Study On Scientific Research Evaluation In Univesities Based On DEA Efficiency

Efficiency evaluation is a common task in economic circles or in social activities. The scientific efficiency evaluation is very important for enterprises, governments or organizations to improve performance, and can help enterprises, governments or organizations to provide high quality or quantity of products and services with the least consumption of costs.The dissertation employs Data Envelopment Analysis (DEA) to evaluate efficiency. DEA is a useful tool to evaluate the relative efficiencies among Decision Making Units (DMUs) with multiple inputs and multiple outputs. Charnes et al proposed the first DEA model, i.e. CCR model, in 1978. Henceforth DEA has become one of the most important methods for efficiency evaluation. In these years DEA has been widely used in efficiency evaluation for enterprises, governments, financial organizations, universities, airports, hospitals and so on.The current dissertation has studied the rating and ranking problem for universities. The shortcoming of traditional DEA methods in ranking is that the weight systems are not unique. The dissertation proposed two approaches to induce common weight in ranking so that all decision making units (universities) can be given efficiency scores and ranked under a set of common weights which can be accepted widely. Therefore the obtained ranking results from the proposed approaches are more believable.The first approach (A Common Weights DEA model based on standard samples) is administration-oriented under the centralized decision-making mechanism. The administration sector determines some universities as standard samples, and the standard sample universities make decision on the weight system which should be accepted by all of other universities as common weight.The second approach (A MOLP (Multi-objective linear programming)-DEA model) is democracy-oriented under the decentralized decision-making mechanism. All universities take part in the decision on the common weight, and the common weight should be accepted by more than a certain percentage, such as a half or 2/3, of universities.Besides, a technique about measuring the reliability of the results from peer review systems is developed in the dissertation. The technique can help to adjust the peer review results according to the reliabilities of different experts, which can improve the reliability of the peer review system.