Hello, I have a general data analysis question. I recently visited a
lab where they are testing a new treatment and they had done the
experiment several times on different dates. They repeated the
experiment 3-5 times per day. And then for practical reasons they
repeated the whole procedure for 5 days.(they wanted a large sample
size but practically they couldn't handle more than 5-10 experiments
per day). However there might have been some extra variation between
different days because the experimenter changed, although same
procedure was being followed.
Below are the data:
Control data:
Cday1=c(5,2,5,3,4);
Cday2=c(2,1,3,1);
Cday3=c(7,6,4,11,10);
Cday4=c(5,13,8,4,10,6);
Cday5=c(21,8, 5, 5,11);
Treatment data:
Tday1=c(17,11,25,21,16);
Tday2=c(17,7,12);
Tday3=c(16,18,4,20,18,25);
Tday4=c(17,20,29,17,19);
Tday5=c(14,31,28,34);
Then they decided to do a paired t.test on the "mean" per day to
measure whether
they can detect a difference between the Control and the Treatment,
something like:
t.test(c(3.8,1.75,7.6,7.66,10),c(18,12,16.83,20.4,26.75),paired=T)
But I thought there was something wrong in that procedure, something
missing but I couldn't figure out what exactly, my feeling was that
they were not capturing the variability in the measurements taken
within one day. I thought maybe the solution could be in linear mixed
effects models (lme) but could that be used to have some sort of a
p-value (or other) to say there is a difference or not between the
2 conditions. Or maybe other procedures? Any ideas? Thanks
--
View this message in context:
http://www.nabble.com/Repeated-Measures-%28lme-%29-tp16012010p16012010.html
Sent from the R help mailing list archive at Nabble.com.
