Constrained Inference for Linear Mixed Effects Models

Estimation and inference for linear models where some or all of the fixed-effects coefficients are subject to order restrictions. This package uses the robust residual bootstrap methodology for inference, and can handle some structure in the residual variance matrix.


============================= PROJECTED DEVELOPMENT:

  • Better handling of environments
  • Better formula parsing (hopefully to remove dependence on lme4::lFormula)
  • Write automated tests
  • Improvements to GUI in shiny_clme(), particularly:
    • Allow for custom constraints to be specified.
  • Implement unconstrained model testing all pairwise comparisons.
  • Add argument to not bootstrap the random effect residuals. See Morris (2002) doi:10.1016/S0167-7152(02)00041-X
    • How does lme4 handle this, just use the observed random effects?
  • Enable parallel processing for the bootstrap samples (foreach, dopar).
  • Generalized LME models ============================= VERSION HISTORY:

<<<<<<< HEAD

Version 2.0-11, 2019-02-07

  • Removed a default value for seed in summary.clme
  • Fixed error from setting names to NULL

Version 2.0-10, 2018-05-18

  • Fixed out-of-sync metadata.
  • Model now assumes ordred factor for constrained effect. If not, issues warning and attempts to force ordered=TRUE .

Version 2.0-9, 2018-02-01

  • Fixed error in naming to prevent error with R 3.4.3

Version 2.0-8, 2017-11-16

  • Streamlined of formula parsing (less messy).
  • Fixes to plot.clme() for tree orders.
  • Fixed error in removing duplicate patterns from auto-generated contraints.
  • Added option to specify node in shiny_clme.
  • shiny_clme will produce code used to run the specified model.

Version 2.0-5, 2016-11-07

  • Added reference to JSS manuscript: Jelsema and Peddada (2016)

Version 2.0-4, 2015-07-01

  • Fixed error in logLik.clme.
  • Changed maintainer email.
  • Added explicit BIC.clme method (instead of forcing custom use of AIC.clme)
  • Removed the bootstrap computations from clme() and put them into summary.clme().
  • summary.clme() now returns a list of values (largely equivalent to output of clme()) but does not automatically print output to console.
  • print.summary.clme() prints the output of summary.clme to the console.
  • Improvements to the shiny_clme() application:
    • Added "model running" indicator
    • Enabled tab-delimited and xlsx formatted data input
    • Multiple tabs for output.

Version 2.0-3, 2015-05-14

  • Noted nsim is in the fitted object of clme().
  • Fixed error with method VarCorr() (failed unless "lme4" or "nlme" was loaded, redefined method VarCorr)
  • Replaced vignette with static document.
  • Implemented work-around to prevent error (stemming from package isotone) when there was only 1 constraint.

Version 2.0-2, 2015-01-20

  • Fixed typo in printing of log-likelihood, AIC, BIC
  • Added ellipses to MINQUE function to prevent errors from arguments em.eps and em.iter
  • Improvements to shiny_clme() application

Version 2.0-1, 2014-12-05

  • Minor updates to several functions

Version 2.0-0, 2014-12-01

  • Major changes to arguments, particularly:
    • Formula interface implemented
    • Removed necessity for custom PAVA algorithms (implemented function 'activeSet' from package isotone)
    • Nks replaced with group vector input
    • Qs removed, each random term now gets individual variance
  • Provided more common methods for fitted objects

Version 1.0-0, 2014-09-09

  • Initial release of package

=========================== Author / Maintainer: Casey M. Jelsema Assistant Professor Department of Biostatistics West Virginia University [email protected]

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


2.0-12 by Casey M. Jelsema, a year ago

Report a bug at

Browse source code at

Authors: Casey M. Jelsema [aut, cre] , Shyamal D. Peddada [aut]

Documentation:   PDF Manual  

GPL-3 license

Imports nlme, methods, isotone, stringr, prettyR, stats, openxlsx, graphics

Depends on shiny, lme4, MASS

Suggests testthat, nnet

See at CRAN