Non-parametric, multivariate, repeated measures analysis. CLM?
On Jul 8, 2013, at 13:14 , Torvon wrote:
Sorry, don't often post on this mailinglist and thought gmail would automatically reply to all.
That's gmail goof that I have seen before. It actually does copy to the list, if you "reply all", but it sends a copy to me with no indication that it has been cc'ed! I usually wait and see whether that is the case, but in this case there was 15 minutes inbetween and I didn't wait long enough. -p
Thanks! Eiko On 8 July 2013 13:11, peter dalgaard <pdalgd at gmail.com> wrote: On Jul 8, 2013, at 12:59 , Torvon wrote:
Peter, Thank you for your reply. Shouldn't there be a (random) effect of participant somewhere? I.e. a clmm() model? Possibly also random interactions between participant:time and participant:symptoms. Yes, the example output I posted was a CLM, not a CLMM. I will use random intercept models eventually (random slopes are not possible with CLMM), but for the sake of simplicity lets stick with CLM for now (I have 35.000 rows, so random effects in the model do take substantial amount of time to converge with ordered data).
It does look like you can have random interactions, though.
Now, what I am looking for is the overall "symptom * time" interaction term. How do I obtain this value?
If you have the correct error model (and enough participants), it should be a matter of comparing the model with deterministic part index*time to the one with index+time. Presumably, anova() is your friend.
So this would look like the following?
m1<-clm (symptoms ~ index+time, data = data)
m2<-clm (symptoms ~ index*time, data = data)
anova (m1, m2)
no.par AIC logLik LR.stat df Pr(>Chisq)
m1 12 40753 -20365
m2 20 40598 -20279 170.95 8 < 2.2e-16 ***
And that would tell me that, since the time*index model fits the data significantly better than the time+index model, that there is a significant interaction between time and index?
...under the assumptions of the m2 model, yes, that's the gist of it. But please keep it on-list. For various good reasons. For oone, Rune might be reading the list!
Thank you T.
Thank you
T-
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
-- Peter Dalgaard, Professor Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Email: pd.mes at cbs.dk Priv: PDalgd at gmail.com
-- Peter Dalgaard, Professor Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Email: pd.mes at cbs.dk Priv: PDalgd at gmail.com
Peter Dalgaard, Professor Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Email: pd.mes at cbs.dk Priv: PDalgd at gmail.com