The fact that no-one has yet crossed daffodils and elephants (unless
they have...) could in principle mean that
1. No one has thought of how to do it yet.
2. It's not a good idea, so no one has tried.
Nick
[email protected]
Richard Williams
At 10:27 AM 6/3/2009, Nick Cox wrote:
>Can you identify any literature that defines covariance for ordinal
>(graded) responses? I don't usually align myself with "level of
>measurement" arguments, but I would be surprised if the models you
>think you want are even defined in the literature. I'd be happy for
>anyone to expose my enormous ignorance on this detail.
I keep getting requests to write xtoglm and xtgologit2, but I have no
idea how to do it. The fact that xtologit does not exist makes me
think it is a non-trivial problem.
The closest things I know of are the user-written routines -reoprob-
and -regoprob-, both available from SSC.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
