Comments on: R-Sessions 25: Book – Mixed Effects Models in S and S-PLUS (Pinheiro & Bates, 2000)

By: Bernd

Bernd — Tue, 07 Oct 2008 07:47:33 +0000

... and here's the URL: Douglas Bates "Using lme4: Mixed-Effects Modeling in R".

By: Rense Nieuwenhuis

Rense Nieuwenhuis — Tue, 07 Oct 2008 07:20:05 +0000

Ah yes, that makes sense. Somehow, when I read ‘hierarchical model’ my brain reads something like ‘model with any kind of hierarchy’, bringing it to some sort of synonym. However, taking a more strict understanding of the term is much more clear. Thanks again!

Oh: for those who do not know what Bernd refers to: We both attended a tutorial session on mixed models by Douglas Bates at the useR! 2008 conference (in Dortmund).

By: Bernd

Bernd — Tue, 07 Oct 2008 06:09:47 +0000

It’s on Doug’s slides (p39, “Models with crossed random effects”): “Many people believe that mixed-effects models are equivalent to hierarchical linear models (HLMs) or ‘multilevel models’. This is not true. The plate and sample factors in fm2 are crossed. They do not represent levels in a hierarchy.”

By: Rense Nieuwenhuis

Rense Nieuwenhuis — Thu, 02 Oct 2008 10:10:25 +0000

Ah, thanks for the comment. I must admit I can’t remember what Douglas Bates stated about the difference between mixed and multilevel models. I always thought these terms were used rather interchangeably. Perhaps you can elaborate?

By: Bernd

Bernd — Wed, 01 Oct 2008 11:53:56 +0000

I really like the book as it offers a thorough discussion of the linear and non-linear mixed effects models (BTW: not multilevel models, Doug Bates mentioned that issue in Dortmund ;-). But chapter 2 is really hard stuff, especially for a non-statistician.