Random effects panel data

[Sjaakie de Boer]

Hi, I am fairly new to Stata and statisitcs overall. For my thesis I am conducting a random effects panel data. With T much larger than N. I am wondering whether I need to be testing for autocorrelation, heteroskedasticty, cross-sectional dependence, and multicollinearity. Now my question is whether there are other important things I am overlooking or I am doing things wrong? I tested for autocorrelation using a Woolridge test xtserial. I am wondering how to test for heteroskedasiticty in my model. Cross-sectional dependence using a Breusch-Pagan LM test xttest2. Multicollinearity using a VIF test. Hopefully someone can guid me into the right direction. As for me the information on the internet looks quite overwhelming and contradictory, and I have the feeling I am doing entirely wrong. Thanks in advance!

[Sjaakie de Boer]

I think I need to test for Heteroskedasticity using a Breusch-Pagan test > estate hettest ?

[George Ford]

xtscc will you give you Dkraay errors, which adjust for hetero/autocorr/spatial corr (a type of cluster error) and is well suited to the data (small N, large T).

ignore multicollinearity, unless you suspect you've got a real problem.

http://fmwww.bc.edu/repec/bocode/x/xtscc_paper

[Sjaakie de Boer]

xtscc will you give you Dkraay errors, which adjust for hetero/autocorr/spatial corr (a type of cluster error) and is well suited to the data (small N, large T).

ignore multicollinearity, unless you suspect you've got a real problem.

http://fmwww.bc.edu/repec/bocode/x/xtscc_paper

Thanks for your response! Should I test for the issues or should I just use the Driscoll-Kraay standard errors?

[George Ford]

I wouldn't bother. You've fixed any problems that could arise.

[Sjaakie de Boer]

Oke, thanks for your help!