Cluster Size in Multilevel Models: The Impact of Sparse Data Structures on Point and Interval Estimates in Two-Level Models

• Activity Number: 149
• Type: Contributed
• Date/Time: Monday, August 4, 2008 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Bethany A. Bell-Ellison*+ and John M. Ferron and Jeffrey D. Kromrey
• Companies: University of South Florida and University of South Florida and University of South Florida
• Keywords: multilevel modeling ; sample size ; survey research

Abstract

Sample size at each level is important to consider when estimating multilevel models. Although general sample size guidelines have been suggested, the nature of social science survey research (e.g., large number of level-2 units with few individuals per unit) often makes such recommendations difficult to follow. This Monte Carlo study focuses on the consequences of level-2 sparseness on the estimation of fixed and random effects coefficients in terms of model convergence and both point estimates and interval estimates as a function of the level-1 sample size, number of level-2 units, proportion of singletons (level-2 units with one observation), collinearity, intraclass correlation, and model complexity. SAS IML was used to simulate 1000 data sets across 5760 conditions. Results are presented in terms of statistical bias, confidence interval coverage, and rates of model non-convergence.