Abstract
The problem of the estimation of parameters determined statistically from physical measurements is discussed. Emphasis is placed on the fundamental role played by the prior probability distribution for the parameter. The validity of "maximum likelihood" estimation is examined with particular reference to the case of the estimation of a parameter which actually has an unique but (originally) unknown magnitude. Situations in which the prior probability distribution for the parameter is completely unknown are treated and a method is described for the calculation of this distribution from appropriate experimental data. Many examples are given throughout from the field of cosmic radiation.
DOI:https://doi.org/10.1103/RevModPhys.25.818
©1953 American Physical Society