Fit a generalized estimating equation (GEE) model with fixed additive correlation structure

The geekin function fits generalized estimating equations but where the correlation structure is given as linear function of (scaled) fixed correlation structures.

geekin(formula, family = gaussian, data, weights, subset, id, na.action,
  control = geepack::geese.control(...), varlist, ...)

Arguments

formula	See corresponding documentation to glm.
family	See corresponding documentation to glm.
data	See corresponding documentation to glm.
weights	See corresponding documentation to glm.
subset	See corresponding documentation to glm.
id	a vector which identifies the clusters. The length of id should be the same as the number of observations. Data must be sorted so that observations on a cluster are contiguous rows for all entities in the formula. If not the function will give an error
na.action	See corresponding documentation to glm.
control	See corresponding documentation to glm.
varlist	a list containing one or more matrix or bdsmatrix objects that represent the correlation structures
…	further arguments passed to or from other methods.

Value

Returns an object of type geeglm.

Details

The geekin function is essentially a wrapper function to geeglm. Through the varlist argument, it allows for correlation structures of the form

R = sum_i=1^k alpha_i R_i

where alpha_i are(nuisance) scale parameters that are used to scale the off-diagonal elements of the individual correlation matrices, R_i.

See also

lmekin, geeglm

Examples

 # Get dataset
 library(kinship2)
 library(mvtnorm)
 data(minnbreast)

 breastpeda <- with(minnbreast[order(minnbreast$famid), ], pedigree(id,
                   fatherid, motherid, sex,
                   status=(cancer& !is.na(cancer)), affected=proband,
                   famid=famid))

set.seed(10)

nfam <- 6
breastped <- breastpeda[1:nfam]

 # Simulate a response

# Make dataset for lme4
df <- lapply(1:nfam, function(xx) {
            as.data.frame(breastped[xx])
            })

mydata <- do.call(rbind, df)
mydata$famid <- rep(1:nfam, times=unlist(lapply(df, nrow)))

y <- lapply(1:nfam, function(xx) {
            x <- breastped[xx]
            rmvtnorm.pedigree(1, x, h2=0.3, c2=0)
            })
yy <- unlist(y)

library(geepack)

geekin(yy ~ 1, id=mydata$famid, varlist=list(2*kinship(breastped)))
#> 
#> Call:
#> geekin(formula = yy ~ 1, id = mydata$famid, varlist = list(2 * 
#>     kinship(breastped)))
#> 
#> Coefficients:
#> (Intercept) 
#>  -0.1709465 
#> 
#> Degrees of Freedom: 225 Total (i.e. Null);  224 Residual
#> 
#> Scale Link:                   identity
#> Estimated Scale Parameters:  [1] 0.8873601
#> 
#> Correlation:  Structure = userdefined    Link = identity 
#> Estimated Correlation Parameters:
#>   alpha:1 
#> 0.2659875 
#> 
#> Number of clusters:   6   Maximum cluster size: 47 
#> 

# lmekin(yy ~ 1 + (1|id), data=mydata, varlist=list(2*kinship(breastped)),method="REML")