Simulation of complex survival distributions

摘要: This paper describes a method for generating sample survival distributions from hypothetical population, as would be required running Monte Carlo simulations. The is based on the concept of quincunx. Cases are entered into life table and allowed to drop out or die during each interval with probabilities that mirror population. By repeating this process many times tracking results, researcher able study sampling distribution effect size indices test statistics, can generate empirical estimates power precision planned studies. Unlike other methods commonly used purpose, model proposed here allows define population in which hazard rates and/or attrition vary substantially one time point next, may case clinical trials studies processing times. requires less than 100 lines code runs some 10,000 simulations per hour microcomputer.