On Mon, 8 Dec 2008, Sven Schreiber wrote:
Am 08.12.2008 18:29, Allin Cottrell schrieb:
> A few things to report:
>
> 1) Jack has added Interval regression, the shiny new "intreg"
> command. It's documented and in the GUI and has an example
> script.
>
> 2) I've just done something that was requested a while back,
> namely enabled the "restrict" command for nls, mle and gmm.
> Please test and report any misbehavior!
That's great, thanks a lot! I will test, but can't promise
anything this week.
Thanks.
> 3) I've added accessors $Fstat and $chisq which retrieve
the
> F-stat or chi-square test for overall fit from the last model.
> (Most model types have an Fstat but no chisq; the exception
> is the limited-dependent variable models logit, probit and
> intreg.)
are the d.o.f. always obvious for these stats? or are these
actually matrices that also give you the d.o.f and p-value?
(just a shot in the dark)
$Fstat and $chisq return scalars, but IMO the df should be clear
enough to people who are likely to make use of these things: one
has the $ncoeff accessor and one knows how to subtract 1 ;-)
Allin.