Testing for the Absence of Random Effects in a Two-way Nested Design with Mixed Effects Model: A Nonparametric Approach |
| |
Authors: | Tathagata Banerjee Atanu Biswas |
| |
Institution: | (1) Indian Institute of Management Ahmedabad, Vastrapur, Ahmedabad, 380015, India;(2) Applied Statistics Unit, Indian Statistical Institute, 203 B. T. Road, Kolkata, 700 108, India |
| |
Abstract: | Two-way nested design with mixed effects model arises in many practical situations. In the classical analysis of variance
set-up, a test for the absence of the random effects is obtained under the assumption that the random effects and the errors
are normally distributed. The present paper avoids this assumption and provides an asymptotically distribution-free test procedure
for the above problem. The asymptotic null distribution of the test statistic is obtained. Actual implementation of the test
is straight forward given the prior information on quantiles of the intra-block differences of observations. In the absence
of such information, working test procedures are proposed. The performances of these tests are compared with the classical
analysis of variance test through simulations. The tests are then illustrated by some real data sets. |
| |
Keywords: | Analysis of variance Asymptotically distribution free Normal error |
本文献已被 SpringerLink 等数据库收录! |
|