| As the main body of the postgraduate education,postgraduates’ development is not only the focus of the government and schools,but also related to the quality of postgraduate education.According to the statistical data from the Ministry of Education of the People’s Republic of China in 2019,nearly 85%of postgraduates were master degree candidates.And the percentage is still increasing.Therefore,it is important to discuss the aspects in the training of master degree candidates,and to improve the master’s program.The study on the development of postgraduates has entered a mature stage,but there are still three theoretical issues that need to be resolved critically:(1)most scholars have only studied the factors that affect the development of postgraduates from one aspect in previous studies;(2)a unified analysis framework of postgraduates’development has not been established in an integrated way,so it is difficult to fully reveal the interrelationships between the factors;(3)previous studies lacked quantitative analysis on the development satisfaction of postgraduates.Without analyzing the priority of factor optimization,the optimization direction could not be determined.To solve these three issues,this study proposed the following solutions.Based on the results of relevant literature and investigating,the factors affecting the development of master degree candidates were preliminarily screened out.After that,we designed a pre-investigation questionnaire with Likert Five-point Scale method.The affecting factors were further determined by pre-investigation and analysis,and the questionnaires was optimized.After the formal investigation,the reliability and validity of the questionnaires are excellent.By using the exploratory factor analysis,the factors were classified into five categories:psychological characteristics,growth ability,growth incentives,growth resources,and master degree candidates’ social network.These five categories constituted a multi-dimensional analysis framework for the development of master degree candidates.We selected the optimization model comparatively better reflecting the complex relationships between factors with the full model exploration technology of structural equation and existing researches.By using multiple regression analysis,we examined the mechanism of action between the factors.It was found that there are 7 direct effect paths and 4 indirect effect paths between factors.Finally,the complex relationship between the factors was determined,which can provide theoretical support for the of construction Bayesian network topology.To solve the third issue,we proposed to build a hybrid model combining structural equation model and Bayesian network.As the bridge connecting structural equation model and Bayesian networks,the value of latent variables were calculated by factor score method and Bayesian estimation respectively.After testing,it was verified that the Bayesian estimation method was more effective in this article.Then,according to the structural equation model and the value of latent variables,a Bayesian network was constructed successfully.Through the inverse reasoning function of the Bayesian network,an optimized path was determined from the perspective of the whole and different categories,this path could improve the development satisfaction of master degree candidates. |
