A Comparison of Classical and Bayesian Statistical Analysis in Operational Testing.
GEORGIA INST OF TECH ATLANTA
Pagination or Media Count:
This research is devoted to investigating how Bayesian statistical analysis differs from classical statistical analysis in the context of operational testing. The specific aspects of operational testing which are considered are the power resulting from a hypothesis test and the expected loss, or risk, resulting from a decision. First it is shown that it is quite difficult to develop a meaningful measure of comparison between Bayesian and classical analysis in the framework of hypothesis testing. Using the power of the hypothesis test as a measure of comparison, it is shown that under certain conditions classical statistical procedures lead to more powerful tests than Bayesian procedures. It is then shown that Bayesian statistical procedures are superior to classical procedures in the framework of minimizing expected loss or risk. Author
- Statistics and Probability