Heteroskedasticity and robust standard errors
When error spread varies with x, OLS stays unbiased but its usual standard errors do not: compute robust errors, test with LM = nR², and weight when the pattern is known.
About 35 minutes8 activities Paper packet