S8 - Multiple Regression (complete ?)

Question 1

What is R squared?

Answer 1

R^2: the proportion of the variation in Y that is explained by X; goodness of fit

R^2 [0,1]

Caveat:
Adding any variable, no matter how irrelevant, will usually raise R-squared at least a little

for comparing the fit of two models, adjusted R-square is a better measure

Question 2

What is the problem with skewed variables?

Answer 2

• skewed variables can produce outliers
• right-skewed variables are a common problem

-> can be transformed with log

Question 3

Adding any additional variable to a model never decreases the R-squared (R^2)? True or False

Answer 3

True

Question 4

When violated, the assumption of homoscedasticity lead to biased estimates of the coefficients. True or False?

Answer 4

False

Question 5

When violated, the assumptions of linearity and mean independence lead to biased estimates of the coefficients.

True or False?

Answer 5

True