Multiple Linear Regression 2

Lecture 18

John Zito

Duke University
STA 199 Spring 2026

2026-03-25

Ann Lee’s announcement?

Bookbagging reminder again

Recap: linear regression

One numerical

One categorical

One numerical; one categorical

Multiple numerical

FAQ

Does it matter what the baseline is?

Sometimes the application suggests a natural choice. As long as you understand and you’re consistent and your communication is clear, often doesn’t matter

Is interaction the same as separate simple models? Same estimates, but different UQ

A second look at \(R^2\)

Model selection

AE