Testing Equality of Treatment Effects in a Two-stage Censored Allocation Design

Abstract

Tests for the equality of treatment effects are developed for a two-stage sampling design with failure censoring in each stage. Depending on the results in the first stage, adaptive failure censoring is adopted in the second stage. Assuming exponential responses, tests for comparing the response distributions are obtained and some related small sample and asymptotic properties are worked out. Relevant competitor is also suggested and compared with the proposed procedure.