ncor_ind: A Negative Control Outcome Regression for Eliminating...

In yuyy-shandong/NCOR:

Description

A Negative Control Outcome Regression for Eliminating Unobserved Confounding in Time-series Studies

Usage

ncor_ind(
  data = data,
  pre_outc_name = "y1",
  expo_namem = "x",
  post_outc_name = "y3",
  centre = "centre",
  method = c("regression", "IVW"),
  boot_no = NULL
)

Arguments

data	an optional data frame containing the variables in the model.
pre_outc_name	the name of pre-exposure outcome
expo_namem	the name of exposure
post_outc_name	the name of post-exposure outcome
centre	the number of reacher centre or time series fragments
method	method of estimation
boots_no	the number of bootstrap for IVW estimation

Value

causal the casual effect estimation

lag the lag causal effect estimation

Examples

NN <- 10
data_total <- NULL
for(i in 1:NN){
c0 <- 0.5
dia <- 0.5
diff_c1 <- 0
r <- 0.5
c1 <- rnorm(1,0.5,1)
c2 <- rnorm(1,0.5,1)
N <- 100
Sigma <- matrix(c(1,r,r^2,r,1,r,r^2,r,1),3,3)
u <- MASS::mvrnorm(n=N, mu=rep(0,3), Sigma=Sigma)
colnames(u) <- c('u1','u2','u3')
u <- data.frame(u)
u1 <- u$u1
u2 <- u$u2
u3 <- u$u3

eps_y1 <- rnorm(N,0,0.1)
eps_y3 <- rnorm(N,0,0.1)
eps_x2 <-  rnorm(N,0,0.1)
y1 <- c1*u1 + eps_y1
x2 <- c2*u2 + eps_x2
y3 <- c0*x2 + c1*u3+diff_c1*u3 + dia*y1 + eps_y3
data_cenre <- data.frame(x2,u1,u2,u3,y1,y3,i)
data_total <- rbind(data_total,data_cenre)

}

model <- ncor_ind (data=data_total, pre_outc_name='y1',expo_namem = 'x2',
                  post_outc_name='y3',centre='i',method='IVW',boot_no=1000)

model

yuyy-shandong/NCOR documentation built on April 26, 2020, 12:44 a.m.