pscoreSelect {cem}

Heuristic search of the best propensity score model specification

Description

Heuristic search of the best propensity score model specification

Usage

pscoreSelect(formula, data, C.L=2*(pnorm(-1,0,1)), C.Q=0.1,verbose=1) 

Arguments

data

the original data.

formula

formula type specification.

C.L

if at leat one likelihood ration test statistic is greater than C.L, then the covariate with higher likelihood ration test statistic is added linearly to the model.

C.Q

if the highest likelihood ratio statistic is greater than C.Q, then interaction terms are included in the pscore model specification.

verbose

integer, prints more info on screen.

Details

This is a tool to help the user to the search for the best propensity score model specification along the lines suggested by Imbens and Rubin (forthcoming). The output of the function is a model formula to be passed to glm or such, in order to estimate the propensity score model and then perform propensity score matching.

This tool is useful in combination with imbspace.plot.

Value

val

an invisible object of class list.

Author(s)

Richard Nielsen

References

Iacus, King, Porro (2011) doi:10.1198/jasa.2011.tm09599

Iacus, King, Porro (2012) doi:10.1093/pan/mpr013

Iacus, King, Porro (2019) doi:10.1017/pan.2018.29

Imbens, Rubin (2015) doi:10.1017/CBO9781139025751

See Also

cemspace

Examples

set.seed(123)
data(LL)
mod <- pscoreSelect( treated ~ age + education + black+ married + nodegree + 
 re74 + re75 + hispanic + u74 + u75, data=LL)
print(mod)