ABSTRACT
This paper develops methodology for statistical analysis with censored and missing data in medical researches. The author gives the examples, in which the appearance of missing or censored cases is not random (nonignorable censoring mechanisms), and how to resolve the problem.
Subject(s)
Biomedical Research/ethics , Data Interpretation, Statistical , Ethics, Medical , Animals , Biomedical Research/methods , HumansABSTRACT
The paper analyses the problem of censored samples in experimental medicine. Some cases in such samples have indefinite digital values. Censoring may result from death of some animals in the test group of the disease studied. When the animals that survived in the test group are compared to the control group in which all the animals survived, the difference between the groups is conventionally interpreted as the result of the disease pathogenesis, but as well it can be a result of censoring. The author gives the examples of censored samples and how to manage the problem.