Potsdam Mind Research Repository

Reproducible Research

  • Increase font size
  • Default font size
  • Decrease font size
Home Research Topics Other Experiments and Analyses Bates et al. (2015) Parsimonious mixed models

Bates et al. (2015) Parsimonious mixed models

E-mail

Bates, D., Kliegl, R., Vasishth, S., & Baayen, R.H. (2015, arXiv). Parsimonious mixed models.

The analysis of experimental data with mixed-effects models requires deci- sions about the specification of the appropriate random-effects structure. Recently, Barr, Levy, Scheepers, and Tily 2013 recommended fitting ‘maximal’ models with all possible random effect components included. Estimation of maximal models, however, may not converge. We show that failure to converge typically is not due to a suboptimal estima- tion algorithm, but is a consequence of attempting to fit a model that is too complex to be properly supported by the data, irrespective of whether estimation is based on maxi- mum likelihood or on Bayesian hierarchical modeling with uninformative or weakly infor- mative priors. Importantly, even under convergence, overparameterization may lead to uninterpretable models. We provide diagnostic tools for detecting overparameterization and guiding model simplification.

To download, use right-click or CTRL-click and then "save as..."
Download this file (Bates_etal.ParsimoniousMixedModels_v2.2015.pdf)Bates_etal.ParsimoniousMixedModels_v2.2015.pdf[ ][ ]320 Kb20/06/2019 18:14
Last Updated on Thursday, 20 June 2019 18:41  

Add comment



See also


Copyright © 2020 Potsdam Mind Research Repository (PMR2). All Rights Reserved.
Joomla! is Free Software released under the GNU/GPL License.