Learn R Programming

PSAgraphics (version 2.1.3)

cat.psa: Compare balance graphically of a categorical covariate as part of a PSA

Description

Given predefined strata and two level treatment for a categorical covariate from a propensity score analysis, cat.psa draws pairs of side by side barplots corresponding to control and treatment for each stratum.

Usage

cat.psa(
  categorical,
  treatment = NULL,
  strata = NULL,
  catnames = NULL,
  catcol = "terrain.colors",
  width = 0.25,
  barlab = c("A", "B"),
  barnames = NULL,
  rtmar = 1.5,
  balance = FALSE,
  B = 1000,
  tbl = TRUE,
  cex.leg = 1,
  ...
)

Value

If tbl is TRUE, then a matrix is returned containing the proportions of each category, and in each treatment level and stratum that were used to draw the bargraph.

Arguments

categorical

Vector or N X 3 dataframe or matrix. If a vector, then represents a categorical covariate that is being balanced within strata in a PSA. If categorical has three columns, then the second and third are assumed to be the treatment and strata respectively. Missing values are not allowed. May be factor or numeric.

treatment

Binary vector or factor of same length as continuous representing the two treatments.

strata

A vector or factor of same length as continuous indicating the derived strata from estimated propensity scores. Strata are ordered lexicographically in plot.

catnames

List of names in order of the categories; used in the plot legend. Default is 1:n.

catcol

List of colors used for the categories, default is terrain.colors.

width

Controls width of bars, default = 0.25.

barlab

Binary list of single treatment character labels for the bars, default is c("A", "B"). These are defined in a legend by barnames.

barnames

Binary list of treatment names used in the legend; by default names are taken from treatment.

rtmar

Numeric. Governs size of right margin allocated for legend. Default = 1.5

balance

Logical. If TRUE a call is made to functions bal.cs.psa and bal.cws.psa. The former provides a reference histogram and ad hoc balance statistic, the second provides bootstrapped p-values for the two-way table formed in each statum. Default is FALSE.

B

Numeric; passed to bal.cs.psa governing size of reference histogram generated. Default is 100.

tbl

Logical; if TRUE, then a matrix of the proportions used in the creation of the bargraph is returned.

cex.leg

Numeric; value of cex (governing font size) passed to legend. Default = 1.

...

Other graphical parameters passed to plot.

Author

James E. Helmreich James.Helmreich@Marist.edu

Robert M. Pruzek RMPruzek@yahoo.com

Details

Pairs of bars are graphed side by side so that comparisons may be made within each stratum and across strata. If balance is TRUE, then the histogram represents an ad hoc balance measure of the given strata as compared to randomly generated strata. The p-values provided on the bargraph are bootstrapped in a standard fashion via randomly generated treatment divisions within given strata. For continuous covariates use box.psa.

See Also

bal.cs.psa, bal.cws.psa, box.psa

Examples

Run this code

categorical<-sample(1:7,1000,replace=TRUE)
treatment<-sample(c(0,1),1000,replace=TRUE)
strata<-sample(5,1000,replace=TRUE)
cat.psa(categorical,treatment,strata)

data(lindner)
attach(lindner)
lindner.ps <- glm(abcix ~ stent + height + female +
      diabetic + acutemi + ejecfrac + ves1proc,
      data = lindner, family = binomial)
ps<-lindner.ps$fitted
lindner.s5 <- as.numeric(cut(ps, quantile(ps, seq(0, 1, 1/5)),
      include.lowest = TRUE, labels = FALSE))
cat.psa(stent, abcix, lindner.s5, xlab = "stent")

lindner.s10 <- as.numeric(cut(ps, quantile(ps, seq(0, 1, 1/10)),
      include.lowest = TRUE, labels = FALSE))
cat.psa(ves1proc,abcix, lindner.s10, balance = TRUE, xlab = "ves1proc")

#Using a rpart tree for strata
library(rpart)
lindner.rpart<-rpart(abcix ~ stent + height + female + diabetic +
      acutemi + ejecfrac + ves1proc, data=lindner, method="class")
lindner.tree<-factor(lindner.rpart$where, labels = 1:6)
cat.psa(stent, abcix, lindner.tree, xlab = "stent")
cat.psa(ves1proc, abcix, lindner.tree, xlab = "ves1proc")

Run the code above in your browser using DataLab