rbounds (version 2.1)

iv_sens: Function to calculate Rosenbaum bounds for IV Estimator based on Wilcoxon sign rank test.

Description

iv_sens performs a non-parametric, instrumental variable sensitivity analysis on matched pairs following the logic of the Neyman-Rubin framework for causal inference. The function supports a variable-valued instrument.

Usage

iv_sens(Rt, Rc, Dt, Dc, Gamma = 6, GammaInc = 1)

Arguments

Rt,Rc

Vectors of observed response outcomes for matched treatment and control observations, respectively.

Dt,Dc

Vectors of observed doses for matched observations, respectively. This is level of dose encouraged by the instrument.

Gamma

Upper-bound on gamma parameter.

GammaInc

To set user specified increments for gamma parameter.

Value

rbounds.

Details

Given matched pairs of observations on an instrument Z, which encourages dose D, this function performs a Rosenbaum's bounds sensitivity analysis. Note that matching is done on levels of the instrument. See example below.

References

Angrist, Joshua D., Imbens, Guido W., and Rubin, Donald B. (1996). "Identification of Causal Effects Using Instrumental Variables." Journal of the American Statistical Association 91/434, pp. 444--455.

Rosenbaum, Paul R. (1996). "Comment." Journal of the American Statistical Association 91/434, pp. 465--468.

Rosenbaum, Paul R. (2002). Observational Studies. Springer-Verlag.

Rosenbaum, Paul R. (2010). Design of Observational Studies. Springer-Verlag.

See Also

See also data.prep, binarysens, hlsens, Match, mcontrol

Examples

## Example from Rosenbaum (2010, ch. 5).

data(AngristLavy)

#Match on Economic Status Across Levels of the Instrument
rr <- Match(Y=AngristLavy$avgmath, Tr=AngristLavy$z, X=AngristLavy$pct_disadv, 
estimand ="ATC", M=2,  replace=FALSE)

#Extract Matched Outome Data
ctrl <- AngristLavy$avgmath[rr$index.control]
trt <- AngristLavy$avgmath[rr$index.treated]

#Extract Matched Doses
#Doses Encouraged By Instrument - Here Class Size
csize.trt <- AngristLavy$classize[rr$index.treated]
csize.ctrl <- AngristLavy$classize[rr$index.control]

#Run Sensitivity Analsyis
iv_sens(trt, ctrl, csize.trt, csize.ctrl, Gamma=2, GammaInc=.1)