Scaling of Cmin [Regulatives / Guidelines]

posted by zizou – Plzeň, Czech Republic, 2017-02-03 14:49 (3058 d 11:22 ago) – Posting: # 17006
Views: 12,600

(edited on 2017-02-03 18:47)

Dear VStus,
it looks like you reduced it too much.


❝ I have reduced Liu's dataset to have:

❝ subject - 20 levels (1-20)

❝ sequence - 2 levels (RT and TR)

❝ period - 2 levels (1 or 2)

❝ replication - 2 levels (1 or 2)

❝ AUC for Reference formulation.


When you reduce period for 2 levels - I think the change makes period to be equal to replication.
Period should still have 4 levels (1 or 2 or 3 or 4). If I am not mistaken.

Edit1: My mistake I don't have a data/reference and I thought the data was from 4 periods. After repeated reading I noticed that there were two periods only with created "repetition" term. I am sorry, I was too fast with replaying and I lacked the quality.

Edit2: If I get it right (I hope so) there is something like this scheme:
  period: 1    2
seq 1: ---TT---RR
seq 2: ---RR---TT

When we are interested only in R (per1+seq1 and per2+seq2 are not used at all). So in evaluation of only R data, Period 1 is equal to Sequence 2 (in the term of categorizing data) and Period 2 is equal to Sequence 1, so it's also known from subject No, when the data was obtaind (period 1 or 2). So period doesn't have any new information (the information is provided in other terms).
You may try:
anova(lm(log(AUC) ~ seq + rep + subj:seq, data=data, na.action=na.exclude))
(I'm curious if the results differ against the results with period in.)


Best regards,
zizou

Complete thread:

UA Flag
Activity
 Admin contact
23,424 posts in 4,927 threads, 1,677 registered users;
28 visitors (0 registered, 28 guests [including 12 identified bots]).
Forum time: 03:12 CEST (Europe/Vienna)

Complex, statistically improbable things are by their nature
more difficult to explain than
simple, statistically probable things.    Richard Dawkins

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