Basic problem of psychological measurements is disparity between psychological analysis of measurement process and corresponding mathematical models. Mathematical model of any process is valid if there is prior idea of the process nature. When using probability distributed with regard to the general principle there is no solution of the problem, the most problematic aspects are just avoided. In this case the subject of the research is handled as a black box. Using the results of direct methods of stimulus psychological measurements having no physical measures it is possible to construct a metric scale of intervals and relations. In the scales of this kind there is no particular unit of measurement since corresponding acceptable transformations enable to use a conditional unit.