profiles.drm: Creates a support matrix for 'drm' for profile probability...

In drm: Regression and association models for repeated categorical data

Description

Creates a support matrix for the calculation of all possible profile probabilities using the ML estimates obtained from drm.

Usage

profiles.drm(n.categories, n.repetitions, structure = "exchangeable")

Arguments

n.categories	Number of respose variable categories
n.repetitions	Number of response repetitions within a cluster
structure	Association structure indicator of the model to be fitted: "exchangeable" if the structure is exchangeable, otherwise either "M" or "M2"

Details

Creation of this support object is useful if the cluster size is large and if several models are fitted to the same dataset. Avoids creating the same matrix over and over again for each model fit. See examples for drm.

Value

If structure="exchangeable", returns a matrix with row and column size equal to n.categories to the power of n.repetitions. For temporal structures "M" and "M2", the result is an array.

See Also

drm

Examples

## use marijuana-data
data(marijuana)

## create matrix for profiles:
W.mari <- profiles.drm(3, 5, "M")

## fit the model using W.mari:
fit <- drm(y~age+cluster(id)+Time(age), data=marijuana,
           subset=sex=="male",pmatrix="W.mari", save.profiles=FALSE,
           dep=list("M",tau12~function(a=1,b=0) a+b*c(0:3),
                    tau21~function(a=1,b=0) a+b*c(0:3)),print=0)

## a speedy way of saving the profile probabilities to the object:
fit <- drm(y~age+cluster(id)+Time(age), data=marijuana,
           subset=sex=="male",pmatrix="W.mari", save.profiles=TRUE,
           dep=list("M",tau12~function(a=1,b=0) a+b*c(0:3),
                    tau21~function(a=1,b=0) a+b*c(0:3)),print=0,
           start=coef(fit))

drm documentation built on May 29, 2017, 7:24 p.m.