[Gretl-users] Re: Paired T-Test Kruskal-Wallis test C API
Am 06.05.2021 um 09:33 schrieb Riccardo (Jack) Lucchetti:
On Wed, 5 May 2021, Sven Schreiber wrote:
> (@Jack, BTW: I guess the issue is practically moot now, but I think the
> package requirement would have to be 1.9.12 instead of 1.9.4.)
Do you think it worthwhile to update the package just for this? I had a
brief look at the package to see if there was something else worth
adding to justify raising the version number, but I couldn't think of
anything.
As I said, such an old version requirement is probably irrelevant now.
There may be a couple of minor points, though:
- the help text could perhaps use some reformatting (and typo: returs)
- About the grouping variable: is this supposed to be a binary dummy to
define the samples? Or at least discrete? If so, it doesn't seem to be
checked, I could execute the GUI function with arbitrary variables there
apparently. If not, maybe some more explanation could help.
- In general it's not quite clear to me why the second function combines
a different interface (list instead of series) with a different
statistic (median instead of mean). Again, maybe some explanation might
be good.
thanks
sven