used (Mb) gc trigger (Mb) limit (Mb) max used (Mb)
Ncells 597388 32.0 1356698 72.5 NA 700248 37.4
Vcells 1100546 8.4 8388608 64.0 32768 1963378 15.0cat("\f") # Clear the consolegraphics.off() # Clear the chartsDiff-in-Diff Assumptions
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 ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr 1.1.4 ✔ readr 2.1.5
✔ forcats 1.0.0 ✔ stringr 1.5.1
✔ ggplot2 3.5.1 ✔ tibble 3.2.1
✔ lubridate 1.9.3 ✔ tidyr 1.3.1
✔ purrr 1.0.2 ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag() masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsEXAMPLE 1: Estimating Diff-in-Diff
Getting sample data.
'data.frame': 70 obs. of 9 variables:
$ country: Factor w/ 7 levels "A","B","C","D",..: 1 1 1 1 1 1 1 1 1 1 ...
$ year : int 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 ...
$ y : num 1.34e+09 -1.90e+09 -1.12e+07 2.65e+09 3.01e+09 ...
$ y_bin : num 1 0 0 1 1 1 1 1 1 1 ...
$ x1 : num 0.278 0.321 0.363 0.246 0.425 ...
$ x2 : num -1.108 -0.949 -0.789 -0.886 -0.73 ...
$ x3 : num 0.2826 0.4925 0.7025 -0.0944 0.9461 ...
$ opinion: Factor w/ 4 levels "Str agree","Agree",..: 1 3 3 3 3 1 3 1 3 4 ...
$ op : num 1 0 0 0 0 1 0 1 0 0 ...
- attr(*, "datalabel")= chr ""
- attr(*, "time.stamp")= chr " 3 Jan 2020 11:28"
- attr(*, "formats")= chr [1:9] "%14.0g" "%8.0g" "%10.0g" "%9.0g" ...
- attr(*, "types")= int [1:9] 253 252 255 254 254 254 254 254 254
- attr(*, "val.labels")= chr [1:9] "country" "" "" "" ...
- attr(*, "var.labels")= chr [1:9] "Country" "Year" "Outcome Y" "Binary outcome Y" ...
- attr(*, "expansion.fields")=List of 35
..$ : chr [1:3] "_dta" "_dta" "st"
..$ : chr [1:3] "_dta" "st_ver" "2"
..$ : chr [1:3] "_dta" "st_id" "new_period"
..$ : chr [1:3] "_dta" "st_bt" "elapsed"
..$ : chr [1:3] "_dta" "st_bd" "liberalization"
..$ : chr [1:3] "_dta" "st_o" "_origin"
..$ : chr [1:3] "_dta" "st_orig" "time entrydate"
..$ : chr [1:3] "_dta" "st_orexp" "entrydate"
..$ : chr [1:3] "_dta" "st_s" "365.25"
..$ : chr [1:3] "_dta" "st_bs" "365.25"
..$ : chr [1:3] "_dta" "st_d" "_d"
..$ : chr [1:3] "_dta" "st_t0" "_t0"
..$ : chr [1:3] "_dta" "st_t" "_t"
..$ : chr [1:3] "_dta" "st_ornl" "1"
..$ : chr [1:3] "_dta" "st_orvn" "begin"
..$ : chr [1:3] "_dta" "_lang_list" "default"
..$ : chr [1:3] "_dta" "_lang_c" "default"
..$ : chr [1:3] "_dta" "ReS_i" "country"
..$ : chr [1:3] "_dta" "ReS_ver" "v.2"
..$ : chr [1:3] "_dta" "ReS_j" "year"
..$ : chr [1:3] "_dta" "ReS_str" "0"
..$ : chr [1:3] "_dta" "ReS_Xij" "schooling"
..$ : chr [1:3] "year" "destring" "Characters removed were:"
..$ : chr [1:3] "_dta" "st_exvn" "regtrans"
..$ : chr [1:3] "_dta" "st_exnl" "-66 -77 98"
..$ : chr [1:3] "_dta" "st_exexp" "."
..$ : chr [1:3] "_dta" "st_enexp" "entrydate"
..$ : chr [1:3] "_dta" "__xi__Vars__To__Drop__" "_Icountry_2 _Icountry_3 _Icountry_4 _Icountry_5 _Icountry_6 _Icountry_7"
..$ : chr [1:3] "_dta" "__xi__Vars__Prefix__" "_I _I _I _I _I _I"
..$ : chr [1:3] "_dta" "_TStvar" "year"
..$ : chr [1:3] "_dta" "_TSpanel" "country"
..$ : chr [1:3] "_dta" "_TSdelta" "+1.0000000000000X+000"
..$ : chr [1:3] "_dta" "_TSitrvl" "1"
..$ : chr [1:3] "_dta" "tis" "year"
..$ : chr [1:3] "_dta" "iis" "country"
- attr(*, "version")= int 12
- attr(*, "label.table")=List of 2
..$ agree : Named int [1:4] 1 2 3 4
.. ..- attr(*, "names")= chr [1:4] "Str agree" "Agree" "Disag" "Str disag"
..$ country: Named int [1:7] 1 2 3 4 5 6 7
.. ..- attr(*, "names")= chr [1:7] "A" "B" "C" "D" ...stargazer(mydata,
type = "text")
=============================================================================
Statistic N Mean St. Dev. Min Max
-----------------------------------------------------------------------------
year 70 1,994.500 2.893 1,990 1,999
y 70 1,845,072,181.000 3,015,166,973.000 -7,863,482,880 8,941,232,128
y_bin 70 0.800 0.403 0 1
x1 70 0.648 0.468 -0.568 1.446
x2 70 0.134 1.371 -1.622 2.530
x3 70 0.762 1.446 -1.165 7.169
op 70 0.500 0.504 0 1
-----------------------------------------------------------------------------table(mydata$year)
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
7 7 7 7 7 7 7 7 7 7 Create a dummy variable to indicate the time when the treatment started. Lets assume that treatment started in 1994. In this case, years before 1994 will have a value of 0 and 1994 onwards a value of 1.
mydata$time = ifelse(test = mydata$year >= 1994,
yes = 1,
no = 0)Create a dummy variable to identify the group exposed to the treatment.
In this example lets assumed that countries with code 5,6, and 7 were treated (
=1).Countries 1-4 were not treated (
=0). If you already have this skip this step.
Interaction Term
Create an interaction between time and treated. We will call this interaction ‘did’.
mydata$did = mydata$time * mydata$treatedEstimating the DID estimator
Call:
lm(formula = y ~ treated + time + did, data = mydata)
Residuals:
Min 1Q Median 3Q Max
-9.768e+09 -1.623e+09 1.167e+08 1.393e+09 6.807e+09
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.581e+08 7.382e+08 0.485 0.6292
treated 1.776e+09 1.128e+09 1.575 0.1200
time 2.289e+09 9.530e+08 2.402 0.0191 *
did -2.520e+09 1.456e+09 -1.731 0.0882 .
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 2.953e+09 on 66 degrees of freedom
Multiple R-squared: 0.08273, Adjusted R-squared: 0.04104
F-statistic: 1.984 on 3 and 66 DF, p-value: 0.1249The coefficient for did is the differences-in-differences estimator. The effect is significant at 10% with the treatment having a negative effect.
Estimating the DID estimator
Using the multiplication method, no need to generate the interaction.
Call:
lm(formula = y ~ treated * time, data = mydata)
Residuals:
Min 1Q Median 3Q Max
-9.768e+09 -1.623e+09 1.167e+08 1.393e+09 6.807e+09
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.581e+08 7.382e+08 0.485 0.6292
treated 1.776e+09 1.128e+09 1.575 0.1200
time 2.289e+09 9.530e+08 2.402 0.0191 *
treated:time -2.520e+09 1.456e+09 -1.731 0.0882 .
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 2.953e+09 on 66 degrees of freedom
Multiple R-squared: 0.08273, Adjusted R-squared: 0.04104
F-statistic: 1.984 on 3 and 66 DF, p-value: 0.1249stargazer(didreg, didreg1,
type = "text")
=====================================================================
Dependent variable:
---------------------------------------
y
(1) (2)
---------------------------------------------------------------------
treated 1,775,969,673.000 1,775,969,673.000
(1,127,561,848.000) (1,127,561,848.000)
time 2,289,454,651.000** 2,289,454,651.000**
(952,963,694.000) (952,963,694.000)
did -2,519,511,630.000*
(1,455,676,087.000)
treated:time -2,519,511,630.000*
(1,455,676,087.000)
Constant 358,143,950.000 358,143,950.000
(738,162,503.000) (738,162,503.000)
---------------------------------------------------------------------
Observations 70 70
R2 0.083 0.083
Adjusted R2 0.041 0.041
Residual Std. Error (df = 66) 2,952,650,012.000 2,952,650,012.000
F Statistic (df = 3; 66) 1.984 1.984
=====================================================================
Note: *p<0.1; **p<0.05; ***p<0.01EXAMPLE 2
data <- read.csv("us_fred_coastal_us_states_avg_hpi_before_after_2005.csv")
stargazer(data,
type = "text")
================================================
Statistic N Mean St. Dev. Min Max
------------------------------------------------
HPI_CHG 48 0.022 0.017 -0.006 0.061
Time_Period 48 0.500 0.505 0 1
Disaster_Affected 48 0.208 0.410 0 1
NUM_DISASTERS 48 3.208 2.143 1 10
NUM_IND_ASSIST 48 8.583 14.946 0 55
------------------------------------------------did_model_results <-
lm(data = data,
formula = HPI_CHG ~ Time_Period * Disaster_Affected)
summary(did_model_results)
Call:
lm(formula = HPI_CHG ~ Time_Period * Disaster_Affected, data = data)
Residuals:
Min 1Q Median 3Q Max
-0.023081 -0.007610 -0.000171 0.004656 0.035981
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.037090 0.002819 13.157 < 2e-16 ***
Time_Period -0.027847 0.003987 -6.985 1.2e-08 ***
Disaster_Affected -0.013944 0.006176 -2.258 0.0290 *
Time_Period:Disaster_Affected 0.019739 0.008734 2.260 0.0288 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.01229 on 44 degrees of freedom
Multiple R-squared: 0.5356, Adjusted R-squared: 0.504
F-statistic: 16.92 on 3 and 44 DF, p-value: 1.882e-07table(data$Time_Period,
data$Disaster_Affected
)
0 1
0 19 5
1 19 5# Create a two-way table with labels
raw_table <- table(Time_Period = ifelse(test = data$Time_Period == 0, yes = "pre", no = "post"),
Treatment_Status = ifelse(data$Disaster_Affected == 0, yes = "control", no = "treatment")
)
raw_table Treatment_Status
Time_Period control treatment
post 19 5
pre 19 5?tapply # Apply a function to each cell of a ragged array, that is to each (non-empty) group of values given by a unique combination of the levels of certain factors.
# Calculate means for each group
mean_table <- tapply(X = data$HPI_CHG,
INDEX = list(data$Time_Period,
data$Disaster_Affected),
FUN = mean)
mean_table 0 1
0 0.037090020 0.02314612
1 0.009242792 0.01503835# Calculate means for each group
mean_table <- tapply(data$HPI_CHG, list(Time_Period = ifelse(data$Time_Period == 0, "pre", "post"),
Treatment_Status = ifelse(data$Disaster_Affected == 0, "control", "treatment")),
mean)
# Display the table
print(mean_table) Treatment_Status
Time_Period control treatment
post 0.009242792 0.01503835
pre 0.037090020 0.02314612# Calculate DiD effect
DiD_effect <- (mean_table[1, 2] - mean_table[2, 2]) - (mean_table[1, 1] - mean_table[2, 1])
print(DiD_effect)[1] 0.01973946Source:
Identifying Assumptions
Difference-in-differences (DiD) analysis relies on certain assumptions to identify the causal effect of a treatment. The key threats to identification or implicit assumptions in DiD include:
Parallel Trends Assumption:
Assumption: The average treatment and control groups would have followed parallel trends in the absence of the treatment.
Implication: Any difference in trends post-treatment can be attributed to the treatment effect.
-
Threat: If there are pre-existing trends or shocks that affect the treatment and control groups differently, the parallel trends assumption may be violated.
Common Time Trends:
Assumption: Unobservable factors affecting outcomes in both groups evolve similarly over time.
Implication: Controls for unobserved time-varying factors.
Threat: If there are unobserved time-varying factors that systematically differ between treatment and control groups and are related to the outcomes, the estimates may be biased.
No Spillover Effects:
Assumption: The treatment does not have direct or indirect effects on the control group.
Implication: The treatment effect is isolated to the treated units.
Threat: If there are spillover effects, where the treatment affects the control group, it could confound the DiD estimates.
No Other Contemporaneous Events:
Assumption: There are no other contemporaneous events that systematically affect the treatment and control groups differently.
Implication: Any observed changes in outcomes can be attributed to the treatment.
Threat: If other events occur simultaneously and affect the groups differently, they may confound the estimates.
Sufficient Time Periods:
Assumption: There are enough pre-treatment and post-treatment time periods to observe trends.
Implication: Provides information to identify and control for trends.
Threat: With insufficient time periods, it may be challenging to establish pre-treatment trends or control for changes in trends.
Homogeneous Treatment Effects:
Assumption: The treatment effect is homogeneous across all units.
Implication: Assumes that the treatment effect does not vary systematically across different units.
Threat: If treatment effects vary, aggregating them may obscure heterogeneity.
It’s important to note that, like in any statistical method, the validity of DiD results depends on the robustness of these assumptions.
Violations of these assumptions can lead to biased estimates. Researchers often perform sensitivity analyses and robustness checks to assess the impact of potential violations on the results. Additionally, employing matching methods or incorporating additional control variables can help address some of these threats.