# Load required libraries
library(readxl)
library(dplyr)
library(ggplot2)
library(broom)
library(stargazer)

# Import the Excel file
data <- read_excel("mock_ER_visits_DiD.xlsx")

# Clean and prepare the data
data <- data %>%
  mutate(
    state = factor(state),
    year = as.numeric(year),
    treat = as.factor(treat),
    post = as.factor(post),
    treat_post = as.factor(treat_post)
  )

# Run the DiD regression model
model_did <- lm(er_visits ~ treat + post + treat_post, data = data)

# Display regression results using stargazer
stargazer(model_did, type = "text", title = "Difference-in-Differences Regression", digits = 3)

# Visual inspection of parallel trends assumption
avg_trend <- data %>%
  group_by(state, year) %>%
  summarise(mean_er = mean(er_visits), .groups = 'drop')

ggplot(avg_trend, aes(x = year, y = mean_er, color = state)) +
  geom_line(size = 1.2) +
  geom_vline(xintercept = 2021, linetype = "dashed", color = "black") +
  labs(
    title = "Parallel Trends Check: ER Visits by State",
    y = "Mean ER Visits per 1000 Residents",
    x = "Year"
  ) +
  theme_minimal()