The problems, associated with sensitivity of quantitative multiattribute decision-making methods to the initial data and the reliability of the decision obtained by using these methods, are considered in the presented dissertaition. The methods discussed are applied to solving decision-making problems, consisting in the selection of one alternative out of the available several options, when a decision-making person seeks to achieve a number of aims, rather than a single aim, and the alternatives are evaluated based on a set of attribute, which may be conflicting. The research object is sensitivity of the popular quantitative multiattribute decision-making methods: TOPSIS, SAW and COPRAS to the initial data errors. The initial data inaccuracy may be caused by the errors made by a person, entering the data. The estimates of the alternatives may have some er-rors, which are not taken into account, when the considered methods are used with these data. The analysis of the related works has shown that there are me-thods for evaluating sensitivity with respect to one attribute, based on varying the significance values, such as the method of the most critical criterion (Trian-taphyllou 2000) and to many attributes - Tornado diagram. However, these methods do not show either the errors of solution results or the reliability level of the result obtained.

The main aim of the dissertation is the evaluation of the stochastic nature of multiattribute decision-making methods, based on quantitative measurement. The application areas of the created methods are the design of decision support systems as well as sensitivity and reliability analysis of decision-making methods. The main problems to be solved in the work are as follows: first to deter-mine sensitivity of multiattribute decision-making methods, and, second, to offer the methods for evaluating sensitivity of multiple attribute decision-making methods and the reliability of decisions made by using them. The first problem is stated, taking into account a possibly random quality of the data used in the alternatives‘ evaluation. The second problem is associated with the evaluation of the reliability and biases of the decision obtained by the considered methods.

The dissertaition consists of introduction, 4 chapters, Conclutions, References and 3 Annexes. In the introduction, the research problem, topicality of the work, a description of the research object, aim and tasks of the work, as well as the research methods, novelty of the research work, practical use of the results obtained and the defended statments are presented. 12 articles are published on the theme of dissertation: five of them are published in the rewiewed journals, while seven – in other journal. 8 reports were delivered at the conferences, orga-nized in Lithuania and other countries.

Raktažodžiai: The multiattribute decision making methods, a sensitivity analysis, a decision support systems, a reliability of the decision, a biase of the decision