BIOMETRY RESEARCH GROUP

Mathematica Packages

Missing Binary Outcome with an Auxiliary Variable and All-or-None Compliance

(written by Stuart G. Baker)

These files are based on the manuscript by Baker SG. Analyzing a randomized cancer prevention trial with a missing binary outcome, an auxiliary variable, and all-or-none compliance. Journal of the American Statistical Association 2000;95:43-50.

It runs in Mathematica 3.0 and requires the files listed below.

modall.m	main program (which loads others)
modcorA.m	model A core program
modA.m	model A
modcorAM.m	model AM core program
modcorC.m	model C core program
modC.m	model C
modcorAC.m	model AC core program
modAC.m	model AC
modvar.m	functions for some variance calculations (MP transformation)
modsim.m	functions for simulations

To reproduce the calculations in the manuscript load modall.m and use the functions:

modA[{1.2,1}, n, "NUL"], modA[{1.2,1.1}, n, "ALT"],
modC[{1.2}, n, "NUL"], modC[{1.2}, n, "ALT"],
modAC[{1.2,1}, n, "NUL"], modAC[{1.2,1.1}, n, "ALT"].

The parameter n is the number of simulations where n=0 gives an analytic approximation only. The value 1.2 is the ratio of the probability of missing given finasteride to the probability of missing given placebo. The values 1 and 1.1 are the ratios of true positive rates in the finasteride group to those in the placebo group.