Skip to contents

Enhanced Event Study Plot for Difference-in-Differences

did_event_study.Rd

Creates a comprehensive event study (dynamic DiD) plot with confidence bands, reference period annotation, pre-trends test annotation, and support for multiple estimators on the same plot. Works with fixest, did (Callaway-Sant'Anna), and manually supplied coefficients.

Usage

did_event_study(
  models,
  ref_period = -1,
  ci_type = c("pointwise", "simultaneous"),
  conf_level = 0.95,
  show_pretrend_test = TRUE,
  colors = NULL,
  shapes = NULL,
  title = "Event Study: Dynamic Treatment Effects",
  x_label = "Period Relative to Treatment",
  y_label = "Coefficient Estimate",
  vline_color = "gray30",
  dodge_width = 0.3
)

Arguments

models

Named list of model objects or data frames. Each element is one of:

  • A fixest model with i() interaction coefficients.

  • A data frame with columns period, estimate, se (and optionally ci_lower, ci_upper).

  • An aggte object from the did package.

ref_period

Integer. Reference period (normalized to 0). Default -1.

ci_type

Character. Confidence interval type: "pointwise" (default) or "simultaneous" (wider, controls overall error rate).

conf_level

Numeric. Confidence level. Default 0.95.

show_pretrend_test

Logical. Annotate with joint pre-trend test result. Default TRUE.

colors

Character vector. Colors for each model. If NULL, uses a built-in palette.

shapes

Integer vector. Point shapes. Default: different shapes per model.

title

Character. Plot title.

x_label

Character. X-axis label. Default "Period Relative to Treatment".

y_label

Character. Y-axis label. Default "Coefficient Estimate".

vline_color

Character. Color of the treatment timing vertical line. Default "gray30".

dodge_width

Numeric. Horizontal dodge for multiple estimators. Default 0.3.

Value

A ggplot2 object.

Examples

if (FALSE) { # \dontrun{
data(base_stagg)

# Single estimator (TWFE)
mod1 <- feols(y ~ i(time_to_treatment, ref = -1) | id + year, base_stagg)

# Sun-Abraham
mod2 <- feols(y ~ sunab(year_treated, year) | id + year, base_stagg)

did_event_study(
  models  = list("TWFE" = mod1, "Sun-Abraham" = mod2),
  ref_period = -1
)
} # }