DHARMa (version 0.1.3)

testOverdispersionParametric: Parametric overdipersion test

Description

This function implements a parametric dispersion test based on comparing the residual devialce to the residual degrees of freedom that is commonly used, with the purpose of benchmarking against the nonparametric tests of DHARMA

Usage

testOverdispersionParametric(model)

Arguments

model

a fitted model object. See details for possible models

Details

The general idea of such as test is to copy GLM wisdom that we can define a dispersion parameter as residual deviance / residual degrees of freedom. For a model with correct dispersion, this parameter

1. Should be on average 1 2. Be chi2 distributed with df = rdf

For GL(M)Ms, we have to answer three questions

1. What is the residual deviance 2. What are the rdf 3. Is the distribution still chisq

There are quite a few implementations of this idea, e.g. https://stat.ethz.ch/pipermail/r-sig-mixed-models/2011q1/015392.html (implemented in blmeco::dispersion_glmer), http://glmm.wikidot.com/faq, and the code from Harrison, X. A. Using observation-level random effects to model overdispersion in count data in ecology and evolution PeerJ, 2014, 2, e616 The implementation here follows the suggestion in http://glmm.wikidot.com/faq, which is based on dividing the pearson residuals by the (probably not completely accurate) rdf, and testing this against a chi2 distribution with df = rdf.

See Also

testSimulatedResiduals, testSimulatedResiduals, testZeroInflation, testTemporalAutocorrelation, testSpatialAutocorrelation