Re: FW: st: Random effects probit

["Guillaume Frechette" <gurst1@hotmail.com>]

Daniel M�ller <d.mueller@agr.uni-goettingen.de> wrote:
> Guillaume,
>
> could you pls add the Stata answer you mention in your post on missing std
> err.
>
> thanks, cheers,
> Daniel
Ooops, sorry about that. Here it is:

Dear Guillaume,

In general, it means that the standard error could not be identified. 
Another
way of saying the same thing is that the variance-covariance matrix does not
have full rank. collinearity is the most common cause. The fact that the
problem arises when you cl uster is not surprising. If you think of clusters
as superobservations, clustering can drastically reduce the rank of your 
data
matrix (design matrix), i.e., in reality you have fewer observations, you 
can
drastically reduce the overall variation in your data , making collinearity
more likely to be present.


Sincerely,

Allen McDowell

_________________________________________________________________
MSN Photos is the easiest way to share and print your photos: 
http://photos.msn.com/support/worldwide.aspx

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/