Pick an assignment from the list below. Consider what would need to change in the communication approach as the audience moves from 1 to 2, from 2 to 3? What changes as the genre moves from A to B, from B to C, from C to D?
Sample Resource: “5 levels approach” of https://www.wired.com/video/series/5-levels
Meta assignment: Pick an assignment from the list below. Consider what would need to change in the communication approach as the audience becomes more mixed. For example, what if you need to write an executive summary for a team with some statisticians and some people with only a baseline math literacy?
A1: Consider the problem of small sample sizes from both a methodological and a representation point of view. Write a formal report or give a formal presentation that communicates answers to the following questions to an audience of other statisticians. What should statisticians be aware of in terms of the methods they use, and the conclusions they draw? Who may or may not be fully represented in data based on small sample sizes, and what are the implications of not considering these points of view in conclusions drawn from a dataset?
Sample Resource: https://data-feminism.mitpress.mit.edu/pub/h1w0nbqp/release/3
Sample Resource: https://pudding.cool/2020/03/census-history/
A2: For a piece of writing that you have completed for this class, whether this is an explanation of a p-value in a homework answer or a longer-form piece of writing, exchange responses with a peer and go through a peer review process. (Revise your work based on the feedback you receive.
Sample Resource: https://www.tandfonline.com/doi/abs/10.1080/10691898.2010.11889489 Sample Resource: https://dsc-capstone.org/2024-25/assignments/methodology/02/
A3: For a data analysis that you have completed for this class, give a presentation of your findings to a targeted audience of your choice. Use the time constraint provided by your instructor.
Sample Resource: “5 levels approach” of https://www.wired.com/video/series/5-levels Sample Resource: https://da4all.github.io/student-showcase/
B1: Write an abstract for a presentation based on a data investigation you have completed, perhaps as a course project. Consider both the space constraint of the abstract itself and the time constraint of the presentation. For example, the Joint Statistical Meetings, a major conference for professional statisticians, has a 1200-character abstract limit (including blank spaces) for a roughly 20-minute talk.
Sample Resource: https://ww2.amstat.org/meetings/jsm/2025/submissions.cfm Sample Resource: https://ww2.amstat.org/meetings/jsm/2022/contributedsessions.cfm
B2: Explain a topic (such as Simpson’s paradox or the confusion of the inverse) to a non-technical audience such as a policymaker who has familiarity with only the basics of statistics.
Sample Resource: https://www.allendowney.com/blog/2021/05/25/in-search-of-simpsons-paradox/
B3: Specify a targeted audience, for example, a decision-maker of some kind, and for a chosen homework question, write a paragraph that communicates the takeaway to that audience.
Sample Resource: https://da4all.github.io/cards/many-ways-to-write-a-statistic
C1: Read the ethics codes that are provided, and then draft your own code of ethics,GAISE College Recommendation 7 that you think the statistics and data science community should follow. For at least one element of your code of ethics, defend its inclusion in the code to a group of statisticians who are considering adopting your code.
Sample Resource: https://www.oreilly.com/radar/of-oaths-and-checklists/ Sample Resource: https://www.amstat.org/your-career/ethical-guidelines-for-statistical-practice Sample Resource: https://datasciencebydesign.org/blog/writing-a-modelers-manifesto-for-more-transparent-ethical-data-science Sample Resource: https://da4all.github.io/cards/write-your-own-data-advocacy-values-statement
C2: Teach someone with math literacy the importance of data privacy and the limitations of anonymity. Do this by constructing an example where the data seems anonymized, but due to small counts, individuals can be identified. How would you explain this to a researcher who is deciding how to report findings and share data upon publication of their work?
Sample Resource: https://rss.onlinelibrary.wiley.com/doi/full/10.1111/1740-9713.01608
C3: Choose a dataset, either one we have used in class, GAISE College Recommendation 4 or another that you think is important to be publicly available. Convince a policymaker to support and maintain this dataset as a public good.
Sample Resource: https://hdsr.mitpress.mit.edu/pub/m3fk4fah/release/2 Sample Resource: https://www.data-liberation-project.org/
D1: Consider a software feature or applet, GAISE College Recommendation 6 that you have taken advantage of in this course. Write a blog post tutorial aimed at statisticians who are not yet familiar with the tool that walks them through how to use the tool and explains why it might be useful to them in their work.
Sample Resource: https://rebeccabarter.com/blog#category=R Sample Resource: https://lukebenz.com/post/ncaahoopr_v1.5/
D2: Pick a cartoon from the CAUSE Cartoon Caption Contest. This can be the current cartoon or one from the past (no peeking at the winning caption!). Write a caption for it and discuss with a classmate the meaning and effectiveness of the caption for an audience who has some math literacy but not specialized statistical knowledge.
Sample Resource: https://www.CAUSEweb.org/cause/caption-contest/
D3: Describe a scenario where two variables are correlated but one does not cause the other. Specify a targeted audience, and create an edutainment piece for that audience that explains the difference in the form of a cartoon, joke, poem, or song.
Sample Resource: https://www.tylervigen.com/spurious-correlations Sample Resource: https://causeweb.org/cause/resources/fun/references
Additional Resources
Annotated bibliography
Examples
Assessments