What extra assumptions are introduced in multiple variable regression?
- No exact collinearity between X variables
- No specification bias
What does an estimate in multiple linear regression mean?
The change in Y caused by a change in x, holding all other variables constant
How does the correlation between regressors affect the error of the estimates
Greater the correlation, higher the error
Why do we use Adjusted R^2?
Because normal R^2 can be increased just by adding junk regressors. Adjusted R^2 compensates for the number of variables
When can we compare R^2 values?
-When sample size is the same
-When dependant variables are the same
Give the formula for adjusted R^2
Does R^2 have any intrinsic properties that might favour its use over other calculations?
Nope, pretty arbitrary
What is the Gross/Simple correlation coefficient?
Shown as r1-2, where 1 = Y and i > 1 = Xi. Shows the correlation between two variables
What is the partial correlation coefficient?
The correlation between two variables, eliminating the correlation effect from some other variables. Shown as r12.34, where the effects from 3 and 4 are eliminated