do_multboot: Multiplier bootstrap to construct uniform confidence bands...

In matteobonvini/sensitivitypuc: Sensitivity Analysis via the Proportion of Unmeasured Confounding (PUC)

Description

do_multboot is the main function to implement the multiplier bootstrap to estimate uniformly valid bands around the bounds.

Usage

do_multboot(n, ndelta, psihat, sigmahat, ifvals, alpha = 0.05, B = 10000)

Arguments

n	sample size.
ndelta	number of delta parameters used.
psihat	Estimate of either lower or upper bound curve minus (epsilon x an estimate of the appropriate quantile). It is a (num eps points) x (ndelta) matrix.
sigmahat	Estimate of the variance of psihat at each value of eps grid. It is a (num eps points) x (ndelta) matrix.
ifvals	n x (num eps points) x ndelta array. It represents the "influence function" values associated with psihat.
alpha	confidence level. Default is 0.05.
B	number of rademacher rvs sampled. Default is 10000.

Value

a ndelta-dimensional vector calpha equal to the z-score used to construct bands of the form psi(eps) \pm calpha * sigma(eps).

References

Kennedy, E. H. (2019). Nonparametric causal effects based on incremental propensity score interventions. Journal of the American Statistical Association, 114(526), 645-656.

matteobonvini/sensitivitypuc documentation built on Dec. 9, 2020, 2:24 a.m.