Plot of imbalance space diagnostic tool for CEM
Plot of imbalance space diagnostic tool for CEM
imbspace.plot(obj, group="1", data, explore=TRUE, verbose=1)
obj |
an object of class |
group |
character string denoting group id. Defaults to |
data |
data for running additional matching solutions. |
explore |
if |
verbose |
integer, controls output. |
For an interactive device a two panels plot is given. On the left panel the user can select a CEM solution and the number of cutpoints used in that matching solution is plotted as a parallel plot on the right plot. On exit (right-click on the left panel), the function returns all the cem solutions highlighted in the last selection of the user.
For non-interactive devices, only the space of the solutions are plotted.
This plot shows the tradeoff in matching as a function of imbalance and sample size.
The imbalance of the raw data is represented as a red plot and the initial CEM solution as a green plot. All solutions below the green dot and left to it are better than the user choice in terms of imbalance and number of units matched.
tab |
an invisible object containing the selection of cem solutions and their coarsenings. |
Stefano Iacus, Gary King, and Giuseppe Porro
Iacus, King, Porro (2011) https://doi.org/10.1198/jasa.2011.tm09599
Iacus, King, Porro (2012) https://doi.org/10.1093/pan/mpr013
Iacus, King, Porro (2019) https://doi.org/10.1017/pan.2018.29
require(cem)
data(LL)
set.seed(123)
mat <- cem("treated", LL, drop=c("re78","treated"),
cut=list(age=4, edu=4, re74=3, re75=3), keep.all=TRUE)
mat
imb.raw <- L1.profile(LL$treated, LL[, mat$vars], M=250, plot=FALSE)
imbsp <- imbspace(mat, LL,depth=2, raw.profile=imb.raw,
maximal=6, minimal=2, fixed=c("hispanic", "black", "married",
"nodegree","u74","u75"), plot=FALSE)
tmp <- plot(imbsp,data=LL,explore=interactive())
tmpPlease choose more modern alternatives, such as Google Chrome or Mozilla Firefox.