R/kmlate.R
In pedrohcgs/kmte: An R Package for Treatment Effects with Censored Outcomes
Defines functions
kmlate
Documented in
kmlate
#' Kaplan-Meier Local Average Treatment Effect
#'
#' \emph{kmlate} computes the Local Average Treatment Effect for possibly right-censored outcomes.
#' The estimator relies on the availability of an Instrumental variable Z, and on a monotonicity assumption.
#' To implement the estimator, we make use of an instrumental propensity score approach.
#' For details of the estimation procedure, see Sant'Anna (2016a), 'Program Evaluation with
#' Right-Censored Data'.
#'
#'
#'@param out vector containing the outcome of interest
#'@param delta vector containing the censoring indicator (1 if observed, 0 if censored)
#'@param treat vector containing the treatment indicator (1 if treated, 0 if control)
#'@param z vector containing the binary instrument
#'@param xpscore matrix (or data frame) containing the covariates (and their
#' transformations) to be included in the instrument propensity score estimation.
#' Instrument Propensity score estimation is based on Logit.
#'@param b The number of bootstrap replicates to be performed. Default is 1,000.
#'@param ci A scalar or vector with values in (0,1) containing the confidence level(s)
#' of the required interval(s). Default is a vector with
#' 0,90, 0.95 and 0.99
#'@param trunc scalar that defined the truncation parameter. Default is NULL, which does not perform any kind of
#' truncation in the computation of the ATE. When trunc is different than NULL, all outcomes which values greater
#' than trunc are truncated.
#'@param cores number of processesors to be used during the bootstrap (default is 1).
#' If cores>1, the bootstrap is conducted using snow
#'
#'@return a list containing the local average treatment effect estimate, late,
#' and the bootstrapped \emph{ci} confidence
#' confidence interval, late.lb (lower bound), and late.ub (upper bound).
#'@export
#'@importFrom stats glm
#'@importFrom parallel makeCluster stopCluster clusterExport
#'@importFrom boot boot.ci boot
#-----------------------------------------------------------------------------
kmlate <- function(out, delta, treat, z, xpscore, b = 1000, ci = c(0.90,0.95,0.99),
trunc = NULL, cores = 1) {
#-----------------------------------------------------------------------------
# first, we merge all the data into a single datafile
fulldata <- data.frame(cbind(out, delta, treat, z, xpscore))
#-----------------------------------------------------------------------------
# Next, we set up the bootstrap function
boot1.kmlate <- function(fulldata, i, trunc1 = trunc){
#----------------------------------------------------------------------------
# Select the data for the bootstrap (like the original data)
df.b=fulldata[i,]
#----------------------------------------------------------------------------
# Dimension of data matrix df.b
dim.b <- dim(df.b)[2]
# Next, we rename the variable in xpscore to avoid problems
xpscore1.b <- df.b[, (5:dim.b)]
datascore.b <- data.frame(y = df.b[, 4], xpscore1.b)
#-----------------------------------------------------------------------------
# estimate the propensity score
pscore.b <- stats::glm(y ~ ., data = datascore.b,
family = binomial("logit"))
df.b$pscore <- pscore.b$fit
#-----------------------------------------------------------------------------
# sample size
n.total.b <- as.numeric(length(df.b[, 1]))
# subset of treated individuals with instrument Z equal to 1
data.treat.z1.b <- subset(df.b, df.b[, 3] == 1 & df.b[, 4] == 1)
# subset of treated individuals with instrument Z equal to 0
data.treat.z0.b <- subset(df.b, df.b[, 3] == 1 & df.b[, 4] == 0)
# subset of control individuals with instrument Z equal to 1
data.control.z1.b <- subset(df.b, df.b[, 3] == 0 & df.b[, 4] == 1)
# subset of control individuals with instrument Z equal to 0
data.control.z0.b <- subset(df.b, df.b[, 3] == 0 & df.b[, 4] == 0)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for treated with z=1
data.treat.z1.b <- kmweight(1, 2, data.treat.z1.b)
n.treat.z1.b <- as.numeric(length(data.treat.z1.b[, 1]))
data.treat.z1.b$w <- data.treat.z1.b$w * (n.treat.z1.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for treated with z=0
data.treat.z0.b <- kmweight(1, 2, data.treat.z0.b)
n.treat.z0.b <- as.numeric(length(data.treat.z0.b[, 1]))
data.treat.z0.b$w <- data.treat.z0.b$w * (n.treat.z0.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for control with z=1
data.control.z1.b <- kmweight(1, 2, data.control.z1.b)
n.control.z1.b <- as.numeric(length(data.control.z1.b[, 1]))
data.control.z1.b$w <- data.control.z1.b$w * (n.control.z1.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for control with z=0
data.control.z0.b <- kmweight(1, 2, data.control.z0.b)
n.control.z0.b <- as.numeric(length(data.control.z0.b[, 1]))
data.control.z0.b$w <- data.control.z0.b$w * (n.control.z0.b/n.total.b)
#-----------------------------------------------------------------------------
# Let's put everything in a single data
df.b <- data.frame(rbind(data.treat.z1.b, data.treat.z0.b,
data.control.z1.b, data.control.z0.b))
#-----------------------------------------------------------------------------
# Compute weigths for treatment and control groups
w11km.b <- ((df.b$treat * df.b$z * df.b$w) / df.b$pscore)
w10km.b <- ((df.b$treat * (1 - df.b$z) * df.b$w) / (1 - df.b$pscore))
w01km.b <- ((1 - df.b$treat) * df.b$z * df.b$w / df.b$pscore)
w00km.b <- ((1 - df.b$treat) * (1 - df.b$z) * df.b$w / (1 - df.b$pscore))
kappa11 <- mean(df.b$treat * df.b$z / df.b$pscore)
kappa10 <- mean(df.b$treat * (1-df.b$z) / (1 - df.b$pscore))
kappa01 <- mean((1 - df.b$treat) * df.b$z / df.b$pscore)
kappa00 <- mean((1 - df.b$treat) * (1-df.b$z) / (1 - df.b$pscore))
w1km.b <- w11km.b - w10km.b
w0km.b <- w01km.b - w00km.b
kappa1 <- kappa11 - kappa10
kappa0 <- kappa01 - kappa00
w1km.b <- w1km.b / kappa1
w0km.b <- w0km.b / kappa0
#-----------------------------------------------------------------------------
# Compute Counterfactual Local Average Outcomes, E[Y(1)|C] and E[Y(0)|C], and the LATE
meany1km.c <- sum(w1km.b * df.b$out)
meany0km.c <- sum(w0km.b * df.b$out)
if (is.null(trunc1) == FALSE){
meany1km.c <- sum(w1km.b * df.b$out * (df.b$out <= trunc1))
meany0km.c <- sum(w0km.b * df.b$out * (df.b$out <= trunc1))
}
late <- meany1km.c - meany0km.c
#-----------------------------------------------------------------------------
return(cbind(meany1km.c, meany0km.c, late))
}
#-----------------------------------------------------------------------------
# Number of bootstrap draws
nboot <- b
#----------------------------------------------------------------------------
#COmput the bootstrap
if (cores == 1){
boot.kmlate <- boot::boot(fulldata, boot1.kmlate, R = nboot)
}
if (cores > 1){
cl <- parallel::makeCluster(cores)
#clusterExport(cl, "kmweight")
parallel::clusterSetRNGStream(cl)
boot.kmlate <- boot::boot(fulldata, boot1.kmlate, R = nboot, parallel = "snow", ncpus = cores)
parallel::stopCluster(cl)
}
#----------------------------------------------------------------------------
# Compute Counterfactual Average Outcomes, E[Y(1)|C] and E[Y(0)|C], and the LATE
meany1km.c <- boot.ci(boot.kmlate, type="perc", index=1)$t0
names(meany1km.c) <- "E[Y(1)|Complier]"
meany0km.c <- boot.ci(boot.kmlate, type="perc", index=2)$t0
names(meany0km.c) <- "E[Y(0)|Complier]"
late <- boot.ci(boot.kmlate, type="perc", index=3)$t0
names(late) <- "LATE"
#----------------------------------------------------------------------------
#Compute the confidence interval for ate
if (length(ci) == 1){
late.lb <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[4]
late.ub <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[5]
}
if (length(ci) >1){
late.lb <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[,4]
late.ub <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[,5]
}
late.lb <- matrix(late.lb,length(ci),1)
late.ub <- matrix(late.ub,length(ci),1)
rownames(late.ub) <- paste(names(quantile(1, probs = ci)), 'CI: UB')
rownames(late.lb) <- paste(names(quantile(1, probs = ci)), 'CI: LB')
colnames(late.ub) <- "LATE"
colnames(late.lb) <- "LATE"
#----------------------------------------------------------------------------
# Return these
list(late = late,
meany1.c = meany1km.c,
meany0.c = meany0km.c,
#boot = boot.kmlate,
late.lb = late.lb,
late.ub = late.ub)
}
pedrohcgs/kmte documentation built on May 24, 2019, 11:46 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions kmlate
Documented in kmlate
#' Kaplan-Meier Local Average Treatment Effect
#'
#' \emph{kmlate} computes the Local Average Treatment Effect for possibly right-censored outcomes.
#' The estimator relies on the availability of an Instrumental variable Z, and on a monotonicity assumption.
#' To implement the estimator, we make use of an instrumental propensity score approach.
#' For details of the estimation procedure, see Sant'Anna (2016a), 'Program Evaluation with
#' Right-Censored Data'.
#'
#'
#'@param out vector containing the outcome of interest
#'@param delta vector containing the censoring indicator (1 if observed, 0 if censored)
#'@param treat vector containing the treatment indicator (1 if treated, 0 if control)
#'@param z vector containing the binary instrument
#'@param xpscore matrix (or data frame) containing the covariates (and their
#' transformations) to be included in the instrument propensity score estimation.
#' Instrument Propensity score estimation is based on Logit.
#'@param b The number of bootstrap replicates to be performed. Default is 1,000.
#'@param ci A scalar or vector with values in (0,1) containing the confidence level(s)
#' of the required interval(s). Default is a vector with
#' 0,90, 0.95 and 0.99
#'@param trunc scalar that defined the truncation parameter. Default is NULL, which does not perform any kind of
#' truncation in the computation of the ATE. When trunc is different than NULL, all outcomes which values greater
#' than trunc are truncated.
#'@param cores number of processesors to be used during the bootstrap (default is 1).
#' If cores>1, the bootstrap is conducted using snow
#'
#'@return a list containing the local average treatment effect estimate, late,
#' and the bootstrapped \emph{ci} confidence
#' confidence interval, late.lb (lower bound), and late.ub (upper bound).
#'@export
#'@importFrom stats glm
#'@importFrom parallel makeCluster stopCluster clusterExport
#'@importFrom boot boot.ci boot
#-----------------------------------------------------------------------------
kmlate <- function(out, delta, treat, z, xpscore, b = 1000, ci = c(0.90,0.95,0.99),
trunc = NULL, cores = 1) {
#-----------------------------------------------------------------------------
# first, we merge all the data into a single datafile
fulldata <- data.frame(cbind(out, delta, treat, z, xpscore))
#-----------------------------------------------------------------------------
# Next, we set up the bootstrap function
boot1.kmlate <- function(fulldata, i, trunc1 = trunc){
#----------------------------------------------------------------------------
# Select the data for the bootstrap (like the original data)
df.b=fulldata[i,]
#----------------------------------------------------------------------------
# Dimension of data matrix df.b
dim.b <- dim(df.b)[2]
# Next, we rename the variable in xpscore to avoid problems
xpscore1.b <- df.b[, (5:dim.b)]
datascore.b <- data.frame(y = df.b[, 4], xpscore1.b)
#-----------------------------------------------------------------------------
# estimate the propensity score
pscore.b <- stats::glm(y ~ ., data = datascore.b,
family = binomial("logit"))
df.b$pscore <- pscore.b$fit
#-----------------------------------------------------------------------------
# sample size
n.total.b <- as.numeric(length(df.b[, 1]))
# subset of treated individuals with instrument Z equal to 1
data.treat.z1.b <- subset(df.b, df.b[, 3] == 1 & df.b[, 4] == 1)
# subset of treated individuals with instrument Z equal to 0
data.treat.z0.b <- subset(df.b, df.b[, 3] == 1 & df.b[, 4] == 0)
# subset of control individuals with instrument Z equal to 1
data.control.z1.b <- subset(df.b, df.b[, 3] == 0 & df.b[, 4] == 1)
# subset of control individuals with instrument Z equal to 0
data.control.z0.b <- subset(df.b, df.b[, 3] == 0 & df.b[, 4] == 0)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for treated with z=1
data.treat.z1.b <- kmweight(1, 2, data.treat.z1.b)
n.treat.z1.b <- as.numeric(length(data.treat.z1.b[, 1]))
data.treat.z1.b$w <- data.treat.z1.b$w * (n.treat.z1.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for treated with z=0
data.treat.z0.b <- kmweight(1, 2, data.treat.z0.b)
n.treat.z0.b <- as.numeric(length(data.treat.z0.b[, 1]))
data.treat.z0.b$w <- data.treat.z0.b$w * (n.treat.z0.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for control with z=1
data.control.z1.b <- kmweight(1, 2, data.control.z1.b)
n.control.z1.b <- as.numeric(length(data.control.z1.b[, 1]))
data.control.z1.b$w <- data.control.z1.b$w * (n.control.z1.b/n.total.b)
#-----------------------------------------------------------------------------
# Compute Kaplan-Meier weigth for control with z=0
data.control.z0.b <- kmweight(1, 2, data.control.z0.b)
n.control.z0.b <- as.numeric(length(data.control.z0.b[, 1]))
data.control.z0.b$w <- data.control.z0.b$w * (n.control.z0.b/n.total.b)
#-----------------------------------------------------------------------------
# Let's put everything in a single data
df.b <- data.frame(rbind(data.treat.z1.b, data.treat.z0.b,
data.control.z1.b, data.control.z0.b))
#-----------------------------------------------------------------------------
# Compute weigths for treatment and control groups
w11km.b <- ((df.b$treat * df.b$z * df.b$w) / df.b$pscore)
w10km.b <- ((df.b$treat * (1 - df.b$z) * df.b$w) / (1 - df.b$pscore))
w01km.b <- ((1 - df.b$treat) * df.b$z * df.b$w / df.b$pscore)
w00km.b <- ((1 - df.b$treat) * (1 - df.b$z) * df.b$w / (1 - df.b$pscore))
kappa11 <- mean(df.b$treat * df.b$z / df.b$pscore)
kappa10 <- mean(df.b$treat * (1-df.b$z) / (1 - df.b$pscore))
kappa01 <- mean((1 - df.b$treat) * df.b$z / df.b$pscore)
kappa00 <- mean((1 - df.b$treat) * (1-df.b$z) / (1 - df.b$pscore))
w1km.b <- w11km.b - w10km.b
w0km.b <- w01km.b - w00km.b
kappa1 <- kappa11 - kappa10
kappa0 <- kappa01 - kappa00
w1km.b <- w1km.b / kappa1
w0km.b <- w0km.b / kappa0
#-----------------------------------------------------------------------------
# Compute Counterfactual Local Average Outcomes, E[Y(1)|C] and E[Y(0)|C], and the LATE
meany1km.c <- sum(w1km.b * df.b$out)
meany0km.c <- sum(w0km.b * df.b$out)
if (is.null(trunc1) == FALSE){
meany1km.c <- sum(w1km.b * df.b$out * (df.b$out <= trunc1))
meany0km.c <- sum(w0km.b * df.b$out * (df.b$out <= trunc1))
}
late <- meany1km.c - meany0km.c
#-----------------------------------------------------------------------------
return(cbind(meany1km.c, meany0km.c, late))
}
#-----------------------------------------------------------------------------
# Number of bootstrap draws
nboot <- b
#----------------------------------------------------------------------------
#COmput the bootstrap
if (cores == 1){
boot.kmlate <- boot::boot(fulldata, boot1.kmlate, R = nboot)
}
if (cores > 1){
cl <- parallel::makeCluster(cores)
#clusterExport(cl, "kmweight")
parallel::clusterSetRNGStream(cl)
boot.kmlate <- boot::boot(fulldata, boot1.kmlate, R = nboot, parallel = "snow", ncpus = cores)
parallel::stopCluster(cl)
}
#----------------------------------------------------------------------------
# Compute Counterfactual Average Outcomes, E[Y(1)|C] and E[Y(0)|C], and the LATE
meany1km.c <- boot.ci(boot.kmlate, type="perc", index=1)$t0
names(meany1km.c) <- "E[Y(1)|Complier]"
meany0km.c <- boot.ci(boot.kmlate, type="perc", index=2)$t0
names(meany0km.c) <- "E[Y(0)|Complier]"
late <- boot.ci(boot.kmlate, type="perc", index=3)$t0
names(late) <- "LATE"
#----------------------------------------------------------------------------
#Compute the confidence interval for ate
if (length(ci) == 1){
late.lb <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[4]
late.ub <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[5]
}
if (length(ci) >1){
late.lb <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[,4]
late.ub <- boot.ci(boot.kmlate, type="perc", index=3, conf = ci)$percent[,5]
}
late.lb <- matrix(late.lb,length(ci),1)
late.ub <- matrix(late.ub,length(ci),1)
rownames(late.ub) <- paste(names(quantile(1, probs = ci)), 'CI: UB')
rownames(late.lb) <- paste(names(quantile(1, probs = ci)), 'CI: LB')
colnames(late.ub) <- "LATE"
colnames(late.lb) <- "LATE"
#----------------------------------------------------------------------------
# Return these
list(late = late,
meany1.c = meany1km.c,
meany0.c = meany0km.c,
#boot = boot.kmlate,
late.lb = late.lb,
late.ub = late.ub)
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.