plot.PanelBalance: Plot covariate balance results Create figures displaying...

In PanelMatch: Matching Methods for Causal Inference with Time-Series Cross-Sectional Data

Plot covariate balance results Create figures displaying covariate balance results for one or more PanelMatch configurations. Users can customize these visualizations.

Description

Plot covariate balance results Create figures displaying covariate balance results for one or more PanelMatch configurations. Users can customize these visualizations.

Usage

## S3 method for class 'PanelBalance'
plot(
  x,
  ...,
  type = "panel",
  reference.line = TRUE,
  legend = TRUE,
  ylab = NULL,
  include.treatment.period = TRUE,
  include.unrefined.panel = TRUE,
  legend.position = "topleft"
)

Arguments

x	PanelBalance object
...	additional parameters to be passed to base::plot()
type	character specifying which type of plot to produce. Can be either "panel" or "scatter". When "panel," covariate balance results for covariates are shown over the lag period. When "scatter," the figure has the following characteristics. Each point on the plot represents a specific covariate at a particular time period in the lag window from t-L to t-1. The horizontal axis represents the covariate balance for this particular variable and time period before refinement is applied, while the vertical axis represents the post-refinement balance value.
reference.line	logical. Include a reference line at y = 0? Only applicable to the panel plot.
legend	logical. Describes whether or not to include a legend.
ylab	character. Y-axis label.
include.treatment.period	Logical. Describes whether or not the treatment period should be included on the panel plot. Default is TRUE.
include.unrefined.panel	logical indicating whether or not unrefined balance plots should be returned for panel plot. Only applicable to panel plot. Default is TRUE.
legend.position	character. Describes where the legend should be placed on the figure. Uses base R syntax.

Value

returns a set of base R plots, depending on the specification of "panel" or "scatter" above. When type = "panel" and include.unrefined.panel = TRUE, two sets of plots are returned. The first set shows covariate balance levels for the specified PanelMatch configurations. The second set shows covariate balance levels for the same PanelMatch configurations, but with all control units receiving equal weight (i.e., balance levels prior to refinement). If include.unrefined.panel = FALSE, only the first set of figures are returned. The sets of figures are both returned in the same order as the PanelMatch configurations specified to get_covariate_balance() that compose the PanelBalance object. When type = "scatter", the visualization described above is produced, with all configurations shown on the same plot with different symbols.

Examples

dem$rdata <- runif(runif(nrow(dem)))
dem.panel <- PanelData(dem, "wbcode2", "year", "dem", "y")
pm.obj <- PanelMatch(lead = 0:3, lag = 4, refinement.method = "mahalanobis", 
                     panel.data = dem.panel, match.missing = TRUE,
                     covs.formula = ~ tradewb + rdata + I(lag(tradewb, 1:4)) + I(lag(y, 1:4)), 
                     size.match = 5, qoi = "att")

# create multiple configurations to compare
pm2 <- PanelMatch(lead = 0:3, lag = 4, refinement.method = "ps.match", 
                  panel.data = dem.panel, match.missing = TRUE,
                  covs.formula = ~ tradewb + rdata + I(lag(tradewb, 1:4)) + I(lag(y, 1:4)), 
                  size.match = 5, qoi = "att")

pb <- get_covariate_balance(pm.obj, pm2,
                            include.unrefined = TRUE,
                            panel.data = dem.panel, 
                            covariates = c("tradewb", "rdata"))
plot(pb, type = "panel", include.unrefined.panel = TRUE)
plot(pb, type = "scatter")
# only show refined balance figures
plot(pb, type = "panel", include.unrefined.panel = FALSE)

PanelMatch documentation built on April 3, 2025, 6:34 p.m.