Paper Summary
Share...

Direct link:

A Partially Nested Multilevel Model for Analysis of District Data Including Charter or Private Schools

Sat, April 6, 2:15 to 3:45pm, Fairmont Royal York Hotel, Floor: Convention Floor, Concert Hall

Abstract

The purpose of this paper is to describe an adapted partially nested multilevel model suitable for analyzing datasets containing both public schools clustered within districts and unclustered charter or private schools. We identify the adapted model, explain when and how it is useful, and contrast it with existing approaches to multilevel modeling with unclustered schools. In an empirical example, results of the new model are compared to several standard approaches to charter and private school analysis including traditional multilevel modeling approaches and cluster-robust regression. Our results show that other modeling approaches may be masking important differences between public and nonpublic schools through aggregation. Annotated syntax files and recommendations for future simulation are provided.

Authors