1

I'm doing an econometric analysis to try to explain the gender wage gap. Here's my regression:

enter image description here

Bp-test, White test outputs and residuals VS fitted values plot (rvfplot command in stata)

enter image description here

So for the bp-test i'm rejecting the null hypothesis of homoscedasticity. For the White test i'm NOT rejecting the null hypothesis of homoscedasticity. White test shouldn't be more general than bp test? so shouldn't it be more precise than bp test? Why they give opposite outputs?

From the residuals plot i can see there's no much difference in variance exept for some outliers. Am i wrong about this interpretation?

Thanks to anyone's help/criticism!! :)

edit: Is there any wrong question in my post? I'm looking around forums from 4 days and didn't find anything really helpful. Getting crazy!! Please is there anyone who can suggest me the right way to handle this kind of problem?

Stramike
  • 13
  • 6
  • The White test is more general, since it also (in its general specification) non linear combinations of the independent variables. Which brush pegan does not – Repmat Jun 10 '16 at 21:21
  • Exactly! So if white test takes in much more informations shouldn't it be more reliable than BP-test? – Stramike Jun 10 '16 at 23:04

0 Answers0