Plots

Event-study, level, and multi-model comparison plots in the package’s figure style (highlight color #E64173, zero/reference lines, direct labels).

R stack_plot(), stack_plot_levels(), stack_plot_compare()   ↔︎   Stata stacked plot

Usage

# event-study coefficients from a fitted model
stack_plot(model, combine = c("overlay", "facet"), include_overall = TRUE, ...)

# Q-weighted level outcomes from a stack
stack_plot_levels(stack_data, outcome_var, weight_var = "q_weight",
                  by_group = FALSE, combine = c("facet", "overlay"), ...)

# overlay several models' event studies
stack_plot_compare(models, label_position = c("direct", "legend", "none"), ...)

Key arguments

Argument Function Description
model stack_plot A fit from stackreg(). A stackreg_groups object triggers per-cohort overlay/facet.
combine stack_plot, stack_plot_levels "overlay" (marker area = cohort weight) or "facet" (one panel per cohort).
include_overall stack_plot Overlay the pooled estimate for grouped objects.
stack_data, outcome_var stack_plot_levels Stack and outcome for the levels view.
weight_var stack_plot_levels "q_weight" (default) or "q_weight_ps".
models stack_plot_compare A named list of models; names label the series.
label_position stack_plot_compare "direct" (default), "legend", or "none".
show_avg_att, title, xlab, ylab, point_color, … all Cosmetic overrides. 
stacked plot [, bygroup combine(str) title(str) noatt twoway_options]

* the level plot is a flag on the levels subcommand:
stacked levels depvar, plot

Options

Option Description
bygroup Plot the last stacked reg, bygroup: overlay (marker size = weight) or, with combine(facet), one panel per cohort; after model(att), a cohort coefficient plot.
combine(str) overlay (default) or facet.
title(str) Plot title.
noatt Suppress the average-post-ATT reference line.
twoway_options Passed through to Stata’s twoway.

stacked plot draws the event study from the most recent stacked reg. The level plot is stacked levels depvar, plot (see levels).

R ↔︎ Stata mapping

R Stata
stack_plot(model) stacked plot (uses the last stacked reg)
stack_plot(groups_obj) / combine stacked plot, bygroup combine()
show_avg_att = FALSE noatt
title title()
stack_plot_levels(stack, y) stacked levels y, plot
stack_plot_compare(models) (no single subcommand; overlay twoway manually, or re-plot each stacked reg)

Value

R functions return a ggplot2 object (or draw a base-R plot and return NULL invisibly if ggplot2 is absent). stack_plot_levels() attaches the underlying means as a "levels_data" attribute.

Stata draws a twoway graph in the current graph window; combine with graph export to save.

Example

library(stacked)
data(medicaid)
stack <- build_stack(medicaid, "year", "state", "adopt_year",
                     kappa_pre = 3, kappa_post = 2)
model <- stackreg(stack, "uninsured", cluster_var = "state")

stack_plot(model, ylab = "Effect on uninsured rate")

stack_plot_levels(stack, "uninsured", ylab = "Uninsured rate")

stacked use medicaid, clear
stacked build, time(year) unit(state) adopt(adopt_year) kpre(3) kpost(2)
stacked reg uninsured, cluster(state)

stacked plot, title("Effect of Medicaid expansion on uninsured rate")
stacked levels uninsured, plot

* by-cohort event study, marker size = aggregation weight
stacked reg uninsured, cluster(state) bygroup
stacked plot, bygroup
stacked plot, bygroup combine(facet)

See also