Dear Scott, I believe that normally the display formula is just used to construct graphs of the data, although when a grouped-data object is passed in the data argument to lme(), the grouping structure can be obtained from the object and need not be specified explicitly. As well, it is possible to specify additional "outer" and "inner" covariates beyond the principal covariate given in the display formula, and these can be used in graphs of the data. All this is described, if I remember right, in the section on grouped-data objects in Pinheiro and Bates. I hope that this helps, John
At 07:50 PM 2/6/2004 -0500, Scott Rifkin wrote:
I'm trying to set up a mixed model to solve using lme. It will have 3 fixed effects, two random effects and two interaction terms. I've been reading Pinheiro's and Bates's book on the nmle library, but find the part about display functions to be unclear. When creating a groupedData object from a data.frame, you need to enter a function of the form: response ~primary|grouping Does this display function serve any purpose later on when specifying the model in the call to lme? If so, how would I include more terms in primary or grouping (I tried primary1+primary2+... as per one post in the archives, but it didn't work). If not, is it just for purposes of displaying the data?
----------------------------------------------------- John Fox Department of Sociology McMaster University Hamilton, Ontario, Canada L8S 4M4 email: jfox at mcmaster.ca phone: 905-525-9140x23604 web: www.socsci.mcmaster.ca/jfox