Propensity Score Diagnostics Panel — propensity

Evaluates overlap, common support, and calibration of estimated propensity scores. All three plots are returned individually and as a combined figure.

Usage

propensity_diagnostics(
  ps_scores,
  treatment,
  data = NULL,
  method = c("logit", "rf", "any"),
  title = NULL
)

Arguments

ps_scores: Numeric vector of estimated propensity scores (values in (0, 1)).
treatment: Integer or logical vector of the same length as ps_scores. Treatment indicator (1 = treated, 0 = control).
data: Data frame or NULL. If supplied, it is used only for labelling purposes.
method: Character. Label for the propensity score estimation method shown in plot titles (e.g. "logit", "rf"). Default "any".
title: Character or NULL. Optional overall title prefix.

Value

A named list:

overlap_plot: ggplot2 density overlap plot.
histogram_plot: ggplot2 histogram by treatment group.
calibration_plot: ggplot2 calibration plot.
combined_plot: Patchwork combined figure (or overlap_plot if patchwork is unavailable).
overlap_stats: Named numeric vector: min_treated, max_control, common_support_lo, common_support_hi, pct_off_support.

Details

Comprehensive Propensity Score Diagnostics

Produces a set of diagnostic plots and statistics for evaluating a propensity score model: (1) an overlap density plot, (2) a histogram by treatment group with trimming bounds, and (3) a calibration plot comparing predicted treatment probability with observed treatment rate by decile. The three panels are assembled into a combined figure via patchwork if available.

Examples

set.seed(42)
n   <- 400
x1  <- rnorm(n)
x2  <- rnorm(n)
lp  <- -0.5 + 0.8 * x1 + 0.5 * x2
ps  <- 1 / (1 + exp(-lp))
trt <- rbinom(n, 1, ps)

diag <- propensity_diagnostics(ps_scores = ps, treatment = trt)
diag$combined_plot

diag$overlap_stats
#>       min_treated       max_treated       min_control       max_control 
#>        0.04643335        0.86449771        0.03180591        0.82869973 
#> common_support_lo common_support_hi   pct_off_support 
#>        0.04643335        0.82869973        0.50000000