What are the best steps to follow in a panel data analysis? (using stata)

#1
Hello,

I would like to know your opinion on the steps I’m following in analyzing panel data in stata.. Are the steps I’m following ok or not? If you consider that I should look for something else besides what I will list below please tell me.

Here are the steps:

1) check unit roots for all variables (ipshin, levinlin)
2) if there is unit root, I will do panel cointegration tests (xtwest) for only the variables that I have unit roots
3) if there is a panel cointegration, I take the first difference to remove cointegration
4) Fixed and random effects regression and IV regression
5) Hausman test for checking whether my model have fixed or random type
6) normality test for residual to check whether they are normally distributed or not (could you please tell me how to run this for panel data?)
7) check for serial correlation (xtserial)
8) Granger causality test (do you know how to check for this in stata?)
10) check for heterokedastcity (xtgls).

I appreciate your advice.
Thanks in advance
Daniela