Statistical methods rely on valid data. These data either already
exist, or can at least be generated with defined effort and/or budget.
Reliability and safety analyses however deal with poor data, and -even
with budget- there is almost never a chance to improve data quality.
Statistical analyses yield defined
uncertainties based on defined data quality. Reliability and safety
analyses however are "per se" uncertain.
Statistical methods must be mathematically correct, whereas reliability and safety analyses must be technically correct.