mmiGEE function

Multi-model inference for GEE models

mmiGEE is a multimodel inference approach evaluating the relative importance of predictors used in GEE.

@details It performs automatically generated model selection and creates a model selection table according to the approach of multi-model inference (Burnham & Anderson, 2002). QIC is used to obtain model selection weights and to rank the models. Moreover, mmiGEE calculates relative variable importance of a given model. Finally, this function requires that all predictor variables be continuous .

mmiGEE(object, data, trace = FALSE)

Arguments

• object: A model of class GEE.
• data: A data frame or set of vectors of equal length.
• trace: A logical indicating whether or not to print results to console.

Returns

mmiGEE returns a list containing the following elements

• result: A matrix containing slopes, degrees of freedom, quasilikelihood, QIC, delta, and weight values for the set of candidate models. The models are ranked by QIC.
• rvi: A vector containing the relative importance of each variable in the regression.

Details

Calculates the relative importance of each variable using multi-model inference methods in a Generalized Estimating Equations framework implemented in GEE.

Examples

# data (for demonstration only)
library(MASS)
data(birthwt)

# impose an artificial (not fully appropriate) grid structure

x <- rep(1:14, 14)
y <- as.integer(gl(14, 14))
coords <- cbind(x[-(190:196)], y[-(190:196)])

## Not run:

formula <- formula(low ~ race + smoke +  bwt)

mgee <- GEE(formula,
            family = "gaussian",
            data = birthwt,
            coord = coords,
            corstr = "fixed",
            scale.fix = TRUE)

mmi <- mmiGEE(mgee, birthwt)
## End(Not run)

References

Burnham, K.P. & Anderson, D.R. (2002) Model selection and multimodel inference. Springer, New York.

Carl G & Kuehn I, 2007. Analyzing Spatial Autocorrelation in Species Distributions using Gaussian and Logit Models, Ecol. Model. 207, 159 - 170

See Also

GEE, qic.calc, MuMIn

Author(s)

Gudrun Carl, Sam Levin