Bioequivalence and Bioavailability Forum • ML vs. REML

ML vs. REML [🇷 for BE/BA]

posted by Helmut – Vienna, Austria, 2015-04-20 18:35 (4079 d 12:24 ago) – Posting: # 14717
Views: 31,868

Hi ElMaestro,

❝ There is still something that really bothers me and that is the log Likelihood difference of R versus WNL. With REML I think optimisation switches back and forth be in terms of the estimates. […]

Maybe, maybe not. PHX’ manual only tells me:

The linear mixed effects model is:

y = Xβ + Zγ + ε,
V = Variance(y) =ZGZ^T + R.

Let θ be a vector consisting of the variance parameters in G and R. The full maximum likelihood procedure (ML) would simultaneously estimate both the fixed effects parameters β and the variance parameters θ by maximizing the likelihood of the observations y with respect to these parameters. In contrast, restricted maximum likelihood estimation (REML) maximizes a likelihood that is only a function of the variance parameters θ and the observations y, and not a function of the fixed effects parameters. Hence for models that do not contain any fixed effects, REML would be the same as ML.

❝ ML would only be plain and simple covariance matrix fiddling, I think (?).

Well, we do have fixed effects, right? At least in PHX/WNL, only REML is implemented. IIRC, REML is recommended by Patterson/Jones somewhere.

Thread closed. Please continue over there.

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Thread locked

Complete thread:

Reproducing Bear results in Phoenix mittyri 2015-04-14 08:35
- Bug in bear? Helmut 2015-04-14 14:07
  - Bug in bear? mittyri 2015-04-14 20:43
    - Bug in bear? Helmut 2015-04-15 02:17
  - use the same dataset? yjlee168 2015-04-20 10:13
- Bugs in bear with replicated demo data set? yjlee168 2015-04-15 09:38
- Bugs fixed in bear? yjlee168 2015-04-17 01:21
  - Bugs fixed in bear? Helmut 2015-04-17 14:25
    - Bugs fixed in bear? yjlee168 2015-04-17 18:50
      - Bugs fixed in bear? ElMaestro 2015-04-17 23:14
        
        lme() in bear yjlee168 2015-04-18 11:12
        
        lme() in bear ElMaestro 2015-04-18 11:28
        
        lme() in bear yjlee168 2015-04-18 12:02
        
        lme() in bear ElMaestro 2015-04-18 13:42
        
        Mean means Helmut 2015-04-18 14:08
        
        Mean means ElMaestro 2015-04-18 16:01
        
        Mean means Helmut 2015-04-19 00:59
        
        Mean means ElMaestro 2015-04-19 09:14
        
        LL and AIC Helmut 2015-04-19 11:08
        
        Confused ElMaestro 2015-04-19 11:42
        
        lsmeans for mixed model in R yjlee168 2015-04-19 23:36
        
        lsmeans() & lme() Helmut 2015-04-20 01:28
        
        lsmeans() & lme() yjlee168 2015-04-20 10:23
        lsmeans() & lme() ElMaestro 2015-04-20 10:35
        
        keep it simple! Helmut 2015-04-20 14:34
        
        keep it simple! ElMaestro 2015-04-20 15:39
        
        ML vs. REMLHelmut 2015-04-20 16:35
        
        lme() with NA in R yjlee168 2015-04-19 23:44
      - Dataset Helmut 2015-04-18 13:30
        
        Dataset and lme() in bear yjlee168 2015-04-18 23:14