panel data estimation

[johnyath]

Hi, i have a problem and everyone is telling me different opinions. I want to find the determinants of leverage for US from 1987-2012. First i will test the US market and then some sectors. I am using panel data and i don’t know what i have to put in panel options. In cross section and period there is the choice to put fixed or random and in coefficient covariance method ordinary, white cross section or white diagonal. I tried with fixed to both and ordinary. Then i tried with fixed to both and white diagonal to the coefficient covariance method. I have different results. Do you have any suggestion of which is the most robust method, or way to decide which to use? Thanks in advance!

PS: my model is leverage= c+a1tangibility+a2growth+a3( some more possible determinants)....+u