plantTrialLmmFitCompSel: Model fit, comparison and selection
In timflutre/rutilstimflutre: Timothee Flutre's personal R code

plantTrialLmmFitCompSel

R Documentation

Model fit, comparison and selection

Description

For a plant field trial, fit by maximum likelihood (ML) a global linear (mixed) model with all the specified terms, then fit by ML various sub-models and compare them (AICc) or test each term (F-test for fixed effects and likelihood ratio test for random variables), finally select the best model and, if there are random variables, re-fit it using restricted maximum likelihood (ReML).

Usage

plantTrialLmmFitCompSel(
  glob.form,
  dat,
  part.comp.sel = "fixed",
  alpha.fixed = 0.05,
  alpha.random = 0.1,
  ctl = NULL,
  saved.file = NULL,
  nb.cores = 1,
  cl = NULL,
  verbose = 1
)

Arguments

`glob.form`	character containing the formula for the global model
`dat`	data frame with all the columns required by `glob.form`; if the response contains an inline function (e.g., log or sqrt), the untransformed response may still be required
`part.comp.sel`	part(s) of the model (fixed and/or random); if only "fixed", the lm/lme4 and MuMIn packages will be used, otherwise the lmerTest package will be used
`alpha.fixed`	for lmerTest, threshold on p values of fixed effects below which they are kept
`alpha.random`	for lmerTest, threshold on p values of random effects below which they are kept
`ctl`	if NULL, will be the output of `lmerControl` with default arguments
`saved.file`	name of the file in which to save all model fits or from which to load all model fits (for MuMIn); for lmerTest, only the fit of the global model will be saved
`nb.cores`	number of cores; ignored with lmerTest
`cl`	an object of class "cluster"; if NULL, will be created automaticall based on `nb.cores`; ignored with lmerTest
`verbose`	verbosity level (0/1)