A Sequential Test for Contiguously Correlated Inventory Differences

The nature of statistical significance testing of a sequence of inventory differences (IDs) is that single test criteria are often used in the implementation of a test even though the IDs and hence the tests are correlated. As a consequence, the experienced false alarm rate for the sequence can be considerably larger than that presumably designed into the single test. Thus one is led to a consideration of test criteria that better address the false alarm rate problem. This paper is intended to do this. An application of Wald's sequential probability ratio test that only contiguous IDs are correlated results in one such test. 1 to sequences of IDs under the assumption