The systems of preferences of expertise participants are an important indicator of the influence of the human factor on decision making. Their applied aspect lies in the use of an integral assessment of the investment attractiveness of objects of expertise/projects to solve multicriteria problems, as well as the establishment of “compromises” in the requirements for the degree of expression of investment attractiveness features inherent in each object/project. The system of advantages is an ordered series of specified features (n = 18): from more significant (significant, attractive, etc.) to less significant.
The implementation of a multi-step technology and algorithm for identifying and rejecting marginal thoughts, eliminating the “systematic error of the survivor” made it possible to identify four subgroups from the initial sample of experts numbering m = 90 people (mС = 30 people, mН = 12 people, mМ = 11 people, mТ = 6 people) whose internal group consistency of opinions about the significance of features of investment attractiveness satisfies the range of system-information criteria at an unusually high level of significance for human factor studies a = 1 %. It is substantiated that the group system of preferences of members of the mС subgroup should be considered basic. The opinions of marginal experts form a subgroup of mU = 31 people.
The degree to which experts differentiate the significance of features of investment attractiveness in the process of compiling them is determined by the number of “related” ranks and is taken into account when determining the Kendall dispersion coefficient of concordance (agreement). It is proposed to apply the entropy of the fragmentation of features for the same purpose. For each of the m subjects, normalized entropy indicators were determined, which were generalized both for group m and for subgroups mС, mН, mМ, mТ. Using the Student’s test, a statistically probable (a = 1%) agreement between the average entropy indicators was established. Therefore, the criteria for dividing them into subgroups-clusters according to the applied technology for identifying and screening out marginal thoughts and eliminating the “systematic survivor bias” are important.
The paradoxical nature of the research hypothesis has been established, since it is logical to assume that the more competent the expert, the more strictly he will order the studied features of investment attractiveness, and therefore the less entropy of ranks should then be observed in his system of advantages. On the other hand, the same high level of expert competence can lead to his conscious caution in ordering the studied traits, and therefore the use of a larger number of “connected (middle)” ranks, which will contribute to an increase in their entropy.
For the mС subgroup, recognized as the basic one, it was found that greater entropy is characteristic of a more significant feature of the investment attractiveness of the objects of examination. The well-known approach to determining entropy concordance coefficients did not turn out to be effective under the conditions of our research and needs further development.
Taking into account the issues highlighted, further steps are outlined for the development of information-entropy technologies for expert research.
