didididid.r

library(DeclareDesign)
library(tidyverse)
library(ggridges)

design <-
  declare_population(
    N = 1000,
    X1 = rbinom(N, 1, .5),
    X2 = rbinom(N, 1, .5),
    X3 = rbinom(N, 1, .5),
    X4 = rbinom(N, 1, .5),
    X5 = rbinom(N, 1, .5),
    Y = rnorm(N)
  ) +
  declare_estimator(Y ~ X1, model = lm_robust, term = "X1", label = "D") +
  declare_estimator(Y ~ X1*X2, model = lm_robust, term = "X1:X2", label = "DiD") +
  declare_estimator(Y ~ X1*X2*X3, model = lm_robust, term = "X1:X2:X3", label = "DiDiD") +
  declare_estimator(Y ~ X1*X2*X3*X4, model = lm_robust, term = "X1:X2:X3:X4", label = "DiDiDiD") +
  declare_estimator(Y ~ X1*X2*X3*X4*X5, model = lm_robust, term = "X1:X2:X3:X4:X5", label = "DiDiDiDiD")
  
simulations <- simulate_design(design)
ggplot(simulations, aes(x = estimate, y = estimator_label)) +
  geom_density_ridges() +
  theme_ridges() +
  theme(axis.title = element_blank())

variance_df <- 
simulations %>%
  group_by(estimator_label) %>%
  summarise(variance = var(estimate))

variance_df$variance[2:5]/ variance_df$variance[1]