Software : Validation


RCTdesign (and S+SeqTrial before it) have been extensively validated. Below you will find links to a general description of the ways that the code has been validated, as well as links to some of the R script files used to verify the accuracy and precision of the RCTdesign routines. These script files are provided in order that can be examined and extended by users who are charged with validation of their programs related to regulatory submissions.

  1. Overview of Validation Methods and selected results. This Sweave document sources the R script files given below.   

  2. RCTvalidate.R: something of a master validation R script file that describes and sources other scripts.

  3. RCTvalidateExpMeanJ2.R: compares numerically integrated expected MLE to theoretical results.

  4. RCTvalidateCriticalValues.R: compares RCTdesign results to critical values previously reported in the statistical literature.

  5. RCTvalidateGSDesign.R: compares RCTdesign results to those calculated by gsDesign(), an R package written independently by Keaven Anderson.

  6. RCTvalidateRSeq.R: compares vectorized computation of test statistics in RCTdesign simulation routines to the corresponding computations in the R base package.

  7. RCTvalidateOCSimulate.R: compares numerically integrated operating characteristics (stopping probabilities, power, ASN) to simulated results.

  8. RCTvalidateInference.R: verifies internal consistency of inference returned by RCTdesign with operating characteristics, as well as using simulations to verify coverage probabilities of CI and bias of estimators. 

  9. RCTvalidateErrSpendExmpls.R: compares results from current version of RCTdesign to any results stored from previous RCTdesign versions.