Educational Panel Surveys: Statistical Issues

Abstract

In 2001, Taiwan will launch the first national panel survey of students -- the Taiwan Education Panel Surveys (TEPS). TEPS covers two cohorts of students who will be followed at least till the end of high school. TEPS is designed to provide multidimensional (student, class, school, teacher, parent) and multilevel (student, class, school, region) information and large sample size that will permit a great variety of complex multivariate analysis required for testing theories in education, sociology, psychology, economics, and related social sciences. The most fundamental outcomes of interest are analytic skills and psychological well-being. Rich information offers valuable opportunities and tough challenges. I will highlight some of the statistical issues involved and hope to obtain advice and suggestions from a statistical audience.