Event-study plot for a did_impute result — event

Produces a ggplot2 event-study figure from a did_impute result object or from explicitly supplied coefficient and standard-error lists. Pre-trend estimates keyed pre<k> are placed at $x = -k$; horizon estimates keyed tau<k> at $x = +k$. Confidence intervals use the critical value $z_{1-\alpha/2}$.

Usage

event_plot(
  results_obj = NULL,
  pretrends = NULL,
  pretrends_std = NULL,
  effects = NULL,
  effects_std = NULL,
  plot_type = c("rcap", "rarea"),
  significance_level = 0.05,
  together = FALSE,
  xlab = "Time Relative to Treatment",
  ylab = "Coefficient",
  title = NULL,
  pretrends_color = "blue",
  effects_color = "red",
  ...
)

Arguments

results_obj: A did_impute result object, or NULL.
pretrends: Named list of pre-trend estimates (keys pre<k>). Used only when results_obj is NULL.
pretrends_std: Named list of pre-trend standard errors.
effects: Named list of effect estimates (keys tau<k>).
effects_std: Named list of effect standard errors.
plot_type: "rcap" (error bars, default) or "rarea" (shaded ribbon).
significance_level: Significance level for CIs (default 0.05, giving 95 percent CIs).
together: Logical; if TRUE pre-trends and effects are combined into a single series (default FALSE).
xlab: X-axis label.
ylab: Y-axis label.
title: Optional plot title.
pretrends_color: Colour for pre-trend series (default "blue").
effects_color: Colour for effects series (default "red").
...: Currently unused.

Value

A ggplot object. The underlying data frame (accessible as p$data) contains columns x, coef, err, and group.

Examples

# Minimal synthetic panel for event-study
set.seed(42)
n_units <- 4; n_t <- 6
df <- expand.grid(i = seq_len(n_units), t = seq_len(n_t))
df$Ei <- ifelse(df$i <= 2, 4L, NA_integer_)
df$y <- 0.5 * (!is.na(df$Ei) & df$t >= df$Ei) + rnorm(nrow(df), sd = 0.2)

res <- did_impute(df, y = "y", i = "i", t = "t", Ei = "Ei",
                  horizons = 0:2, pretrends = 2, minn = 0)
#> The number of treated entities for 'wtr0' is too small for some cohorts. Standard Errors may be wrong; consider using avgeffectsby option, averaging the the effect by treated X post variable.
#> The number of treated entities for 'wtr1' is too small for some cohorts. Standard Errors may be wrong; consider using avgeffectsby option, averaging the the effect by treated X post variable.
#> The number of treated entities for 'wtr2' is too small for some cohorts. Standard Errors may be wrong; consider using avgeffectsby option, averaging the the effect by treated X post variable.
event_plot(res)

event_plot(res, plot_type = "rarea", together = TRUE)