자료포락분석(DEA) 모형에 따른 대학의 효율성과 대학평가결과 비교

A Comparison Between University Evaluation and Its Efficiency as Measured by DEA

  대학종합평가와 DEA 효율성의 관계를 규명하고, DEA 모형에 따라서 효율성이 달라지는지를 규명하고자 하였다. 종합모형을 사용하여 단위지표 데이터를 사용하는가 총량 데이터를 사용하는가, 그리고 한 변수씩 빼거나 더하는 식으로 9개의 변환모형을 포함하여 총 11개 모형을 설정하였다. 분석을 위한 자료는 2005년도 대학종합평가 자료로서 61개 대학이 분석대상이었다. DEA 효율성과 대학종합평가의 상관관계 분석 결과는 11개 모형중 1개 모형에서만 효율성과 대학종합평가 점수 간 상관이 .05% 수준에서 유의하였는데, 그 값은 .235에 지나지 않았기 때문에 대학종합평가가 효율성을 평가하는 것으로 보기는 힘들었다. 그리고 DEA 모형에 따른 효율성 분석결과는 어떤 변수가 배제되거나 추가되는가에 따라서 그리고 데이터의 특성에 따라서도 차이가 나타남을 발견할 수 있었다. 이 결과에 근거하여 대학을 평가할 때 효율성을 고려할 필요가 있지만, 성급한 DEA 적용보다는 종합평가총점과 효율성 간의 관계를 이론적으로 재정립할 수 있도록 연구가 후속될 필요가 있음을 논의하였다.

&nbsp;&nbsp;Does university evaluation include the evaluation of efficiency of the university? Is the efficiency measured by DEA applicable to evaluate the efficiency of universities? To reach answers, first, the university evaluation results were correlated with the efficiency as measured by Data Enveloping Analysis(DEA), and second, consistency of DEA results was tested against the alteration of DEA models. A total of 11 alternating DEA models were made by adding or deleting a variable or by changing scales based on a comprehensive model. Data was from 2005 university evaluation, in which 61 universities participated.<BR>&nbsp;&nbsp;First finding was that there was not substantive evidence that the university evaluation was assessing efficiency. Only one DEA model out of 11 ones showed statistically significant relationship between efficiency and evaluation result. But the correlation coefficient(r= .235) was not substantively high enough.<BR>&nbsp;&nbsp;Second finding was that efficiency of the universities was inconsistent over the alteration of the DEA models. The efficiency was affected by the adding or deleting variable or scale change.<BR>&nbsp;&nbsp;Considering the findings that the current school evaluation system does assess efficiency, and that efficiency is affected by DEA models, it was concluded that a thorough and careful study to set up a DEA model must precede an application of DEA in evulating efficiency of universities in reality.