s-news
[Top] [All Lists]

Defining new corStruct class for lme?

To: s-news@lists.biostat.wustl.edu
Subject: Defining new corStruct class for lme?
From: Andrew Cooper <andrew.cooper@unh.edu>
Date: Tue, 7 Mar 2006 09:47:34 -0500 (EST)

Hello S-Newsers,

I've read through Pinheiro and Bates, searched the help files and s-news archives, and even googled, all to no avail (this question has come up on s-news, just no answers posted).

I think (and please correct me if I'm wrong!), I need to create a new
corStruct class.  Does anyone have a primer (or advice) on how to define a
new corStruct class by specifying a constructor function, and methods for
the functions corMatrix and coef? Details of what I'm trying to do are below...

It feels like what I'm wanting to do is straight-forward (which is probably my first mistake). Samples were taken from four lakes, each lake was visited on multiple days (between 8-12 days per year for 3 years during the summer), and on each visit, 3 samples were taken.

I want account for the correlation between samples taken on the same day from a given lake and the correlation between samples taken from different days of that lake.

I could easily do this with a symmetric correlation structure using Lake as the only level (e.g., corSymm(~1|Lake)) and let the model sort out the details. I'd rather test to see if a simpler structure would be as good (e.g., within-day observations being compound symmetric and between day observations of the same lake be either compound symmetric, symmetric, or AR1). The corStrat() function will allow me estimate different correlation structures (or values) within a hierarchy (e.g. each lake having a different correlation), but that's not really what I want, yet.

Any thoughts would be greatly appreciated! I will post a summary of replies to the list.

Thank you for your help!

Cheers,
Andy

******************************************************************
"What if the Hokey Pokey is all it really is about?" - Jimmy Buffett

Andrew B. Cooper, Ph.D.
Department of Natural Resources
Institute for the Study of Earth, Oceans and Space
Morse Hall 142
University of New Hampshire
Durham, NH  03824

andrew.cooper@unh.edu
603.862.4254, 603.862.0243 (FAX)

http://www.unh.edu/natural-resources/fac-cooper.html
******************************************************************

<Prev in Thread] Current Thread [Next in Thread>
  • Defining new corStruct class for lme?, Andrew Cooper <=