Year: 2011
Type: article
Abstract: : Multilevel modeling is an increasingly popular technique for analyzing hierarchical data. Suppose a data set consists of J level-2 units with n j level-1 units within each level-2 unit, e.g., J schools with n j students per school. If there are no covariates examined at either level, the scenario is identical to simple cluster sampling. Given that one wants to model clustered data, the determination of optimal values for J and n j , for which a closed form solution does not exist, is of interest. The small sample properties of multilevel model parameter estimates provides insight to this problem. I investigate these small sample properties as follows: A fixed data set exists, from which I repeatedly sub-sample according to various specifications of J . After a reasonable model is estimated for the entire data, the same model is estimated for each of these sub-samples. The data used is the National Educational Longitudinal Study (NELS), a large multilevel data set from the U.S. Depart...
Author: David Afshartous
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2098702611