Sequential Hypothesis Testing with Spatially Correlated Count Data

It is well known that sequential hypothesis test procedures can have appreciable cost savings compared to fixed sample size test plans. The first sequential hypothesis procedure was developed by Wald for one-parameter families of distributions and later extended by Bartlett to handle the case of nuisance parameters. However, Bartlett’s procedure requires independent and identically distributed observations. In ecological applications, it is common for data to exhibit spatial correlations. We illustrate the existence of spatial correlations in pest count data by analyzing the spatial structure in a data set of mite counts. The goal of this paper is to show how to incorporate the existence of spatial correlation into a sequential hypothesis testing framework so that applications such as pest management can improve the accuracy of their treat or no-treat decisions.