The F-statistic, t-statistics, p-values and standard errors of the estimated coefficients will all be incorrect in the presence of heteroskedasticity.

However, does this even matter if I am working on a dataset that covers the entire population of interest given that I have no need to generalize beyond the present dataset.

  • $\begingroup$ What do you mean by "covers the entire population of interest?" If you really have the entire population (all possible data), then statistical tests are meaningless. $\endgroup$ – jkpate Apr 22 at 17:11
  • $\begingroup$ It's not an estimate any more, so none of that stuff matters, as you suspect. $\endgroup$ – jbowman Apr 22 at 17:49

Your Answer

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

Browse other questions tagged or ask your own question.