cem (version 1.1.31)

shift.cem: Diagnostic tool for CEM

Description

Diagnostic tools for CEM. Applies leftward and rightward shifts of the cutpoints.

Usage

shift.cem(obj, data, shifts=NULL, verbose=1, plot=TRUE)

Value

tab

an invisible object containing a new cem object

Arguments

obj

and object of class cem

data

the original data

shifts

a vector of proportions of shifts

verbose

controls the level of verbosity

plot

whether to plot a graphic representation of the search

Author

Stefano Iacus, Gary King, and Giuseppe Porro

Details

For each variable, shift all the cutpoints left and right by shifts times the smallest epsilon of the coarsening. Shifting to the right produces a new cell on the left; shift to the left, adds a new cell to the coarsening on the right. Only positive proportions should be used; the algorithm will produce shifting on the left or on the right. The best shifting of the original cem match is produced as output, where best is defined in terms of the maximal total number of matched units mT+mC (see below).

By default, the function returns minimal information about the execution of the algorithm. By setting a value greater than 0 in option verbose more feedback on the process is returned.

Option plot = TRUE plots the number of treated units matched mT, the number of control units matched mC, and the sum mT+mC, as a function of the shifts.

References

Iacus, King, Porro (2011) tools:::Rd_expr_doi("10.1198/jasa.2011.tm09599")

Iacus, King, Porro (2012) tools:::Rd_expr_doi("10.1093/pan/mpr013")

Iacus, King, Porro (2019) tools:::Rd_expr_doi("10.1017/pan.2018.29")

See Also

cem

Examples

# \donttest{
data(LL)
set.seed(123)
m74 <- max(LL$re74, na.rm=TRUE)
s74 <- seq(0,m74,by=sd(LL$re74))
l74 <- length(s74)
if(max(s74) < m74) s74 <- c(s74, m74)

m75 <- max(LL$re75, na.rm=TRUE)
s75 <- seq(0,m75,by=sd(LL$re75))
l75 <- length(s75)
if(max(s75) < m75) s75 <- c(s75, m75)

mybr = list(re74=s74, 
 re75 = s75,
 age = hist(LL$age,plot=FALSE)$breaks,
 education = hist(LL$education,plot=FALSE)$breaks)

mat <- cem(treatment="treated",data=LL, drop="re78",cut=mybr)
mat

shift.cem(mat, data=LL, shifts=seq(0.01, 0.5, length=10), verb=1)
# }