作　者：俞立平 钟昌标
关键词：人文社会科学；高等院校；投入要素盲点；联立方程；BP 神经网络
摘　要：

人文社会科学研究投入产出关系比较复杂，且存在要素盲点问题（即一旦投入要素没有通过统计检验或回归系数显著为负数，就难以比较判断各投入要素的贡献大小）。基于教育部高校人文社会科学省际面板数据，采用回归分析与BP 人工神经网络分析方法进行研究，结果表明：BP 人工神经网络是解决投入要素盲点的有效方法。从投入要素的贡献看，人文社会科学研究研发经费绩效总体不高。从要素关系看，研发人员通过学术论文的传导机制发挥其对学术著作和研究报告的作用，其综合绩效较高。学术论文与学术著作互动关系显著，研发经费与研发人员的互动效应良好。学术著作和研究报告对研发经费具有正向反馈作用，学术论文对研发人员的反馈作用较好。 

 

On the Interactive Relationship between Input and Output Of Humanities and Social Sciences Research in Institutions of Higher Learning

Concurrent Discussions about the Solution to the Blind Spot of Input Factors

Yu Liping（Changzhou University）

Zhong Changbiao（Guangzhou College of Commerce）

Abstract The relationship between input and output of humanities and social sciences research is a little complicated. There is also a blind spot of input factors, which makes it difficult for people to compare and determine the contribution of each input factor once the input factors fail to pass statistical tests or the regression coefficient is significantly negative. To address the lack, this paper proposes a method that uses both regression analysis and BP artificial neural network analysis for research, and describes the applicable conditions of the method. By using inter-provincial panel data of humanities and social sciences in institutions of higher learning from the Ministry of Education, the research results show that BP artificial neural network is an effective method to solve the blind spot of input factors, and promotes traditional input-output analysis theories and methods. The overall performance of R & D funding is not good, and the overall performance of R & D personnel is good. Academic papers and academic monographs interact well as follows: academic papers have a great impact on research reports; academic books and research reports have a good feedback effect on R & D funding; academic papers have a good feedback effect on R  & D personnel; R & D funding and R & D personnel interact well.

Key words      humanities and social sciences; institutions of higher learning; blind spot of input factors; simultaneous equations; BP artificial neural network

 

■    作者简介 俞立平，管理学博士，常州大学商学院教授、博士生导师；江苏 常州 213159；钟昌标，经济学博士，广州商学院粤港澳大湾区电子商务研究中心教授、博士生导师；广东 广州 511363。

[PDF]（下载数：）