salbm (version 1.0)

getEYnomo: Sensitivity Analysis for Binary Missing Data

Description

returns the difference in the expected value of the outcome, Y, when Y is not observed and the expected value Y when Y is observed as a percentage of the expected value of Y when observed.

Usage

getEYnomo( Y, BC, ratio=FALSE)

Arguments

Y

a vector of observed data including NAs.

BC

a vector of salbm estimates

ratio

if true return the ratio 100 * E[Y|non-obs] / E[Y|obs]; otherwise return 100 * ( E[Y|non-obs] - E[Y|obs] )/ E[y|obs]

Value

A vector containing 100 (E[ Y | Y not observed ] - E[ Y | Y is observed ]) / E[ Y | Y is observed ] or 100 E[ Y | Y not observed ] / E[ Y | Y is observed ]

Details

Typically, BC is a vector of salbm at different sensitivity parameters alpha. The function computes: 100 (E[ Y | Y not observed ] - E[ Y | Y is observed ]) / E[ Y | Y is observed ] when ratio is FALSE and computes 100 E[ Y | Y not observed ] / E[ Y | Y is observed ] when ratio is TRUE

Examples

# NOT RUN {
  # trt1 is one of two arms of a clinical trial with 
  # trt1[,K] being the measurments at the last timepoint
  # BC has two columns the first containing sensitivity
  # paramenters and the second their corrisponding salbm
  # estimates

  pdiff  <- getEYnomo( trt1[,K], BC[,2] )
  plot( x = BC[,1], y = pdiff )
# }