Published in: Int. J. Theor. Physics 33, 171 (1994).
Abstract:
We are looking for a way of combining experimentally determined probabilitiesthat should yield maximum predictive power. This concept is defined aspermitting calculation of the accuracy of future experimental results solelyon the basis on the number of runs whose data will serve as input for makingthe prediction. Each probability is transformed to an associated variablewhose uncertainty interval depends only on the amount of data and strictlydecreases with it. We find, that for a probability which is a function of twoother probabilities maximum predictive power is achieved when linearly summingtheir associated variables and transforming back to a probability. Thisrecovers the quantum mechanical superposition principle.