Page 1 of 1
NON normality of residuals in panel data
Posted: Sat Apr 20, 2013 2:03 pm
by gianaris
I am running a regression using panel data and 366 observations and Jarque-Bera statistic is around 70 with Prob = 0.000 which means that the residuals are NOT normally distributed. Is this a problem????
Thanks :D
Re: NON normality of residuals in panel data
Posted: Sat Apr 20, 2013 3:23 pm
by startz
Probably not. Normality of the errors isn't important for much of anything in a regression.
Re: NON normality of residuals in panel data
Posted: Sun Apr 21, 2013 12:49 am
by gianaris
just for
coef and p value
Adjusted R-squared is ok
F-statistic is ok
Prob(F-statistic) is ok
White heterosced. test is ok
correlation is ok
Re: NON normality of residuals in panel data
Posted: Sun Apr 21, 2013 7:26 am
by startz
just for
coef and p value
Adjusted R-squared is ok
F-statistic is ok
Prob(F-statistic) is ok
White heterosced. test is ok
correlation is ok
Non-normality is not a problem for the coefficient and is unlikely to be a problem for the p-value unless the data is extremely non-normal or you have a very small sample.