But, in case of small samples does it not use up some degrees of
freedom?
>From: David Greenberg <[email protected]>
>Reply-To: [email protected]
>To: [email protected]
>Subject: Re: st: time dummy in random-effects model
>Date: Wed, 13 Jul 2005 16:53:04 -0400
>
>The purpose of the time dummies is to take into account effects that
may
>influence all cases in a given year to the same amount. If you think
>they are likely to exist, then it would be wise to put them in. This
can
>help to eliminate a possible source of spuriousness due to common
trends
>in observed variables. David Greenberg, Sociology Department, New York
>University
Of course it does. If you relax the constraint that all time periods
share the same intercept (and random individual effects vary around
that intercept) and estimate (T-1) additional parameters, you use (T-1)
additional d.f. But if the constraint that those T parameters are equal
is rejected by the data, you should not impose it.
