Hi Silviana, Your question about heteroskedasticity persisting even after you check the "robust standard error" box is actually more about econometrics in general, and not gretl-specific. Such questions are generally not what this list is for. That said... The robust std error procedure (in whatever context) doesn't change the method of estimating the coefficients. Unlike WLS, for example, the goal isn't to "fix" the heteroskedasticity. Rather, it's to construct standard errors that should give reliable inference even in the presence of hetero. I hope this helps, Peter Summers High Point University High Point NC Sent from my iPad On Jun 8, 2018, at 11:44 AM, SILVIANA SANTOSO <m37414018(a)john.petra.ac.id&gt; wrote: Hello, I want to ask several question about panel data in gretl. I currently working on my thesis that use panel data. I have already tested my data to choose the right models (pooled/fixed/random). My model is fixed effect models. Then I want to test its normality and heteroscedasticity. I have tried to click on test > heteroscedasticity > groupwise. My model is suffered the heteroskedasticity problem. So I tried to rectify my model from heteroskedasticity by ticking the robust standard error arellano. But by model still suffer of heteroskedasticity problem. So what should I do? And about normality my model is not distributed normaly but I don't know how to rectify it in gretl Hope you may give a solution to my problem. Thank you in advance Sincerely, Silviana Gunarsih Santoso _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu https://linkprotect.cudasvc.com/url?a=http%3a%2f%2flists.wfu.edu%2fmailma... _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users