longitudinal with 2 time points

Thank you John. I agree making baseline as a random factor is not a good idea. 

The data have treatment groups and age and gender for each subject. The purpose 
of the study is to investigate the treatment effect on the change of the study 
endpoint?(glucose level) between week 4?and baseline. I am thinking of several 
models/methods to analye the data:

1. mixed model with fixed time and random intercept:
lmer(y ~ treatment + gender + age + time + (1|subject)??? where time = 0 or 4

2. mixed model with random intercept and random slope
lmer(y ~ treatment + gender + age + time + (time|subject)

3. mixed?model with random intercept but no fixed time factor:
lmer(y ~ treatment + gender + age + (1|subject)

4. calculate delta.y = difference of y between week 4?& baseline
lm(delta.y ~ treatment + gender + age)

5. same as 4, but add baseline as a factor
lm(delta.y ~ baseline.y + treatment + gender + age)

My thinking on these 5 models are: model 1 and 2 have a limitation that they 
impose a linear relationship of y versus time, which may not be sensible with 2 
time points. Model 3 simply treats baseline and week?4 as repeated measures, not 
imposing linear relationship. Model 4 & 5 are based on the difference between 
baseline and week 4, except that model 5 adds baseline as a covariate. The 
reason of adding baseline as covariate is based on assumption that the extent of 
the change of y between week 4 and baseline depends on the?level of baseline.

Anyone has any suggestions on which one you would use?

Thanks!

John


----- Original Message ----
From: John Maindonald <john.maindonald at anu.edu.au>
To: array chip <arrayprofile at yahoo.com>
Cc: r-sig-mixed-models at r-project.org
Sent: Wed, August 11, 2010 12:04:01 AM
Subject: Re: [R-sig-ME] longitudinal with 2 time points

All these are possibilities, except maybe making baseline measurement
a random factor.? This would make sense only if data divide into groups,
and you want the baseline effect to vary randomly from group to group.? 
That may limit your ability to estimate parameters that are of interest.
In most circumstances that I am familiar with, it makes better sense to 
treat baseline effect as fixed.

John.

John Maindonald? ? ? ? ? ? email: john.maindonald at anu.edu.au
phone : +61 2 (6125)3473? ? fax? : +61 2(6125)5549
Centre for Mathematics & Its Applications, Room 1194,
John Dedman Mathematical Sciences Building (Building 27)
Australian National University, Canberra ACT 0200.
http://www.maths.anu.edu.au/~johnm

Hi, I am wondering if it is still meaningful to run a mixed model if a 
longitudinal dataset has only 2 time points (baseline and week 4)? Would it be 

more appropriate to simply take the difference between the 2 time points and 
run 

ANOVA (ANCOVA) on the difference? what about still running mixed model on the 
difference of the 2 time points, but adding baseline measurement as a random 
factor?

Thanks for sharing your thoughts.

John

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models