See you at the finish posterior line

Stats 115 discussion

2024-01-24

Hi there!

I’m Federica Zoe Ricci.

PhD student in Statistics at UCI.

My research: Bayesian models for network data (e.g. social networks) and multi-subject time-series observations (e.g. neuroscience studies).

I also work with prof. Dogucu on Stats Education research

Today’s activity is not research..unless you want to!

Today’s discussion

What do we want to learn today?

  • see examples of where priors come from

  • practice choosing priors and realize that different priors can be reasonable!

  • practice the Beta-Binomial model (that you’ve learned on Monday)

..all of this by means of a game!

What’s the game going to be like?

You’re part of a team

In this game, you race a chosen car on a chosen track

Let’s start by taking a quick look at it: https://www.stat2games.sites.grinnell.edu/games/raceradvanced22.html

Logistics

  • Form teams

  • Distribute material to teams

  • Open letters and distribute roles among team members

  • Each team member, take 2 min to read letter

  • Annotator copies the .qmd discussion handout and opens it

  • Teams work on the discussion handout

    • Team introductions (3 min)
    • Choose prior
    • Play game (Player ID/Group ID; Finishing time!!!)
    • Derive posterior

  • Submit handout on gradescope

Teams results

  • One Danica, one Lewis and one Mario team will present their Quarto files

Some reflections

  • what were the differences between these scenarios?

[thoughts: different prior information direction and strengths, different prior choices, different data, different players!]

  • if different priors can be reasonable, what could be a natural thing to do when analyzing data?

[thoughts: check the sensibility of your results to different priors]

  • Why were prior especially important to this analysis?

[thoughts: only up to 5 observations per team]