Re: random and fixed effects

oercim wrote: 
       I am studying panel data. The unobservable effects are
categorized into two groups as random or fixed. I am very confused with
those "random effects" and "fixed effects" concepts.
        I  know what is definition of the difference between random and
fixed effects. Effects are considered as "fixed effects" assumed model
consists all levels of the assumed population while they are considered
as "random effects"  if the assumed model consists of a few levels from
a popoulation which has a  lot of levels. However, I am not clear do i
understand the subject well. I want to understand the dynamics of the
variation of the dependent variable. What happens if I treat a fixed
effect as random or opposite. It seems as an easy issue but why i cant
see?
      I am very confused. Please help. Thanks a lot iBest regards.


For many (most?) statistical procedures, you need to know the variability of your statistics. The estimated variability of things changes according to whether effects are fixed or random.

If a factor is random, then, when the experiment is repeated, the levels change, which often increases the variability of things over what would it would be if the levels stayed the same.

Mislabel fixed as random or vice-versa, and your variance estimates are messed up--too large or too small.
.