×

zbMATH — the first resource for mathematics

Empirical study of computational techniques used for parameters’ estimation in multivariate linear mixed effects models. (English. French summary) Zbl 1430.62110
Summary: An empirical study was conducted to assess the performance of some computational techniques used in multivariate linear mixed effects models (MLMM). Performances of four computational techniques were compared based on their accuracy, relative bias, relative efficiency and computing time according to sample size, between-variables and within-subjects correlations. The accuracy, relative bias, relative efficiency and computing time in estimating MLMM parameters varied significantly according to the computational techniques, sample size and correlations. Further theoretical developments are required to improve the accuracy of the four methods in estimating variances’ components.
MSC:
62H12 Estimation in multivariate analysis
68Q25 Analysis of algorithms and problem complexity
65Y20 Complexity and performance of numerical algorithms
PDF BibTeX XML Cite
Full Text: DOI Euclid