Abstract
The Klein design is a response-driven random rule to allocate experimental subjects between two treatments. It aims to allocate more subjects to the treatment that is performing better, and therefore it is useful when ethical issues are of prime interest. It behaves asymptotically as the drop-the-loser rule, which is known to have a high degree of compromise between ethics and inferential properties. Besides, the Klein design has a powerful stochastic structure, which permits to obtain exact values, for each sample size n, for its main operating characteristics, such as variability of allocations, expected failure rate and power, selection bias or accidental bias. These properties of the Klein design are thoroughly studied and we obtain exact and asymptotic results.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call