Heteroskedasticity testing

[Albert Otete]

I undertook a test of heteroskedasticity after regression of a number of independent variables against two dependent variables (comp_portbase and comp_bidsuccess). I got mixed results in terms of the prob>chi2 where first one was 0.6508 and the second one was 0.0071. What do these probabilities mean and what is the recommended solution to fix heteroskedasticity?

[Carlo Lazzaro]

Albert:
please do not post attachments but share what you typed and what Stata geve you back via CODE delimiters (see the FAQ on this and other posting-related topics. Thanks).
That said, the usual fix to account for heteroskeadticity under -regress- is invoking -robust- standard errors.

[Albert Otete]

regress comp_bidsuccess devp_isa devp_ifrs devp_tax devp_consult devp_itcaats devp_exams devp_writing devp_selling devp_leadership

ettest devp_isa devp_ifrs devp_tax devp_consult devp_itcaats devp_exams devp_writing devp_selling devp_leadership

Result was as follows after hettest:

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
Ho: Constant variance
Variables: devp_isa devp_ifrs devp_tax devp_consult devp_itcaats
devp_exams devp_writing devp_selling devp_leadership

chi2(9) = 22.61
Prob > chi2 = 0.0071

If the null hypothesis is rejected, how do i proceed?

[Marcos Almeida]

Just a few notes after Carlo's advice:

The command - hettest - has been replaced (I believe, since Stata 9) for - estat hettest - .

I can only see the output of one -hettest -.

You may use - robust - vce estimations.