loopGEE: Loop through multiple generalized estimating equations (GEE)...

In lwjohnst86/rstatsToolkit: Bundles up all my most used functions for doing statistical analysis

Description

Loop through each combination of dependent and independent variables, running a generalized estimating equations model on each combination, and generate a dataframe of the results.

Usage

loopGEE(data, dependent, independent, id, covariates = NULL,
  interaction = NULL, corstr = "exchangeable", family = gaussian,
  conf.int = TRUE, conf.level = 0.95, na.rm = TRUE,
  filter.indep = FALSE, filter.interact = FALSE, adjust.p.value = FALSE)

Arguments

data	Dataset to run GEE on
dependent	The dependent (aka outcome or response) variables. Must be quoted and can have several.
independent	Like dependent, except the explanatory (aka predictor or exposure) variables.
id	The variable to cluster on for GEE, for instance the 'ID' variable for a person in a longitudinal cohort.
covariates	The covariate variables. Can be multiple covariates.
interaction	A single interaction variable.
corstr	a character string specifying the correlation structure. The following are permitted: '"independence"', '"exchangeable"', '"ar1"', '"unstructured"' and '"userdefined"'
family	See corresponding documentation to glm
conf.int	whether to include a confidence interval
conf.level	confidence level of the interval, used only if conf.int=TRUE
na.rm	Remove missing values from the dataset before running GEE. geeglm can't handle any missingness, so sometimes it's necessary to remove missingness.
filter.indep	Logical; Keep only the rows that have the independent variables.
filter.interact	Logical; Keep only the rows that have the interaction variable.
adjust.p.value	Logical; Adjust for multiple comparisons using the False Discovery Rate.

Value

A dataframe of all the GEE analyses, with estimates, confidence intervals, and p-values.

Author(s)

Luke W. Johnston

See Also

tidy

Examples

data(state)
ds <- data.frame(state.region, state.x77)
names(ds)
outcome <- c('Income', 'Murder')
exposure <- c('Population', 'Life.Exp', 'Illiteracy')
covar <- c('Area', 'Frost')
cid <- 'state.region'

loopGEE(ds, outcome, exposure, cid)
loopGEE(ds, outcome, exposure, cid, covariates = covar)
loopGEE(ds, outcome, exposure, cid, covariates = covar, interaction = 'HS.Grad')
loopGEE(ds, outcome, exposure, cid, covariates = covar, interaction = 'Area')
loopGEE(ds, outcome, exposure, cid, corstr = 'ar1', conf.level = 0.99)
loopGEE(ds, outcome, exposure, cid, corstr = 'ar1', conf.int = FALSE)
loopGEE(ds, outcome, exposure, cid, corstr = 'ar1', adjust.p.value = TRUE)
loopGEE(ds, outcome, exposure, cid, covariates = covar, adjust.p.value = TRUE)
loopGEE(ds, outcome, exposure, cid, covariates = covar, adjust.p.value = TRUE,
        filter.indep = TRUE)
loopGEE(ds, outcome, exposure, cid, interaction = 'Area', covariates = covar,
        adjust.p.value = TRUE, filter.interact = TRUE)

lwjohnst86/rstatsToolkit documentation built on May 21, 2019, 9:15 a.m.