An improved approach for "optimization" of multiple policy objectives

• Main Author: Tawfik, Perihan Soliman
• Format: Others
• Published: PDXScholar 1977
• Subjects: Operations research; Decision making > Mathematical models; Mathematical optimization
• Online Access: https://pdxscholar.library.pdx.edu/open_access_etds/568; https://pdxscholar.library.pdx.edu/cgi/viewcontent.cgi?article=1567&context=open_access_etds

This thesis involved three categories of activity; development and testing of an expanded version of ELECTRE II, also the development of a computer software program for ELECTRE II. The expanded version of ELECTRE II took the form of an input aiding questionnaire along with a tailored structure to suit a particular problem. The contents of the questionnaire were based on geueral problem solving concepts (techniques, strategies) gleaned from the systems science literature. This questionnaire assumed a programmed instruction format in contrast to that of an interactive computer software package, so that it would not be prohibitive in terms of expenses in its use. The second part of the research was the comparative testing of group decision quality. Improved ELECTRE II was compared to a competitive method called SPAN, regular ELECTRE II, and unaided group decision-making. The effectiveness of the improved "Front End" ELECTRE II was tested as follows: TREATMENT Group A : Decision using ELECTRE II with the improved Front End. CONTROLS Group B : Unaided decision. Group C : Decision using regular ELECTRE II. Group D Decision using "SPAN" consensus taking method. The hypothesis that ELECTRE II and Front End ELECTRE II provide equally good bases for group decision making as SPAN (which had numerous claims for its effectiveness), was tested using appropriate statistical methods. Results of the experiments showed that the regular ELECTRE II did not perform as well as SPAN. However, the improved version of ELECTRE II developed for this thesis did perform as well as but not better than SPAN. It is important to note, however, that the "experimental" task was clearly not favorable to ELECTRE II. Had the task displayed more complexity, we believe the improved version of ELECTRE II would have outperformed SPAN. We feel that our results provide evidence for the value of this improved version of ELECTRE II which, we hope, will lead to its widespread use.