grantmcdermott
commented
3 years ago FWIW, I'm not sure that you want to do this here, but easy enough to add Sun Abraham to the TWFE comparison too. Admittedly, you'd probably want to think a bit more about colours then...
sa = feols(dep_var ~ sunab(g, year) | unit, df_het, cluster=~state)
iplot(list(es, twfe, sa),
sep = 0.1,
ref.line = -0.5,
col = c("steelblue", "#82b446", "#7570B3"),
xlab = "Relative time to treatment",
main = "Event study: Staggered treatment (comparison)")
# Add the true effects
points(-20:20, true_effects, pch = 0, col = "#b44682")
# Legend
legend("topleft", col = c("steelblue", "#b44682", "#82b446", "#7570B3"), pch = c(20, 0, 17, 15),
legend = c("Two-stage estimate", "True effect", "TWFE", "Sun-Abraham"))
Here's a revamped vignette that fixes typos and dead links, as well as simplifies some of the plotting etc. I'm using
fixest::iplot()for the latter to underscore the fixest link. I hope you don't mind, but I dropped your bespoke ggplot2 theme, since that requires several extra packages and I think all of that stuff gets in the way of emphasising the main did2s features.