Monitoring Change in Academic Achievement of Students and Schools: A Hierarchical Linear Modeling Approach

• Activity Number: 148
• Type: Contributed
• Date/Time: Monday, August 12, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section*
• Author(s): Xin Ma*+
• Affiliation(s): University of Alberta
• Keywords: Multilevel modeling ; Social statistics ; Education statistics ; Longitudinal analysis ; Academic achievement

Abstract

Educational data are often hierarchical (students nested within schools). Most conventional statistical models ignore this data hierarchy, producing inaccurate estimates of standard errors for model parameters. This study aims to develop a hierarchical linear model (HLM) that monitors longitudinal change in academic performance of students and schools, taking data hierarchy into account. A three-level HLM model is developed. At the first level, the measurement model specifies initial achievement status, linear rate of change, and quadratic rate of change that describes the nature of the change: acceleration, deceleration, or constant change. Longitudinal academic achievement data can be modeled in a within-student manner at this level. The second level is the between-student model in which individual change parameters can be modeled with student characteristics. The third level is the between-school model in which school change parameters can be modeled with school characteristics. This general HLM model is applied to data from the Longitudinal Study of American Youth (LSAY), a national six-year (Grades 7 to 12) panel study of mathematics and science education in the United States.