Today’s agenda

Today’s agenda

  • Modeling recap
    • Focus on interpreting slopes in multiple regression models

  • Continue App Ex 5: Forward selection (finish in class)

  • [Time permitting] Review App Ex 5

  • Due Thursday:
    • Start midterm project – at a minimum load data and start reviewing codebook
    • Review peer evaluations and schedule a time to meet with me as a team (details will be emailed)

Modeling recap

Parameter interpretation

  • Slope:
    • Categorical vs. numerical explanatory variable: change vs. difference
    • Simple vs. multiple linear regression: “all else held constant”
  • Intercept:
    • May be an extrapolation
    • May not be meaningful in context of the data

Model fit

  • \(R^2\): % of variability in %y% explained by the model

  • Adjusted \(R^2\): used for model selection
    • Adjusted for sample size and number of explanatory variables

Model selection

  • Backwards: start with full model, take one variable at a time and maximize adjusted \(R^2\)

  • Forwards: start with the single variable yielding the highest adjusted \(R^2\), add one variable at a time and maximize adjusted \(R^2\)

  • If interaction is included, main effects must be included as well

App Ex

Continue App Ex 5

  • Same task/set of variables as before, do forward selection instead

  • 15 mins/team visit with me