 |
Dash Weerasinghe and Timothy Orsak
Dallas Independent School District
Dallas, TX 75204
April 6, 1998
Hierarchical Linear modeling is becoming a widely used technique to model student achievement within and between schools. As such there have been no validation studies done on simulated data where hierarchical modeling is used to recover the characteristics of the original simulated data. This paper describes the results of a large scale simulation study where student data within schools are simulated with known student and school characteristics and hierarchical linear modeling is used to recover these characteristics. Included will be the limitations of HLM with respect to the minimum number of students needed per school to effectively recover the student characteristics, the smallest difference in effect size that can distinguish between two schools and the combination of the above two parameters that can lead to false results.
|