Interpreting White test for heteroskedasticity

#1
Hello,

i am using Gretl software and run a OLS regression on a set of panel data. I am trying to find out if there is hetereskedasticity in the model, but I do not know how to interpret the outcome of White test.

It says:

Null hypothesis: heteroskedasticity not present
Test statistic: LM = 40.5477
with p-value = P(Chi-square (21) > 40.5477) = 0.00637482


How can I tell from this information if there is heteroskedasticity present?

Thank you very much.
 

noetsi

No cake for spunky
#2
I assume you use the normal rules for rejecting the null which is if p is below alpha you reject the null. .006 is below any alpha I have ever seen used so you would reject the null. Since the null is that heteroskedasticity is not present you accept the alternative that it is present.