Hypothesis testing for positive feedback models: some uses of a modified Poisson distribution for loops involving the self‐fulfilling prophecy

Abstract

Briefly reviews the standard Poisson distribution and then examines a set of derivative, modified Poisson distributions for testing hypotheses derived from positive deviation‐amplifying feedback models, which do not lend themselves to ordinary statistically based hypothesis testing. The “reinforcement” or “contagious” Poisson offers promise for a subset of such models, in particular those models with data in the form of rates (rather than magnitudes). The practical difficulty lies in distinguishing reinforcement effects from initial heterogeneity, since both can form negative binomial distributions, with look‐alike data. Illustrates these difficulties, and also opportunities, for various feedback models employing the self‐fulfilling prophecy, and especially for confidence loops, which incorporate particular self‐fulfilling prophecies as part of a larger dynamic process. Describes an actual methodology for testing hypotheses regarding confidence loops with the aid of a “reinforcement” Poisson distribution, as well as its place within sociocybernetics.