Period DFs - EMA code literally in R [Regulatives / Guidelines]

posted by ElMaestro  – Denmark, 2011-08-24 15:04 (5012 d 15:24 ago) – Posting: # 7303
Views: 7,190

Dear d_labes

factor(b$sequence) vanished in nirvana :confused:

❝ Clever or bug?

I have no idea. I am one big question mark.

muddle2 <- lm(log(Cmax)~ sequence + subject%in%sequence + period, data=refdata)

My eyes, my eyes, it hurts it hurts! Subject%in%sequence when all subjects have unique identifiers... Initial muddle matrix will contain 77 columns of pure zeros. Makes no practical difference for the results, though.

drop1(muddle2)


drop1 does not work like in SAS when it comes to sequence. When R drops Sequence to get the deltaSS for this factor it does not automatically recognise that Sequence is already in the model because it is contained in Subject. SAS does that (or at least: should do it, but might obviously be failing here).


Your example with first the sequential muddle with period last and then a drop1 is truly brilliant. Here I would expect exactly the same SS for period between the two anovas. It just isn't the case.

Where's hrotter and jdetlor? There must be an explanation.

Pass or fail!
ElMaestro

Complete thread:

UA Flag
Activity
 Admin contact
23,424 posts in 4,927 threads, 1,671 registered users;
120 visitors (0 registered, 120 guests [including 5 identified bots]).
Forum time: 06:28 CEST (Europe/Vienna)

Don’t undertake a project
unless it’s manifestly important
and nearly impossible.    Edwin H. Land

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5