modComp: Compare covariate influence

In renlund/ucR: Ucr Stat Functions

Compare covariate influence

Description

Compare different subsets of covariates within a model

Usage

modComp(resp, vars, model, covars, data = NULL, uni = TRUE, ci = TRUE, ...)

Arguments

resp	the response variable
vars	character vector containing the names of all variables of interest
model	name of the model
covars	a list containing index vectors on vars. Each entry corresponds to an analysis.
data	location of the variables in vars
uni	logical: should univariate analyses be performed. If integer, this is used as an index on vars to determine which univariate analyses are performed)
ci	if TRUE confidence intervals are included
...	arguments passed to transFormat

Note

This function cannot yet incorporate interactions.

Author(s)

Henrik Renlund

Examples

# Comparing two set of covariates in model 'lm'
DF <- data.frame(x1=c(1,2,3,4),x2=c(3,4,0,1))
DF$y <- 2*DF$x1 + DF$x2 + c(0.1, -0.2, 0.05,0.05)
modComp(
   resp = "y",
   vars = c("x1", "x2"),
   model = lm,
   covars = list(1, 1:2),
   data = DF,
   uni = FALSE,
   ci = TRUE,
   round=2
   )
# Comparing different covariates in model 'coxph'
library(survival)
DF <- data.frame(
   x = c(3,1,2,3,2,4,5,6,4,5,3,2,4,1,1,2,3,4,6,7,8,1,1,2,6,4,2,1,1,3,4),
   y = c(0,1,1,0,1,0,0,0,1,0,0,1,0,0,1,0,0,0,0,0,0,0,1,1,0,1,0,0,1,0,0),
   z =  rep(letters[1:2], length.out=31),
   u = rep(c(1:5), length.out=31)
)
cox_endp <- with(DF, Surv(x,y))
modComp(resp = "cox_endp",
   vars = c("z", "u"),
   model=coxph,
   covars=list(1:2),
   data = DF,
   uni=TRUE,
   ci=FALSE,
   round=1,
   fun=exp
)
# Comparing different covariates in model 'glm' NOTE: must incorporate
# the argument "family='binomial'" by defining a function such that this is
# true
Model <- function(formula, data) glm(formula=formula,
family="binomial",data=data)
modComp(resp = "y",
   vars = c("x", "z"),
   model=Model,
   covars=list(1:2, 2),
   data = DF,
   uni=TRUE,
   ci=FALSE,
   signif=3,
   fun=exp
)
rm(Model, cox_endp, DF)

renlund/ucR documentation built on March 25, 2023, 10:10 a.m.