Statistics for CSAI 2

by Albert Major

1. New integrated teaching platform

• Lecturer introduced the “seamless app”, an interface that combines lecture slides, quizzes, and R coding blocks.
• You can run R code directly in the browser, without loading data manually.
• It mirrors lecture content, but students are free to ignore it.
• Feedback is welcome.

1. Today’s topic: categorical predictors in multiple regression

• You already learned multiple regression with continuous variables.
• You also learned dummy coding with two categories.
• Today expands this to three or more categories.
• You learn four systems:
  • Dummy coding
  • Unweighted effects coding
  • Weighted effects coding
  • Contrast coding
• All systems convert groups into numeric variables.
• You always need G − 1 coded variables for G groups.

1. Quick quiz session (interpreting R output)

   3.1 Interaction question recap

• R output had two main effects and one interaction effect.
• Interaction term was negative and significant.
• Interpretation:
  • The effect of engagement on grades becomes weaker as class size increases.
  • Because the interaction coefficient is negative.

3.2 How to understand an interaction