Estimating Context Effects: A Simulation Study

• In Event: Multilevel Modeling Methods and Applications II

Abstract

In multilevel modeling, the relationships between the criterion and predictors are investigated at different levels. Often, the cluster-level predictors are measured by aggregating the individual-level measures. However, the aggregated cluster-level predictors do not always reliably measure the cluster-level construct. By using aggregated cluster-level predictors, traditional multilevel regression may yield biased estimation of the cluster-level regression coefficient, and therefore of the context coefficient. The bias is impacted by the within-cluster sample size, the cluster-level sample size, and the ICC for the predictor. A comparison is made of the accuracy with which the context coefficient is estimated by the traditional multilevel model and the multilevel latent variable model under a wider range of condition than has been investigated in previous research.

Authors

• Miao Gao, University of Florida

• James Algina, University of Florida