Resources and other notes from ICOTS 10

A collection of interesting resources, recommendations and bits and bobs from the 10th International Conference on Teaching Statistics (ICOTS) in Kyoto, Japan, 8 - 13 July. Thanks to the University of Auckland’s Department of Statistics for funding my attendance, and to my amazing co-author Anna Fergusson, @annafergussonnz, for letting me work with her.

Liza’s one sentence summary

This list is to help me keep track of all the things I want to check after this conference, and I hope it might help you too.


A Goodreads list here.

  • The Art of Data Science: A Guide for Anyone Who Works with Data by Roger Peng and Elizabeth Matsui (as mentioned in Hilary Parker’s keynote, Parker and Peng podcast together)
  • Sprint by Jake Knapp (‘Sprint’ is a design process Hilary Parker worked through at Stitch Fix, where she is a data scientist)
  • Designerly Ways of Knowing by Nigel Cross (also mentioned by Hilary Parker)
  • Dear Data by by Giorgia Lupi and Stefanie Posavec, with forward by Maria Popova (gorgeous data visualisations that inspired some activities in Andee Rubin’s data camps for middle schoolers)
  • Factfulness: Ten Reasons We’re Wrong About the World – and Why Things Are Better Than You Think by Hans Rosling, Ola Rosling and Anna Rosling Rönnlund (Anna was the Tuesday Keynote. It was the topic of a Gates Notes in April.)
  • Weapons of Math Destruction by Cathy O’Neil (highly recommended by Jo Hardin, @jo_hardin47)
  • Thinking, Fast and Slow by Daniel Kahneman (not mentioned by anyone in particular except me)
  • International Handbook of Research in Statistics Education edited by Dani Ben-Zvi, Katie Makar, Joan Garfield
  • Saving Capitalism from Short-Termism: How to Build Long-Term Value and Take Back Our Financial Future by Alfred Rappaport - not sure if this was an intentional recommendation by Chris Wild, but I thought it looked interesting when he had it up on his short-termism slide.








Follow ICOTS10 tweeps on this list. If I missed you, let me know.

  • Help! I can’t find the Hilary Mason, @hmason, tweet asking about what you do first when you get new data. Hilary Parker mentioned it and I’d love to read the responses. Can you help me find it? Hit me up on Twitter.
  • Most of my tweets from this conference are in the below thread.
  • Hadley showed this very readable code when he was a discussant for Session 3A and in another of his talks. It highlights the value of being able to read and get the gist of even unfamiliar code and the value of languages that can do that.



Where do ICOTS folk go between ICOTS? Send me your recommendations!.


  • Resources mentioned by Charlotte Bolch:



Any great recommendations I’ve missed? Let me know, fellow ICOTSers, on Twitter.

Very rough notes from Session 3A

Data Science for ALL - a stroll in the foothills by Jim Ridgway, University of Durham

Take homes: Just because there is a new game in town, doesn’t mean you should forget everything you know. There are new models and new modellers, so adaptation is necessary to stay relevant.

The roots of our discipline come from people wanting to tackle real problems using evidence and collaborate with anyone sympathetic. But more recent critiques suggest we’re stuck in the early 1900s.

“More and different data (open data, text, images, sensor data…)”

Are Statistics and Computing “worlds in conflict”?

The Death of Theory, (Anderson 2008).

Challenges - new analytic techniques, black box models, unanticipated consequences.

Modelling is process NOT content.

“arithmetic is actually a very dangerous thing to teach people as it can be applied in all sorts of bad way”. If I have a letter for house 12 and one for house 8, I can put both in the letter box for house 20 and be done with it.

Tom Leher quote - life is like a sewer. This is a problem for algorithmic models.

Ridgway’s 3 modelling styles

  • Linear, additive (e.g. school physics)
  • Systems models (e.g. school biology)
  • Macro-systemic models (e.g. evolution of the earth; Trump; Brexit) - unstable over time, layers of systems, VERY vulnerable to error

“Models should be viewed as ephemeral things, NOT rock sculptures”

Data Science: Supporting Evidence Informed Decision making

  • Problem exploration
    • Large scale datasets
    • Data visualisation
  • Data collection
    • Measurement issues (what IS poverty?)
  • Analysis
    • What tools?
  • Decision making
    • ‘What if?’, what are the consequences? how will people ‘game’ the system?

Sampling - just having a big sample doesn’t save you. Who are you trying to generalise to?

Teaching Statistics in the Era of Data Science by Alison Gibbs, University of Toronto


Data science as at the intersection of statistics, computation and domain expertise. Want students to develop a rigorous mindset plus more computational and algorithmic thinking, more real world problems.

“Never become so much of an expert that you stop gaining expertise. View life as a continuous learning experience” - Denis Waitley. A good quote to describe adaptive experts. Flexible, innovative, a mindset for continuous learning and can take their knowledge and expertise and work in new contexts.

“A subject area is ultimately about the doing of a subject - using the content in a disciplined way…[not] a march through content.” - Wiggins and McTighe

Small group tutorials are just as important as lectures. Tutorials have an emphasis on communication and group work.

Challenges and opportunities for undergraduate data science major and minor degree programs by Jo Hardin, Pomona College


Donoho 2017 article, on greater data science - data science IS statistics? Jo’s goal for this talk: You will update something in everything class you teach, every time you teach it.

A survey of statistical capstone projects paper - find this article!

Update 1: Go through a full data analysis This could be through a capstone project, Datafest, ASA Data Expo competition, kaggle, Her Hoop Stats

Update 2: Computation in statistical theory Including simulation makes a big difference to student understanding.

Update 3: Ethics Weapons of Math Destruction

Great Wald story about bullet holes on planes. Where do you reinforce a plane? Where there aren’t any bullet holes on the planes you can salvage.

“Crime data doesn’t exist…the only things we have is arrest data”

Update 4: New/augmented learning outcomes Like confounding and cross validation!

Update 5: Faculty development Use the tidyverse.

Update 6: Reproducibility Use R and R Markdown, Git and GitHub. “If they don’t know Git an GitHub, they’re not getting jobs”.

Update 7: Visualisation Not doing it badly, and/or doing it in sophisticated ways and/or doing it creatively.

Discussant: Hadley Wickham

Important skills: SQL, git, json data, Rmarkdown Ability to combine code and prose together in reproducible ways.

Written on July 9, 2018

Explore all tags: event-notes book-recommendations R-code PhD social-media event-planning