RDSmoothnessBound: Lower bound on smoothness constant M in RD designs

Description Usage Arguments Value

View source: R/Cbound.R

Description

Estimate a lower bound on smoothness constant M and provide a lower confidence interval.

Usage

1
2
RDSmoothnessBound(d, s, separate = TRUE, multiple = TRUE, alpha = 0.05,
  sclass = "T")

Arguments

d

RDData object

s

Number of support points that curvature estimates should average over

separate

If TRUE, report estimates separately for data above and below cutoff. If FALSE, report pooled estimates

multiple

If TRUE, use multiple curvature estimates. If FALSE, use a single estimate using only observations closest to the cutoff.

alpha

determines confidence level 1-alpha.

sclass

Smoothness class, either "T" for Taylor or "H" for Hölder class.

Value

Returns a list with the following elements

mu+,mu-

Lower bound of CI for observations above and below cutoff

Z+,Z-

Point estimate used for lower bound

sd+,sd-

Standard deviations of point estimates


kolesarm/RDHonest documentation built on April 3, 2018, 11:08 a.m.