Learn R Programming

ImpactIV (version 1.0)

heter_IV2: Estimation causal effect under Assumption 7 in Ding et al. (2011)

Description

Estimation causal effect under Assumption 6 in Ding et al. (2011) when the second order moment of the error term is not constant.

Usage

heter_IV2(Z, A, M, Y, X, polydegree = 2, step1 = NULL, truncate = 0.25, select = NULL)

Arguments

Z

A vector of the randomization variable.

A

A vector of the first mediator: whether a patient receives antidepressant medication.

M

A vector of the second mediator: whether a patient receives mental health therapy.

Y

A vector of the outcome of interest.

X

A matrix of all the covariates.

polydegree

The order of the polynomial function.

step1

The result of the first step estimation from homo_IV1.

truncate

Truncate the estimated Omega using a positive constant.

select

Using AIC or BIC for variable selection in the polynomial regression, the default is null.

Value

beta
beta coefficients of Z, A, M and AM.
phat
proportion of randomization to the treatment group.
residual
residuals of the first step regression.
se
standard errors of beta coefficients.
zvalue
z-vlues of the beta coefficients.
pvalue
p-values of the beta coefficients.
CI
confidence intervals of the beta coefficients.
COV
covariance matrix of the beta coefficients.
ser
robust version of standard errors of beta coefficients.
zvaluer
robust version of z-vlues of the beta coefficients.
pvaluer
robust version of p-values of the beta coefficients.
CIr
robust version of confidence intervals of the beta coefficients.
COVr
robust version of covariance matrix of the beta coefficients.
N
sample size
G
G is defined in Ding et al. (2010).
W
W is defined in Ding et al. (2010).
Omegahat
Omegahat is is defined in Ding et al. (2010).

Details

For background of the problem, see Ding et al. (2011).

References

Ding, P., Geng, Z. and Zhou, X. H. (2011). Identifying Causal Effect for Multi-Component Intervention Using Instrumental Variable Method: with A Case Study of IMPACT Data. Technical Report.

See Also

Examples

Run this code
##See help for "ImpactIV"

Run the code above in your browser using DataLab