高校教师教学水平评价的概化理论运用

Application of Multivariate Generalizability Theory to Teaching Level Evaluation for College Teachers

当前,高校教师教学水平评价方法存在两个问题:缺乏问卷评卷可靠性分析和难以进行评价的微观分析。概化理论在分析问卷评价的可靠性方面具有考虑多个因素和事后分析的多种优势;概化理论具有微观的分析功能,即微观地处理或尝试在挖掘数据信息,有利于提高高校教师教学水平评价的内部信度,有效监测教师教学水平的质量,从而正确引导教师教学能力的发挥。

Aiming at a non-standardized questionnaire,i.e.Teaching Level Evaluation Questionnaire(Scale)for college teachers,by which five hundred thirty-four college students made an evaluation for sixteen teachers,this paper uses Multivariate Generalizability Theory(MGT)to analyze the dependability of data from macroscopic and the microscopic angles.Moreover,the paper discusses whether MGT can be seen as an effective tool used to probe into data quality in order to provide the necessary guarantee of subsequent data analysis.The results show that:(1)in a vertical data analysis,the teaching level evaluation of non-standardized questionnaires has a high overall reliability(0.78109),while the local reliability for some the variables(dimensions)are not high.It is difficult to guarantee the reliability of the evaluation and evaluation results of vertical data analysis are insufficient;(2)in the horizontal data analysis,the integral dependability for teacher evaluation is high and the reliability is 0.92266,while there is a large difference when the students evaluate the twelfth and fifteenth teachers because there are quite a few conditional standard errors of measurement including estimating synthetic estimate of error variance and estimating synthetic estimate of error standard deviation.It is more controversial for the two teachers when the students evaluate their teaching levels.Because their dependability is not up to the standard,the results for the two teachers cannot be used;(3)Multivariate Generalizability Theory can analyze the dependability of Teaching Level Evaluation Questionnaire(Scale)for College Teachers data in vertical and horizontal aspects from macroscopic and microcosmic angles in order to detect the reliability of evaluation data.Compared with the Classical Test Theory(CTT)and Item Response Theory(IRT),the Multivariate Generalizability Theory reflects a strong advantage,which can be used as an effective 'tool' to analyze such evaluation data.