st: Mixed model degrees of freedom and Stata presentation

[Jordan Silberman <silberman.stata@gmail.com>]

Hi Stata folks,

My department is abandoning SPSS, and instead will begin teaching
courses with a different package--probably R, SAS, or Stata. I've been
a Stata fan for years, and I'm going to give a presentation to the
department on pros/cons of Stata. It seems to me that Stata beats or
at least ties other packages for most criteria, but there is one area
in which Stata seems to be behind--df estimation for mixed models.
Unfortunately, this could be a deal-breaker for many researchers in
our department. I understand that this is not an issue for large
samples, but it is an issue when samples are small. And even when
samples are large enough that differences between z and t significance
levels are trivial, the fact is, reviewers in some fields may still
expect df values to be reported for multilevel models (whether or not
this expectation is justified).

So, a few questions:

1. Are there plans to provide more extensive options for df estimation
(eg, Kenward-Roger, Satterthwaite, etc.) with xtmixed/xtmelogit in the
future? This feature would be extremely helpful, even if just one
estimation method is provided.

2. I have read that Stata statisticians believe there's no defensible
way to estimate df for mixed models. Can anyone explain why this is
so, preferably in language a non-statistician can understand?

3. The solution to this problem offered in Stata is to assume infinite
degrees of freedom. It seems to me, from a statistically naive
perspective, that it is literally mathematically impossible to use a
less defensible solution. It's not possible to provide a df estimate
that is further from the true df value than infinite. But I suspect
that there's more to it than this. Can anyone explain why assuming
that df = infinite is more defensible than other df estimation
methods, even though other methods are mathematically guaranteed to
provide more accurate df estimates?

4. What can I say to researchers who publish in journals in which
reviewers are used to seeing dfs reported with multilevel models about
the possibility of using Stata to estimate their multilevel models?

Any thoughts would be greatly appreciated.

Thanks,
Jordan
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/