6
$\begingroup$

Let's say I've ran a linear regression and I'm checking the model diagnostics. I made a histogram of the residuals and they appear more or less normally distributed as below.

I thought for a long time that normally distributed residuals like below also indicated that the residuals are homeskedastic. However, I am I right in saying that heteroeskedastic residuals like below would also be normally distributed? Or in other words, we cannot tell if residuals are homeskedastic or heteroeskedastic just by plotting a histogram?

enter image description here

$\endgroup$
  • 2
    $\begingroup$ It's not necessarily the case that a marginal distributions of residuals looking normal would indicate homoskedasticity. $\endgroup$ – Glen_b Oct 22 '14 at 13:04
5
$\begingroup$

You're absolutely right, normally distributed does not imply homoskedastic, and your example is great.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.