ABSTRACT
In this chapter, the authors find extensive coverage of various types of estimation problems. Recently, Mukhopadhyay and Sen suggested implementing the larger experiment via several identically replicated smaller “pieces”. One advantage of this approach is that one can now capitalize frequently on often available system of a parallel network and gain much operational convenience and thus cut cost. The other advantage is that one can easily provide an unbiased estimator of the variance of the final combined stopping variable. The authors summarize various purely sequential procedures for estimating the scale parameter, mean and percentiles of the negative exponential distribution. Mukhopadhyay and Chattopadhyay introduced meta-analytic tools by examining how such already available sample resources obtained from independent, but otherwise comparable studies can be fruitfully combined in order to make comparative statements between the treatments. Under two-parameter negative exponential models and different types of sampling strategies, Mukhopadhyay and Chattopadhyay developed appropriate asymptotic second-order characteristics for the final overall treatment comparisons.
