random and fixed effects in panel models

[Lili]

Hello,


I am a beginner with Eviews so I want to ask you guys for some help. Can somebody please explain to me the difference between using no effects, fixed effects and random effects when working with pooled data? I need to test whether some indicators based on financial situation influence the market performance indicators. The indicators are computed for 90 companies and a 5 years period. I have a paper as a model and I saw the person who made it tested also using fixed and random effects.
Another thing that I do not understand is: when I use the random effects either for cross-section or for period, what kind of effect should i use for the other dimension - no effect or fixed effect? I oder words, when i want to use random effects only for cross section/time, what should i select for the other one left? Should i check both NE and FE?
I also need to test for stationarity using unit root test. What do i have to fill in in the window for unit root testing.


Thank you very much!
Lili