5

Question 1

Linear regression model assumptions (6)

Answer 1

1 There is a linear dependency between X and Y .
2 Error term has zero mean
3 Explanatory variable (X) are not correlated to error term
4 Error terms are not correlated (or independent)
5 Error term has constant variance
6 Error term has normal distribution. (A stronger assumption is vector (X, Y ) has bivariate normal distribution.)

Question 2

Most common methods to check Linear regression model assumptions

Answer 2

1 There is a linear dependency between X and Y .
Residual plot.
2 Error term has zero mean.
Ensured by OLS.
3 Explanatory variable (X ) are not correlated to error term.
Ensured by OLS.
4 Error terms are not correlated (or independent).
Hard to test. Residual plot.
5 Error term has constant variance.
Residual plot.
6 Error term has normal distribution.
Shapiro-Wilk normality test for residuals. A stronger assumption (vector
(X, Y ) has bivariate normal distribution) is harder to test.