-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathpps.R
166 lines (107 loc) · 3.74 KB
/
pps.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
#https://sejdemyr.github.io/r-tutorials/statistics/tutorial8.html
install.packages("ResourceSelection")
install.packages("pROC")
install.packages("car")
install.packages("car")
install.packages("rsq")
install.packages("rms")
#additions
install.packages("dplyr")
install.packages("magrittr")
install.packages("MatchIt")
library(dplyr)
library(ggplot2)
library(MatchIt)
library(gridExtra)
library(pROC)
library(ResourceSelection)
library(car)
library(rsq)
library(rms)
# additions
library(magrittr)
dataset <- read.csv(path)
dataset %>%
group_by(treated) %>%
summarise(observations = n(),
mean_math = mean(change),
std_error = sd(change) / sqrt(observations))
#Welch Two Sample t-test
with(dataset, t.test(change ~ treated))
#Covariate means
dataset_cov <- c('sett_mean', 'pop_mean', 'road_mean') #c , 'ag16_sum' ('vv_max', 'vv_min', 'vv_std')
dataset %>%
group_by(treated) %>%
select(one_of(dataset_cov)) %>%
summarise_all(list(~mean(., na.rm = T)))
# T-tests
lapply(dataset_cov, function(v) {
t.test(dataset[, v] ~ dataset[, 'treated'])
})
# GLM
m_ps = glm(treated ~ pop_mean+road_mean+sett_mean,
family = binomial(), data = dataset)
summary(m_ps)
# R2
rsq(m_ps,adj=FALSE,type=c('v','kl','sse','lr','n')) #,data=NULL)
# Multicollinearity
vif(m_ps)
#### ROC and AUC ####
par(pty = "s")
roc(dataset$treated, m_ps$fitted.values, plot=TRUE, legacy.axes=TRUE, percent=TRUE, xlab="False Positive Percentage", ylab="True Postive Percentage", col="#377eb8", lwd=4)
# Propensity score calculation for each observation
prs_df <- data.frame(pr_score = predict(m_ps, type = "response"),
treated = m_ps$model$treated)
#head(prs_df)
# Examining the region of common support
labs <- paste("treatment status:", c("treated", "not treated"))
prs_df %>%
mutate(treated = ifelse(treated == 1, labs[1], labs[2])) %>%
ggplot(aes(x = pr_score)) +
geom_histogram(color = "white") +
facet_wrap(~treated) +
xlab("Probability of being treated") +
theme_bw()
# Matching using MatchIt
dataset_nomiss <- dataset %>% # MatchIt does not allow missing values
select(change, treated, one_of(dataset_cov)) %>%
na.omit()
mod_match <- matchit(treated ~ pop_mean + road_mean + sett_mean,
method = "nearest", data = dataset_nomiss)
# Assessing success of matching
summary(mod_match)
plot(mod_match)
# Get data frame containing only matched observations (distance is the propensity score)
dta_m <- match.data(mod_match)
dim(dta_m)
print(dta_m)
#uses library(gridExtra)
grid.arrange(
fn_bal(dta_m, "sett_mean"),
fn_bal(dta_m, "pop_mean"),
fn_bal(dta_m, "road_mean"),
#fn_bal(dta_m, "ag16_sum"), #+ theme(legend.position = "none"),
nrow = 3, widths = c(1, 0.8)
)
print(dta_m)
###Examining covariate balance###
#Difference-in-means
dta_m %>%
group_by(treated) %>%
select(one_of(dataset_cov)) %>%
summarise_all(funs(mean))
# More formally using t-test (ideally we should not be able to reject the null-hypothesis (being: there is no mean difference) for each covariate)
lapply(dataset_cov, function(v) {
t.test(dta_m[, v] ~ dta_m$treated)
})
# Estimating treatment effect (once we have a satisfactory match)
# three things needed:
#the mean difference, the standard deviation of each group, and the number of data values of each group.
#T-test
with(dta_m, t.test(change ~ treated))
#OLS - No covariates
lm_treat1 <- lm(change ~ treated, data = dta_m)
summary(lm_treat1)
#OLS - With covariates
lm_treat2 <- lm(change ~ treated + pop_mean + road_mean + sett_mean, data = dta_m) #sett_mean
summary(lm_treat2)