General model selection test for comparing linear mixed-effects models

ABDOLREZA SAYYAREH

No Detail Found!

General model selection test for comparing linear mixed-effects models

Author(s)

ABDOLREZA SAYYAREH Department of Computer Science and Statistics, Faculty of Mathematics, K.N. Toosi University of Technology, Tehran, Iran

Abstract

Under various situations, researchers may be interested in comparing nonnested (nonhierarchical) linear mixed-effects models to characterize longitudinal data. The main goal of this paper is to explore and develop separated hypotheses testing to compare misspecified and nonnested linear mixed-effects models based on a proposed estimated variance for the differences of the log-likelihood function. It also has been shown that simplifying the proposed approach gives the variance of Cox's test for mixed-effects model selection. In addition, it is shown that the proposed test could be used to compare fixed and mixedeffects models. To illustrate our approach, a simulation study leads to the results presented in the paper. We consider two nonnested and misspecified models to assess the robustness of the proposed variance to analyze the data on corn and soybean production in 12 Iowa counties, contained in the two data sets corn-soybean and corn-soybean means.

Publication Details

Page(s) 1-24

DOI DOI not available

Published Journal: Journal of Applied Probability and Statistics, Volume: 19, Issue: 1, Year: 2024

Request the full text from PASTIC

Keywords

and phrases, tracking interval, Cox's test, nonnested model selection, EM algorithm, linear mixed-effects model, Vuong's test

Citations

Downloads

Views