On errors related to the use of expert estimates.

The use of expert estimates has become a normal procedure for many applied problems, usually reduced to testing statistical hypotheses. A typical example is the problem of medical diagnostics, when experts are asked, for instance, to estimate the probability of appearance of different symptoms of different diseases. This paper treats the problem of the influence of the accuracy of expert estimates on the properties of the statistical tests used, that is, changes of their critical levels and powers are considered.