On Thu, 29 Jan 2009, Dennis van Amelsfort wrote: > That's just the problem, neither the time dummies nor the time > variable (indicating 2004, 2005 etc..) are significant..
If you'd like to send your data file (probably offlist is better) we might be able to come up with something. From a first glance at your results it looks as if the model is not very well considered: the fixed effects estimator is dropping 17 variables as perfectly collinear; I guess that means they are time-invariant? In principle the random effects estimator can deal with such variables, but with so many of them, I don't know... Allin Cottrell
