Week 3 - Dummy Variables

Question 1

Why add mixed variables into regression?

Answer 1

Have a set of important variables that usually contain mixed scale types (if miss out could be mis-specifying the model)

Want to compare and contrast different variable types int he one regression

Can control for the effects of nuisance variables (Partial out variance shared with Y)

Question 2

Interpreting variable with 2 dummies

Answer 2

B - weight is the only difference between dummy 1 and 0

Dummy 0 = Constant (Predicted score for 0 Group)
Dummy 1 = Constant + b-weight

Question 3

Multiple Level Dummy

Answer 3

g-1 dummy for levels (4 level = 3 dummies)
0 = reference group

Each level will have their own b-weight to compare to the constant

Question 4

Multiple Category Dummy (two sets of categorical dummies)

Answer 4

Controlling for the other dummy, you have higher or lower score on Y for (dummy 1)

Question 5

Mix of continuous and Categorical dummy

Answer 5

Continuous = Change in predicted score for every 1 unit change on (X1) on (Y), when holding (D1) constant 
Dummy = Difference in predicted (Y) score between (dummy 1 and 0) when holding (X1) constant