Unbalanced Panel data - regression

[URSUS]

Hi,

I am new to using eviews and regressions so I really need a little urgent help.

I have a dataset of n = 212 companies and t = 10 years and a total of 1561 observations. I have an unbalanced panel. I'm testing 5 independent variables.

As far as I understand I am supposed to use a "Fixed effects model" for the regression.

I know it is alot to ask but does anyone have any tips on how to move forward and on what steps need to be taken in order to perform such a regression using eviews?

Thanks in advance for your help.

Rauf

[BradleyMorris]

I'm not an expert myself, but have done something similar. You will first need to assess whether your data is distributed normally, you can find this out through the jarque-bera test. If it is 0.01 or lower you'll need to execute non-parametric test, if not you can execute parametric tests like the pearson's correlation or t-test..

Hope this helps you a little. Try posting and bothering everybody around you, you'll get there.

[startz]

I'm afraid this is bad advice. Whether the data is distributed normally is completely irrelevant. The error terms being distributed normally matters a little, but even that isn't very important in large datasets unless the deviation from normality is very large.

[BradleyMorris]

Sorry, it's true that the residual distribution needs to be checked instead of the distribution of the data.

-Starz, could you tell me which buttons to press/commands to give in order to execute a spearman rank test? I want to be able to reflect on the difference between the outcome of a OLS and a spearman rank.. Thanks

[startz]

Make a group, choose View/Covariance analysis and flip Method to Spearman rank-order. That may have what you want.