normal_gps: Generalized Propensity Score for Continuous Treatment Domain

In williazo/gps.continuousnav:

Description

This function is used to fit a normal model for the conditional distribution of the treatment given covariates, and returns the resulting score function values for the observed treatment. After fitting the observed values the user can specify specific fixed treatment values to evaluate the conditional density at these points.

Usage

normal_gps(tx, covs, gps_val = NULL, interact_vars = NULL,
  polynomial_vars = NULL, polynomial_deg = NULL, variable_selection = F)

Arguments

tx	Vector with the continuous treatment value, used for finding the initial parameter MLE's
covs	Matrix of observed covariates
gps_val	Scalar value or vector which contains the values to find the estimated generalized propensity score
interact_vars	Specifies a character subset of the variables from the matrix covs and adds in pairwise interactions between all included covariates
polynomial_vars	Specifies a character subset of variables from the matrix covs to include as polynomial terms
polynomial_deg	Scalar value that identifies to what power the polynomial variable should be raised
variable_selection	Indicator whether to perform variable selection using AIC backwards selection

Details

Currently assumes normal density for the conditional distribution.

Value

Returns a list of objects.

williazo/gps.continuousnav documentation built on May 8, 2019, 6:57 p.m.