Diagnostic plots for drlate fits

Usage

# S3 method for class 'drlate'
plot(
  x,
  type = c("overlap", "balance", "balance_density", "weights"),
  bins = 30,
  geom = c("histogram", "density"),
  var = NULL,
  ...
)

Arguments

x

A fitted drlate() object (with keep_data = TRUE).

type

One of:

"overlap": histograms (or kernel densities, see geom) of the estimated instrument propensity score by instrument arm, with the pstolerance bounds marked. Mass piling up near 0 or 1 signals overlap problems.
"balance": a love plot of standardized mean differences from balance(), unweighted vs IPW-weighted, with the conventional |SMD| = 0.1 reference lines.
"balance_density": kernel densities of the covariates by instrument arm, raw versus IPW-weighted (the Stata latebalance density display). Weighting that balances a covariate brings the two arm densities together in the weighted panel.
"weights": distributions of the implied IPW weights by arm; a long right tail means a few observations dominate the estimate.

bins

Number of histogram bins for "overlap" and "weights".

geom

For type = "overlap", either "histogram" (default) or "density" (a kernel-density overlap matching Stata lateoverlap).

var

For type = "balance_density", an optional character vector selecting covariates to plot; defaults to all model covariates.

...

Currently unused.

Value

A ggplot object.