Am 26.04.2018 um 21:41 schrieb Clive Nicholas:
On 26 April 2018 at 11:11, Sven Schreiber <svetosch(a)gmx.net
*Frankly, I could care less whether such things are
consensus" or not. It is my personal view and I'm happy to give it. If
the OP finds it useful, so much the better. If they prefer to chuck my
post straight in the bin, that's for them.
Well, at the risk of starting religious wars, I'm always happy to hear
arguments and reasons for personal views. But a concrete answer to a
concrete question ("why isn't the Hausman test shown") would also be
nice. For example, what if you asked a question about a multilevel model
and somebody just answers "Bayesian is better anyway" ? My guess is that
that wouldn't help you too much.
*I explained to the OP that s/he could run the R code (if s/he so
wished) through -gretl-. You may have missed that bit.
In fact I haven't. But this line:
wouldn't work from gretl, or at least you would have to prepare stuff first.
Gretl has the "foreign" block apparatus which could be used here, no doubt.
I would welcome
any effort to see HLMs introduced in -gretl-, but it's not really
place to ask for it. If people want it badly enough, they can learn to
code it themselves. I can't, so I won't.*
Fair enough. I'm not an expert in that specific area, either. However,
it seems that at least Stata is using max likelihood there (judging from
their ME manual), so my guess is that for someone who knows these models
better it's not prohibitive, using the ML apparatus (mle block, see ch.
23 of the guide).
But that's not a recommendation for applied users, of course.