报告人：马欣 教授 （美国肯塔基大学）

主持人：李淑惠 博士

报告时间：2020/10/10 周六上午 8:30-9:30 

报告地点/形式：线上报告（钉钉平台）

报告人简介：

Dr. Ma is Full Professor and Program Coordinator for Quantitative and Psychometric Methods at the University of Kentucky. He teaches courses in advanced statistics (e.g., meta-analysis, multilevel data analysis, and data mining techniques). He is a Spencer Fellow of the (U.S.) National Academy of Education, a recipient of the Early Career Contribution Award from the American Educational Research Association, (former) Canada Research Chair, and founder and (former) Director of the Canadian Center for Advanced Studies of National Databases. His research interests include advanced statistical (quantitative) methods, advanced data analysis of large-scale (state, national, and international) surveys, psychology of mathematics education, program evaluation and policy analysis, and organizational (school) effectiveness and improvement. He works to advance quantitative research, using latest statistical theories and models to improve and enhance quantitative analysis on critical issues in educational policy and practice. Among his numerous publications, he is author of the book A national assessment of mathematics participation in the United States: A survival analysis model for describing students’ academic careers. Dr. Ma is a member of the academic committee of the Asian Centre for Mathematics Education.

报告摘要：

Instrument validation is a critically important issue in the measurement and evaluation of human behaviors. Often, data do not come from individuals of interest, but from other individuals with close knowledge or experience about the individuals of interest, creating therefore data hierarchy. Previous methods for instrument validation ignore this data hierarchy. They are not appropriate and may produce serious biases concerning the instruments. This presentation is about a breakthrough in instrument valuation. One illustrative example assesses an instrument that uses responses from teachers to measure principal behaviors. To account for the data hierarchy of teachers nested within schools, we employ a multilevel confirmatory factor analysis and estimate multilevel (teacher and school) reliabilities to examine the instrument. We find that when teachers provide responses as indicators of principal behaviors, the instrument can generate a valid and highly reliable measure or estimate of principal behaviors at the school level. However, data analysis at the teacher level attempting to use teacher perceptions of principal behaviors as either dependent variables or independent variables should be avoided because of the low reliability at the teacher level. Keywords: multilevel confirmatory factor analysis, multilevel reliability analysis.

请前往钉钉官网（https://www.dingtalk.com）下载钉钉后扫描下方群二维码或者搜索群号进行申请。（申请时请一定备注单位+姓名）

群号：32251333