Activity created by Federica Zoe Ricci and Mine Dogucu to practice the Beta-Binomial model.
The activity is designed around the online racing game Racer.
In summer 2024 this work was presented presented at eCOTS and BAYSM. The slides are available here.
Abstract
‘Where do priors come from? Should there only be one correct prior? What can people do with Bayesian statistics in real life?’ These are common questions students have when introduced to Bayesian ideas, and providing satisfying answers can be challenging. We present an activity designed to make students practice the Bayesian sequential-learning framework in a realistic and fun setting, based on the Racer game (from Stat2labs.com). Acting like the analysts of a racing team, students are asked to help their manager decide which type of tires to use for an upcoming race, based both on prior knowledge from their team’s engineers and on data they collect in class by playing the racing game. Some teams discuss with the class: what prior distribution they chose, the data they collected and their posteriors. As some teams received different prior information, students reflect on the legitimacy of choosing different priors and on how they all used Bayesian statistics to quantify uncertainty in this problem. We discuss our experience with implementing this activity for a discussion (i.e., lab) session of an Introduction to Bayesian Data Analysis class with upper-level undergraduates at a research university and the students’ perception of this activity as reported in a survey.
Material
The folder activity-material in the GitHub repository federicazoe/bayes-games includes:
discussion-handout.qmd: template to fill out for each team, that guides through the activity, has questions and code. Each team made their own copy of this file, that they accessed via a link on the course’s website.discussion-handout-team01.qmd: example of how discussion handout completed by one of the teams.team-names.md: names of three team types, with a short bio of the person inspiring the name. Names and bios were used to label team-specific, physical envelopes with team names printed on them e.g. team-danica-1, team-danica-2, …, team-lewis-1, team-lewis-2, …, team-mario-1, team-mario-2, …team-scenarios.md: Content of the letters with prior information and instructions. There were three letters, corresponding to the three team types. A copy of this information was printed for each team member and included in teams’ envelopes.team-member-roles.md: Each team member was responsible for an essential component of the activity (e.g. facilitating team discussion, filling out discussion handout). Cards with role names and their descriptions were included in team envelopes’ and teams distributed them among their members, at random.discussion-slides.qmd: slides used in class to explain and guide the activity.
Video
External Resources
Other game-based activities for teaching statistics: Stat2labs
More online games for teaching statistics: Stat2games
References
Kuiper, S., & Sturdivant, R. X. (2015). Using online game-based simulations to strengthen students’ understanding of practical statistical issues in real-world data analysis. The American Statistician, 69(4), 354-361.
Lesser, L. M., Pearl, D. K., & Weber III, J. J. (2016). Assessing fun items’ effectiveness in increasing learning of college introductory statistics students: Results of a randomized experiment. Journal of Statistics Education, 24(2), 54-62.
Lesser, L. M., & Pearl, D. K. (2008). Functional fun in statistics teaching: Resources, research and recommendations. Journal of Statistics Education, 16(3).
Hoegh, A. (2020). Why Bayesian ideas should be introduced in the statistics curricula and how to do so. Journal of Statistics Education, 28(3), 222-228.
Hu, J., & Dogucu, M. (2022). Content and computing outline of two undergraduate Bayesian courses: Tools, examples, and recommendations. Stat, 11(1), e452.
Funding
|
Federica Zoe Ricci was supported by a fellowship of the HPI Research Center in Machine Learning and Data Science at UCI . |
Acknowledgements
We would like to thank Tyler George, Abhishek Chakraborty and Shonda Kuiper for their USCOTS 2023 workshop Improving students’ communication about data using online statistical games, that inspired this work.
Thanks to Catalina Medina and Zahra Mohslemi for their help with printing the material and preparing team envelopes prior to the discussion (lab) session where the activity was taught.
