Multiple Regression

Question 1

Tolerance

Answer 1

A tolerance = 1-R2
R2 = multiple R-squared
A tolerance of less than 0.1 indicates a collinearity problem.
Has nothing to do with Y.

Question 2

Collinearity

Answer 2

Is correlation between 2 or more predictor variables. When predictor variables in the same regression model are correlated they cannot independently predict the dependent variable.

Question 3

The multiple R2 shows the percent % variability in Y that can be……?

Answer 3

Explained by the model.

Question 4

Adjusted R2

Answer 4

More improved way of assessing if a predictor variable adds value to the model or not. Has a maximum of 1.
Low adjusted R2 means poor model, the higher the adj R2 the better the model.
Pick model with the highest adjusted R2.

Question 5

AIC

Answer 5

Lower the better model.

Question 6

BIC

Answer 6

Lower the better model.

Takes into acc number of explanatory variables.

Question 7

If only one more explanatory variable is added to the model, only a wat test??

Answer 7

a t-test is required to test if this explanatory variable adds value to the model or not.

Question 8

Pick model with the higher or lower adj r2?

Answer 8

Higher.

Question 9

Pick the model with the higher or lower BIC?

Answer 9

Lower

Question 10

Pick the model with the higher or lower AIC?

Answer 10

Lower

Question 11

In multiple regression, do a partial f-test when one of the models being tested is….?

Answer 11

Nested within the original model.