DHARMa (version 0.2.2)

testDispersion: DHARMa dispersion tests

Description

This function performs a simulation-based test for over/underdispersion

Usage

testDispersion(simulationOutput, alternative = c("two.sided", "greater",
  "less"), plot = T, ...)

Arguments

simulationOutput

a DHARMa object with simulated residuals created with simulateResiduals

alternative

a character string specifying whether the test should test if observations are "greater", "less" or "two.sided" compared to the simulated null hypothesis. Greate corresponds to overdispersion.

plot

whether to plot output

...

arguments to pass on to testGeneric

Details

The function implements two tests, depending on whether it is applied on a simulation with refit = F, or refit = T.

If refit = F (not recommended), the function tests if the IQR of the scaled residuals deviate from the null hypothesis of a uniform distribution. Simulations show that this option is not properly calibrated and much less powerful than the parametric alternative testOverdispersionParametric and even the simple testUniformity, and therefore it's use is not recommended. A warning will be returned if the function is called.

If refit = T, the function compares the approximate deviance (via squared pearson residuals) with the same quantity from the models refitted with simulated data. It is much slower than the parametric alternative testOverdispersionParametric, but simulations show that it is slightly more powerful than the latter, and more powerful than any other non-parametric test in DHARMa, and it doesn't make any parametric assumptions. However, given the computational cost, I would suggest that most users will be satisfied with the parametric overdispersion test.

See Also

testResiduals, testUniformity, testOutliers, testZeroInflation, testGeneric, testTemporalAutocorrelation, testSpatialAutocorrelation

Examples

# NOT RUN {
# creating test data

testData = createData(sampleSize = 200, overdispersion = 0.5, randomEffectVariance = 0)
fittedModel <- glm(observedResponse ~ Environment1 , family = "poisson", data = testData)
simulationOutput <- simulateResiduals(fittedModel = fittedModel)

plot(simulationOutput, quantreg = FALSE)

###### Distribution tests #####

testUniformity(simulationOutput)

###### Dispersion tests #######

testDispersion(simulationOutput, alternative = "less") # underdispersion

###### Both together###########

testResiduals(simulationOutput)

###### Special tests ##########

# testing zero inflation
testZeroInflation(simulationOutput)

# testing generic summaries
countOnes <- function(x) sum(x == 1)  # testing for number of 1s
testGeneric(simulationOutput, summary = countOnes) # 1-inflation
testGeneric(simulationOutput, summary = countOnes, alternative = "less") # 1-deficit

means <- function(x) mean(x) # testing if mean prediction fits
testGeneric(simulationOutput, summary = means) 

spread <- function(x) sd(x) # testing if mean sd fits
testGeneric(simulationOutput, summary = spread) 

###### Refited ##############

# if model is refitted, a different test will be called

simulationOutput <- simulateResiduals(fittedModel = fittedModel, refit = TRUE, seed = 12)
testDispersion(simulationOutput)

###### Test per group ##############

simulationOutput = recalculateResiduals(simulationOutput, group = testData$group)
testDispersion(simulationOutput)


# }