Multi-level analysis (Goldstein, 1987; Bryk & Raudenbush, 1992) has recently attracted the interest to the development and applications of level-structured data. The hierarchical models provide a description of the variability within and... more
Multi-level analysis (Goldstein, 1987; Bryk & Raudenbush, 1992) has recently attracted the interest to the development and applications of level-structured data. The hierarchical models provide a description of the variability within and across nested levels. Primary concerns are on the estimation and hypothesis testing of variance and covariance components. As a rule multi-level analysis refers to unbalanced settings and, therefore, relies on approximate tests derived from asymptotical theories. In order to give an indication of the accuracy of these test procedures we will provide some finite sample evidence. Only in a one-way random effects model this is possible with an analytical approach. In this paper we evaluate the small sample properties of the asymptotic tests for variance components in a balanced design. The following step might be an examination of the tests for unbalanced data, which should be carried out with the use of a simulation study. The concept of hierarchical ...
