salbm (version 1.0)

getXCI: Sensitivity Analysis for Binary Missing Data

Description

Given an object returned by salbmCombine the getXCI function extracts difference estimates and confidence intervals by pairs of sensitivity paramaters.

Usage

getXCI( Res, ci=c("lb4","ub4") )

Arguments

Res

an object returned by salbmCombine with confidence intervals

ci

indicates which confidence interval to return. See the help for salbmCombine for the list of confidence interval types.

Value

returns a list with 1. K the value of the last time-point 2. alphas the vector of sensitivity parameters 3. ER1 an n^2 by 5 matrix where n is the length of the alphas vector. ER1 contains expected values of Y at time K for the first treatment group trt1. The columns are alpha1, alpha2, E[Y_K| alpha], lb, ub where lb and ub are the lower bound and upper bound of the confidence interval chosen, alpha1 is the sensitivity parameter associated with treatment group 1 and alpha2 the sensitivity parameter associated with treatment group 2. 4. ER2 as ER1 but for the second treatment arm trt2. 5. ERD the difference in estimates trt2 - trt1. 6. SR1 cumulative sum estimates for trt 1 7. SR2 cumulative sum estimates for trt 2 8. SRD differences in cumulative sum estimates trt 2 - trt 1.

Details

Looks for ECI and SCI in the input object. For each pair of sensitivity parameters ECI contains confidence bands for estimates of E[ Yt | alpha ] and their differences. SCI contains confidence bands for estimates of the cumulative sum E[ Y1+Y2+...+YK | alpha ] and their differences.

Examples

# NOT RUN {
  fplots <- getXCI( Res=Results, lb=c("lb2","ub2") )
  alphas <- fplots$alphas
  ERD    <- fplots$ERD

  ## significant points
  Epts   <-  ERD[ sign(ERD[,4]) == sign(ERD[,5]), ]

  filled.contour( x = alphas, y = alphas,
     z              = matrix(ERD[,3],length(alphas),length(alphas), byrow=TRUE),
     xlab           = expression(paste(alpha, " (Treatment 1)")),
     ylab           = expression(paste(alpha, " (Treatment 2)")),
     nlevels        = 8,
     color.palette  = colorRampPalette(c( "#993404","#D95F0E","#FE9929","#FFD9BE","#FFFFD4"),
                                       space="rgb"),
     plot.axis      = ( points( Epts[,c(1,2)], pch=15, cex=0.8, col = "#777799FF" ) )
  ) 
# }