Developing a Comprehensive Performance Evaluation Model while Enhancing the Resolution of Decision-Making Units in Data Overlay Analysis through a Fuzzy Inference System

The utilization of data envelopment analysis models for assessing and ranking organizational performance is on the rise. One of the important challenges of this model is the diminishing of the decision-making unit’s precision when dealing with a multitude of inputs and outputs. Hence, the aim of the present research was to develop a comprehensive performance evaluation model while enhancing the resolution of decision-making units. To this end, a balanced scorecard was used to identify comprehensive indicators. At the same time, for the first time, two objective and subjective approaches based on factor analysis and fuzzy inference system were used simultaneously to reduce indicators and improve the resolution of decision-making units. This study used an explanatory-descriptive method and was conducted as an applied-developmental research. The statistical population for identifying performance evaluation indicators and developing fuzzy inference rules included the experts of higher education institutions of Semnan city. Moreover, twenty-four higher education institutions of Semnan city were selected for model testing. The researcher made two questionnaires for the data collection. The validity of the questionnaires was confirmed by content and construct validity, respectively. Also, the reliability of the questionnaires was confirmed by Cronbach's alpha value and composite reliability of more than 0.7 respectively. The main accomplishment of the research can be designing a unified model with objective and subjective approaches to improve the resolution of decision units. In this regard, 26 indicators were identified and reduced to 8 structures by factor analysis. Also, the structures were scored by relying on the designed fuzzy inference system. The results demonstrated a significant improvement in the resolution of decision-making units when utilizing the proposed model, in contrast to conventional models which are mostly based on objective and subjective methods. As a result, the number of effective units in the proposed model effectively reduced to 10. Additionally, the results of the Kruskal-Wallis test and the calculation of the standard deviation of the efficiency scores revealed that the proposed model with an average rating of 48.29 and a dispersion of 0.221 has a lower efficiency rating and a greater dispersion as compered to other models. This finding serves as a confirmation of the enhanced resolution achieved by the proposed model