Confidence Intervals as an Extensions of Wigmorean Networks

Abstract

This thesis proposes an extension to the Baconian-Fuzzy probabilities previously developed as part of research into Wigmorean Networks (WigNets), providing a framework for evaluating the appropriate degree of confidence for a derived probability. We first give an overview of the existing structure of such probability evaluation. Based on existing Department of Defense (DoD) and Intelligence Community (IC) confidence evaluation guidelines, we establish criteria for the addition of confidence to the current network structure. Based on this analysis of existing guidelines, we present an empirically grounded framework for the evaluation of the degree of confidence that should be associated with a given probability. We next give justification for the behavior of this framework with a comparison with the measure theory underlying the establishment of other, similar Fuzzy confidence intervals.