The normality assumption and diagnostics

Residuals

What does linear regression assume are normally distributed?

Residuals, not the raw data

What should you check for normality?

Pointless

How useful is testing the raw data for normality, given the assumption is about residuals?

Residual histogram

What plot checks whether residuals look like a bell curve?

Residuals vs fitted plot

What plot checks linearity and constant spread?

No pattern

What should a residuals vs fitted plot look like if assumptions hold?

Curve

What shape in a residuals vs fitted plot suggests a non-linear relationship?

Fan shape

What shape in a residuals vs fitted plot suggests uneven spread?

Linearity

Which assumption says the relationship is a straight line?

Linearity, normality, homoscedasticity

What are the three main regression assumptions?

Accounting for a missing predictor

What can make residuals normal when the raw data are not?

Failing to reject the null

Does a normality test p-value above the threshold prove the data are normal?

Perfect normality

What does a normality test assume?

Common sense

What does the lecture say is often more useful than a normality test?