#Packages list
library(stargazer)
Please cite as: Hlavac, Marek (2022). stargazer: Well-Formatted Regression and Summary Statistics Tables. R package version 5.2.3. https://CRAN.R-project.org/package=stargazer library(dplyr)
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionlibrary(ggplot2)dt1 <- read.csv("us_fred_coastal_us_states_avg_hpi_before_after_2005.csv")\[ HPICHG_i = \beta_0 + \beta_1 Timeperiod_i + \beta_2 Disasteraffected_i + \beta_3 Timeperiod_i \ * \ Disasteraffected_i + \epsilon \]
reg1 <- lm(data = dt1,
formula = HPI_CHG ~ Time_Period + Disaster_Affected + Time_Period * Disaster_Affected)
stargazer(reg1,
type = "text")
=========================================================
Dependent variable:
---------------------------
HPI_CHG
---------------------------------------------------------
Time_Period -0.028***
(0.004)
Disaster_Affected -0.014**
(0.006)
Time_Period:Disaster_Affected 0.020**
(0.009)
Constant 0.037***
(0.003)
---------------------------------------------------------
Observations 48
R2 0.536
Adjusted R2 0.504
Residual Std. Error 0.012 (df = 44)
F Statistic 16.916*** (df = 3; 44)
=========================================================
Note: *p<0.1; **p<0.05; ***p<0.01The control group are states that were not affected by any disaster, more specifically, the observations that have a 0 on Disaster_affected.
The treatment group are states that were affected by any disaster, more specifically, the observations that have a 1 on Disaster_affected.
When using DID, the goal is to isolate the treatment effect on a variable of interest. We look at the difference between the treatment group and the control group. We achieve this by looking at the difference between the average outcome of our treatment group (observed) and the average outcome of a hypothetical treatment group that did not receive the treatment. This works because of the basic assumption that the treatment group would have followed a similar pattern as the control group had the treatment not being applied.
time <- c(1, 2, 3, 4, 5)
treatment_outcomes <- c(11, 12, 13, 14, 20)
control_outcomes <- c(10, 11, 12, 13, 14)
hypothetical_outcomes <- c(11, 12, 13, 14, 15)
df <- data.frame(time = rep(time, 3),
group = rep(c("Treatment", "Control", "Hypothetical"), each = length(time)),
outcomes = c(treatment_outcomes, control_outcomes, hypothetical_outcomes))
# Plotting
ggplot(df, aes(x = time, y = outcomes, color = group, linetype = group)) +
geom_line(size = 1) +
labs(x = "Time", y = "Outcomes", color = "Group", linetype = "Group") +
ggtitle("Difference-in-Differences (DID) Graph") +
scale_linetype_manual(values = c("Treatment" = "solid", "Control" = "solid", "Hypothetical" = "dashed")) +
geom_segment(aes(x = max(time), xend = 5 , y = max(treatment_outcomes), yend = max(hypothetical_outcomes)) ,
color = "black", size = 1) +
theme_minimal() Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.
The blue line is the treatment group
The red line is the control group
The dashed line is the hypothetical case where the treatment group did not receive the treatment
The black line represents the difference in difference effect
mean_table <- tapply(dt1$HPI_CHG, list(Time_Period = ifelse(dt1$Time_Period == 0, "pre", "post"),
Treatment_Status = ifelse(dt1$Disaster_Affected == 0, "control", "treatment")),
mean)
print(mean_table) Treatment_Status
Time_Period control treatment
post 0.009242792 0.01503835
pre 0.037090020 0.02314612DiD_effect <- (mean_table[1, 2] - mean_table[2, 2]) - (mean_table[1, 1] - mean_table[2, 1])
print(DiD_effect)[1] 0.01973946The average outcome of the treatment group would have followed a parallel trend to the control group had the treatment not being applied.
Factors that not accounted for in the model that affect both groups evolve similarly over time.
The treatment does not have direct or indirect effects on the control group
There are no events that affect both the treatment and control group differently
There are enough pre-treatment and post-treatment time periods to observe trends.
The treatment effect is homogeneous across all units.
It is important to understand that these assumptions are the building base of the DID models, if they are violated, then the models could contain bias so each study done with DID have to carefully evaluated to test the robustest and the accuracy of the results.