Unbalanced Panel data (FEM , REM, or Pooled regression)


I have unbalanced panel data. Using SAS, the null hypthesis of Hausman Test was not rejected (means the random effects model is better than fixed one).

however, the null hypthesis of Lagrange-multiplier test was not rejected too (means pooled regression is better than random effects).

Shall I use pooled regression based on that ??

Note: LM tests by SAS is for balanced panel, maybe if I can use the modified one by Baltagi and Li (1990). Any of how to perform the modified one? using software or by hand???