Plot an event-study result

Base plot() method for es_result objects returned by event_study() (or the deprecated run_es()). Draws the event-study curve with pointwise confidence bands as a static ggplot, or — with interactive = TRUE — an interactive plotly chart with hover tooltips (requires the suggested {plotly} package).

Usage

# S3 method for class 'es_result'
plot(
  x,
  ci_level = 0.95,
  type = c("errorbar", "ribbon"),
  time_axis = c("relative", "calendar"),
  interactive = FALSE,
  show_simultaneous = FALSE,
  ...
)

Arguments

x

An es_result object.

ci_level

Confidence level to display (default 0.95). Must be one of the levels the result was estimated at (conf_level in event_study()).

type

"errorbar" (default) or "ribbon". Static plots only.

time_axis

"relative" (default) for event time, or "calendar" for the original time values. See Details.

interactive

Logical; if TRUE, return an interactive plotly chart instead of a ggplot. Default FALSE.

show_simultaneous

Logical; overlay the simultaneous bootstrap CI band (requires bootstrap = TRUE in the originating event_study() call). Default FALSE.

...

Further styling arguments: for static plots vline_val, hline_val, vline_color, hline_color, linewidth, pointsize, alpha, barwidth, color, fill, errorbar_color, theme_style ("bw", "minimal", or "classic"); for interactive plots markersize, show_ribbon, height, width, and the shared color arguments. vline_val defaults to relative time -1, the last period before treatment, on whichever axis is in use. errorbar_color colours the bars alone and follows color unless set.

Value

A ggplot object, or a plotly object when interactive = TRUE.

Details

The default display is black error bars. type = "ribbon" switches to a line with a shaded band and, unless color/fill say otherwise, to the package's muted steel-blue palette. In either display color sets the points (and the line), while errorbar_color can recolour the bars on their own.

time_axis controls what the horizontal axis shows. The default "relative" plots event time (periods since treatment) and labels the axis "Relative Time to Treatment". "calendar" instead plots the original time values (e.g. 2010, or a Date), labels the axis with the time column's name, and moves the dashed vertical reference line to the treatment period. It is available whenever every treated unit adopts in the same period — universal timing, or a staggered design that happens to have a single cohort — and errors otherwise, because with several cohorts one relative period corresponds to several calendar periods.

See also

event_study(), autoplot.es_result()

Examples

if (FALSE) { # \dontrun{
res <- event_study(y ~ 1 | id + year, data = df, time = year, timing = g,
                   estimator = "cs")
plot(res)
plot(res, type = "ribbon", ci_level = 0.9)
plot(res, color = "#1B4965", errorbar_color = "grey50")
plot(res, interactive = TRUE)

# Universal timing: label the axis with calendar years instead
uni <- event_study(y ~ d | id + year, data = df, time = year,
                   timing = 2010)
plot(uni, time_axis = "calendar")
} # }