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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
|
plot.matchit <- function(x, ...)
{
if (identical(eval(x$call$exact),TRUE)){
return(warning("Plot() not appropriate for exact matching. No plots generated"))
}
match.matrix <- x$match.matrix
xdata <- x$data
treata <- model.frame(x$formula,xdata)[,1,drop=FALSE]
pscore <- xdata[,"pscore"]
in.sample <- x$in.sample
covariates <- x$covariates
treat <- as.vector(treata[,1])
names(treat) <- row.names(treata)
covariates <- model.frame(delete.response(terms(x$formula)),xdata)[,,drop=FALSE]
mahvars <- eval(x$call$mahvars)
exact <- eval(x$call$exact)
if (!is.null(mahvars)){
w <- mahvars%in%names(covariates)
if(sum(w)!=length(mahvars)){
md <- as.data.frame(as.matrix(xdata[,mahvars[!w]]))
names(md) <- mahvars[!w]
covariates <- cbind(covariates,md)
}
}
if (!identical(TRUE, exact) & !identical(FALSE, exact)) {
w <- exact%in%names(covariates)
if(sum(w)!=length(exact)) {
ed <- as.data.frame(as.matrix(xdata[,exact[!w]]))
names(ed) <- exact[!w]
covariates <- cbind(covariates,ed)
}
}
psclass <- x$psclass
if(is.null(psclass)){subclass <- 0}else{subclass <- qbins <- max(na.omit(x$psclass))}
nearest <- !is.null(x$match.matrix)
q.cut <- x$q.cut
n <- length(treat)
n0 <- length(treat[treat==0])
n1 <- length(treat[treat==1])
if(is.null(names(treat))){names(treat) <- seq(1,n)}
labels <- names(treat)
clabels <- labels[treat==0]
tlabels <- labels[treat==1]
if(is.null(in.sample)){in.sample <- rep(TRUE,n); names(in.sample) <- names(treat)}
if(nearest==TRUE) {
match.matrix <- match.matrix[in.sample[treat==1],,drop=F]
num.matches <- dim(match.matrix)[2]-apply(as.matrix(match.matrix), 1, function(x) { sum(is.na(x)) })
names(num.matches) <- tlabels[in.sample[treat==1]]
t.units <- row.names(match.matrix)[num.matches>0]
c.units <- na.omit(as.vector(as.matrix(match.matrix)))
matched <-c(t.units,unique(c.units))
weights <- rep(0,length(treat))
names(weights) <- labels
weights[t.units] <- 1
for (cont in clabels) {
treats <- na.omit(row.names(match.matrix)[cont==match.matrix[,1]])
if (dim(match.matrix)[2]>1) {
for (j in 2:dim(match.matrix)[2])
treats <- c(na.omit(row.names(match.matrix)[cont==match.matrix[,j]]),treats)
}
if (length(treats)==0) weights[cont] <- 0
else for (k in 1:length(treats)) weights[cont] <- weights[cont]+1/num.matches[treats[k]]
}
} else {
matched <- names(treat)
weights <- rep(1,length(treat))
names(weights) <- names(treat)
}
weights[!in.sample] <- 0
doverlay <- function(x,treat,weights=NULL,xlab="",main="",lines=FALSE){
xmiss <- !is.na(x)
x <- x[xmiss]
treat <- treat[xmiss]
weights <- weights[xmiss]
minobs <- min(x)
maxobs <- max(x)
dx1 <- density(x[treat==1],from=minobs,to=maxobs)
dx0 <- density(x[treat==0],from=minobs,to=maxobs)
if(!is.null(weights))
{
x1 <- x[treat==1&weights!=0]
x0 <- sample(x[treat==0], size=min(10000,(100*length(x[treat==0]))),
replace=TRUE,prob=weights[treat==0]/sum(weights[treat==0]))
d1 <- density(x1,from=minobs,to=maxobs)
bw <- d1$bw
d0 <- density(x0,from=minobs,to=maxobs,bw=bw)
par(mfrow=c(2,1))
matplot(dx0$x,cbind(dx1$y,dx0$y),type="l",ylim=range(c(dx0$y,dx1$y,d1$y,d0$y)),
ylab="Density",xlab=xlab,main=paste(main,": All Units",sep=""))
legend(minobs,max(c(d1$y,d0$y,dx1$y,dx0$y)), lty=1:2, col=1:2,
legend=c("Treatment","Control"))
matplot(dx0$x,cbind(d1$y,d0$y),type="l",ylim=range(c(dx0$y,dx1$y,d1$y,d0$y)),
ylab="Density",xlab=xlab,main=paste(main,": Matched Units",sep=""))
legend(minobs,max(c(d1$y,d0$y,dx1$y,dx0$y)), lty=1:2, col=1:2,
legend=c("Treatment","Control"))
if (lines==T) abline(v=q.cut,col="grey",lty=1)
par(mfrow=c(1,1))
} else{
matplot(dx0$x,cbind(dx1$y,dx0$y),type="l",ylab="Density",xlab=xlab,main=main)
legend(minobs,max(c(dx1$y,dx0$y)),lty=1:2,col=1:2,legend=c("Treatment","Control"))
if (lines==T) abline(v=q.cut, col="grey", lty=1)
}
}
choice.menu <- function(choices,question)
{
k <- length(choices)-1
Choices <- data.frame(choices)
row.names(Choices) <- 0:k
names(Choices) <- "Choices"
print.data.frame(Choices,right=FALSE)
ans <- readline(question)
while(!ans%in%c(0:k))
{
print("Not valid -- please pick one of the choices")
print.data.frame(Choices,right=FALSE)
ans <- readline(question)
}
return(ans)
}
if(!is.null(pscore))
{
densq <- ("Would you like to see density estimates of the propensity scores?")
denschoice <- c("No","Yes")
densplot <- choice.menu(denschoice,densq)
if(densplot==1){
if(nearest){
doverlay(pscore,treat,weights,main="Propensity Score",lines=T)
#if(length(subclass)!=1 | subclass!=0){abline(v=q.cut,col="grey",lty=1)}
} else {
doverlay(pscore,treat,main="Propensity Score", lines=T)
#if(length(subclass)!=1 | subclass!=0){abline(v=q.cut,col="grey",lty=1)}
}
}
jitq <- ("Would you like to see a jitterplot of the propensity scores?")
jitchoice <- c("No","Yes")
jitplot <- choice.menu(jitchoice,jitq)
if(jitplot==1) {
jitp <- jitter(rep(1,length(treat)),factor=6)-(treat==0)
cwt <- sqrt(weights)
plot(pscore,xlim=range(na.omit(pscore)),ylim=c(-1,2),type="n",ylab="",xlab="Propensity Score",axes=F,main="Distribution of Propensity Scores")
if(length(subclass)!=1 | subclass!=0){abline(v=q.cut,col="grey",lty=1)}
points(pscore[treat==1&weights!=0],jitp[treat==1&weights!=0],pch=18,cex=cwt[treat==1&weights!=0])
points(pscore[treat==1&weights==0],jitp[treat==1&weights==0],pch=5,col="grey",cex=0.5)
points(pscore[treat==0&weights!=0],jitp[treat==0&weights!=0],pch=18,cex=cwt[treat==0&weights!=0])
points(pscore[treat==0&weights==0],jitp[treat==0&weights==0],pch=5,col="grey",cex=0.5)
axis(1)
text(sum(range(na.omit(pscore)))/2,1.5,"Treatment Units")
text(sum(range(na.omit(pscore)))/2,-0.5,"Control Units")
box()
print("To identify the units, use first mouse button; to stop, use second.")
identify(pscore,jitp,names(treat))
}
}
if(nearest){
choices <- c("No",paste("Yes : ",names(covariates)))
question <- "Would you like to see density estimates of any other covariates?"
ans <- -1
while(ans!=0)
{
ans <- as.numeric(choice.menu(choices,question))
if(ans!=0)
if(nearest)
{
doverlay(covariates[,(ans)],treat,weights,main=names(covariates)[ans])
} else {
doverlay(covariates[,(ans)],treat,main=names(covariates)[ans])
}
}
}
if(length(subclass)!=1 | subclass!=0){
choices <- c("No",paste("Yes : Subclass ", 1:qbins))
question <- "Would you like to see density estimates of any subclass covariates?"
ans <- -1
while(ans!=0)
{
ans <- as.numeric(choice.menu(choices,question))
if(ans!=0)
{
question2 <- "Which covariates?"
choices2 <- c(paste(names(covariates)))
ans2 <- as.numeric(choice.menu(choices2,question2))
ans2 <- ans2+1
if(sum(treat[psclass==ans]==1,na.rm=TRUE)<=2){
print("Not enough treatment units in this subclass")
} else if(sum(treat[psclass==ans]==0,na.rm=TRUE)<=2){
print("Not enough control units in this subclass")} else{
doverlay(covariates[,(ans2)][psclass==ans],treat[psclass==ans],main=paste("Subclass ",ans," : ",names(covariates)[ans2]))}
}
}
}
}
|
