Natalia D. Nikolova UNIFORM METHOD FOR ESTIMATION OF INTERVAL SCALING CONSTANTS The paper presents a simulation procedure, called uniform method, for estimation of scaling constants. <...> Those take Numerical experiments demonstrate that the method accounts for the uncertainty in the subjective estimates. <...> The minimal number of replicas that could provide satisfactory precision of the simulation results is also established. <...> Keywords: scaling constants, uniform method, simulation, statistical test. the form of variables, uniformly distributed in their uncertainty interval. <...> The latter are estimated using a specially developed algorithm that employs triple bisection. <...> The task of estimation of scaling constants is brought down to the construction of the distribution of their sum and to the use of a two-tail statistical test to define its value. <...> Final point estimates for the constants are defined depending on the simulated estimate of pvalue and the chosen significance level. 1. <...> INTRODUCTION When choosing between alternatives, the decision maker (DM) usually identifies several objectives of importance. <...> As a result, consequences are modeled as multidimensional vectors, whose coordinates are the values of the parameters of importance for the DM, which measure the degree to which objectives are met. <...> The ideas of utility theory serve to rank such alternatives under risk [1]. <...> The DM must measure her/his preferences over each consequence using the utility function. <...> When consequences are of low dimension, the DM may analyze consequences as a whole. <...> Otherwise, the utility function is decomposed to several fundamental utility functions, built over groups of attributes [2]. <...> The existence of such fundamental utility functions is based on the existence of certain independence conditions between attributes. <...> For example, under the most common utility independence between attributes, the utility function may be decomposed to fundamental utility functions and their scaling constants. <...> The sum of the scaling constants defines whether the utility function should be represented in an additive or in an multiplicative form. <...> The value of the scaling constant for each attribute is elicited subjectively and coincides with the utility of the so-called corner consequence, where all attributes are set to their worst level except for the analyzed attribute, which is set to its best level. <...> When expressing preferences, real DMs show partial non-transitivity. <...> That is <...>