Abstract
Sequential planning with the restriction that the number of groups is absolutely bounded by some fixed number K is considered. These truncated sequential plans will stop sampling after most K groups of observations are collected. Practically, it reflects the following situation. There are only K days available to complete a certain study (e.g., clinical trials, to get FDA approval or investigate conditions of the field before it is time for farming). Every day, a sample of the desired size is selected. Based on the interim analysis performed, an inference is made or another batch of observations is collected as long as total number of groups does not exceed K. Truncated sequentially planned probability ratio tests (TSPPRTs) on a lattice are considered in detail. Construction and methods of evaluation of TSPPRTs are proposed.
ACKNOWLEDGMENT
Many thanks to Professor Michael I. Baron for his helpful ideas, comments, and suggestions.
Notes
Recommended by Wei Liu