Interim Analysis For Normally Distributed Observables

by Seymour Geisser and Wesley Johnson
Technical Report No. 576
School of Statistics
University of Minnesota
and University fo California, Davis
May, 1992

Research supported in part by NIGMS 25271.

Abstract

We address the problem of whether an experiment should be continued or aborted when N observations are in hand and a total of S>N have been scheduled for a decision. A Bayesian predictive approach is used to determine the probability that if one continued the trial with a further sample of size M where N+M>=S one would come to a particular decision regarding some set of parameters. In particular, sampling from a multivariate normal distribution will be discussed.

Click here to download the complete PostScript technical report.
Other formats include the Macintosh Microsoft Word version of this technical report.