S4E19 The Dark Art of Regression Diagnostics
Quantitude
00:00
Is Heteroscedasticity in Your Regression Parameters a Problem?
Having heteroscedasticity in your residuals doesn't bias your estimates of your regression parameters per se. Where that starts to cause problems is when I want to start bracketing predictions with intervals around that. Other things that these things would indicate is as I alluded to before, maybe there's a nonlinear effect that you've omitted. So all of these things are your free radical opioids of I wonder if this might be a problem and then you can act upon it later.
Transcript
Play full episode
