[rkward-devel] Second call for testing: RKWard 0.5.3-pre2

[meik michalke]

hi,

am Donnerstag 22 April 2010 (00:13) schrieb meik michalke:
> am Mittwoch 21 April 2010 (16:48) schrieb Thomas Friedrichsmeier:
> > > Wald_test 
> > > RSM_parameter_estimation 
> > > PCM_parameter_estimation 
[...]
> as for the RSM and PCM  plugin, it looks like it's only a difference in
> variable labelling.  so nothing alarming, but i'll look into it.

ok, this seems to be due to a change in the eRm package from 0.11-0 to 0.12-0, 
indeed regarding only the labelling of the resulting output. i have committed 
an updated output file to svn.

when someone is running the tests with eRm < 0.12-0, it will now fail of 
course.

> the wald test seems to produce some NaNs, that's odd, i'll check that.

hm, now that i looked at it: the odd thing is that i get these NaNs both with 
eRm 0.11-0 and 0.12-0, so on my system the output is "correct" and the test 
passes (the Waldtest help page mentions the elimination of four items, which 
is the case, and i attributed these NaNs to that matter). now mat actually 
gets plausible results. this doesn't look like an error in either the plugin 
or the test to me, but rather like an unforseen behaviour of the Waldtest 
function.

could someone with an installed eRm package please send me his/her output of 
the following?

 data(rsmdat)
 rsm.res <- RSM(rsmdat)
 Waldtest(rsm.res)

(please include R version, eRm version, OS and, for what it's worth, CPU 
architecture)

this is what i get:

<output>
Warning in Waldtest.Rm(rsm.res) :
  
The following items were excluded due to inappropriate response patterns 
within subgroups: 
I2 I3 I4 I5 
Subgroup models are estimated without these items!

Wald test on item level (z-values):

           z-statistic p-value
beta I1.c1       0.388   0.698
beta I1.c2         NaN     NaN
beta I1.c3         NaN     NaN
beta I6.c1      -0.388   0.698
beta I6.c2         NaN     NaN
beta I6.c3         NaN     NaN

Warning messages:
1: In sqrt(diag(solve(parest$hessian))) : NaNs produced
2: In sqrt(diag(lres$W %*% solve(parest$hessian) %*% t(lres$W))) :
  NaNs produced
</output>


viele grüße :: m.eik

-- 
dipl. psych. meik michalke
institut f"ur experimentelle psychologie
abt. f"ur diagnostik und differentielle psychologie
heinrich-heine-universit"at 40225 d"usseldorf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20100424/8b5aaf1c/attachment.sig>