[rkward-devel] new JS functions

[meik michalke]

hi,

thanks for the hints! i see the two functions are really rather superfluous, 
so i removed them again.

Am Mittwoch 19 Oktober 2011, 09:12:03 schrieb Thomas Friedrichsmeier:
> There is more complexity in corner cases, though. <varslots> will readily
> allow you to mix objects which are not inside the same data.frame.

yes, i was aware of that -- how to cope with that would have been my next 
question ;-) 

> You can certainly argue that in general it makes a lot of sense to keep all
> model data inside one data.frame. I'm not sure, we want to force users to do
> that in all cases, though(*).

that was actually not my choice, it's what ezANOVA() expects here. i'll 
probably try to add some other dialog to convert data into one fitting 
data.frame. not sure if it's possible at all, though...

> I had started to adress this problem in the <formula>-tag.

i've seen that, very nice! unfortunately, for ezANOVA() it doesn't really 
help, as you specify only variables, not formulae. i prefer ez nonetheless, 
because it has a remarkably simple interface. i actually didn't get 
greenhouse-geisser correction to work any other way.

> The formula widget does not support random effects, yet, and still has other
> problems, too (it does not work well in run again links), but perhaps it is
> the way to go after all?

it's certainly something! at the moment it's perhaps a little "static", but 
i'm already thinking of a totally different use case here... that is 
multinomial tree models. it would need a whole number of formulae for one 
model, each defining one branch of the tree. but, well... some other time ;-)

> (*): If you do want to force users to use data from a single data.frame,
> only, you'll have to use a two stage solution: One varselector + varslot to
> select the data.frame, then connect that to the "root"-property of a second
> varselector (like in the "Sort Data" plugin).

exactly what i was looking for. i restructured my ANOVA plugin accordingly, 
and it seems to work quite nicely already:
 o http://reaktanz.de/stuff/R/anova_plugin.R
   (needs rkwarddev 0.04-0 to run)


viele grüße :: m.eik

-- 
dipl. psych. meik michalke
institut f"ur experimentelle psychologie
abt. f"ur diagnostik und differentielle psychologie
heinrich-heine-universit"at d-40204 d"usseldorf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20111020/da76f202/attachment.sig>