Week 9 - Multiple Regression II
Groups, Slopes & Interactions
Week 6 taught you that a group is just a predictor. Week 8 taught you that a predictor’s coefficient depends on what else is in the model. This week those two facts meet in a single regression: one categorical predictor, one continuous predictor, and a question neither week could ask.
The additive model — group plus covariate — draws parallel lines: same slope for every group, different heights. That is a strong assumption, and most people never notice they made it. An interaction relaxes it, and lets the slope itself differ from group to group.
Once an interaction is in the model, there is no such thing as “the effect of X.” There is only the effect of X at a particular value of the other variable. That is why we report simple slopes, and why centering stops being cosmetic.
We work two examples deliberately: one where the interaction is real and changes the story (tutoring helps the students who study least), and one where we test for an interaction, do not find it, and report the additive model like adults. Both are correct outcomes. Only one of them usually gets written up honestly.
In-Class Check 3 runs at the start of class — about 20 minutes, open notes, open readings, open past labs. It covers Weeks 7 and 8: R², adjusted R², the residual standard error, the nested F test, and interpreting a coefficient with its holding constant clause.
Prepare
📖Chapter 10 - ROS — the interaction sections
Slides
💻 Multiple Regression II - JAMOVI
The interaction menu is the one thing in JAMOVI that is genuinely hard to find the first time — it lives behind the arrow dropdown ▾ in Model Builder, not in the variable boxes. The JAMOVI deck walks through it twice.
Data
Lecture: Study, Tutoring & Grades (.csv) · Sleep, Parenting & Anxiety (.csv)
In-class activity: <DASS survey data (.csv, 20 MB)
In-Class Activity
For Next Time
📋Midterm Project — due Tuesday 10/27 at 11:59pm
The midterm is due Tuesday. No lab is due that week, which means there is nothing competing for your time and nothing to blame if it is not done. If Part 6 is not drafted by this weekend, it will not be good by Tuesday.
📋Lab 9 - Groups, Slopes & Interactions — due Tuesday 11/3 at 11:59pm, posted this week
Lab 9 is posted now so you can see what is coming, but do the midterm first. The lab assumes everything from today and nothing from after today.
📖Chapter 11 - ROS — assumptions and diagnostics
For four weeks we have been glancing at residual plots and saying “nothing obviously wrong.” Next week each of those glances becomes a named assumption with a diagnostic, a test, and something to do about it when it fails.
Back to course schedule ⏎