Plot propensity (or censoring) score overlap by group

Produces a diagnostic overlap plot for a set of propensity-like scores (e.g., treatment propensity scores or censoring propensities) stratified by a binary group indicator A. The function visualizes the empirical score distributions using semi-transparent histograms on a density scale.

This plot is primarily intended to assess common support / overlap and to diagnose separation or extreme predicted probabilities before matching/weighting steps.

Usage

propensityplot(ps, A)

Arguments

ps

Numeric vector of propensity-like scores, typically in \([0,1]\).

A

Vector defining the grouping variable (e.g., treatment arm or event indicator). Will be coerced to a factor for plotting.

Value

A ggplot2 plot object (class "gg" and "ggplot"). The plot is also printed.

Details

The input A is coerced to a factor and used for coloring and filling the histogram. The y-axis is scaled to density (..density..). The function returns a ggplot2 object (invisibly) and also prints the plot as a side effect, which is convenient in interactive use.

Package note. In a package context, avoid library(ggplot2) inside functions. Instead, use ggplot2:: calls and guard availability via requireNamespace("ggplot2", quietly = TRUE).

See also

ggplot, geom_histogram