groupEffectParameters: Group Indistinguishable Effect Parameters
In hardmanko/CMBBHT: Cell Means Based Bayesian Hypothesis Tests

groupEffectParameters

R Documentation

Group Indistinguishable Effect Parameters

Description

Perform pairwise comparisons of effect levels to determine which effects are indistinguishable from one another. By default, this is based on credible intervals. For people who do not want to, or cannot, use Bayes factors estimated with the Savage-Dickey density ratio, this function allows for another method of determining whether effects are present in the data. In addition, this allows for an analysis of which factor levels differ from one another.

Usage

groupEffectParameters(postEffects, method = "credInt", CIP = 0.95,
  priorEffects = NULL, equalBF = 3)

Arguments

`postEffects`	A matrix of posterior effect parameters, as from `getEffectParameters`.
`method`	The method to use. `"credInt"`: Use credible intervals of the difference between effects to determine equality. `"BayesFactor"`: Use Bayes factors of a hypothesis test that the difference between effects is 0.
`CIP`	Used if `method == "credInt"`. Credible interval proportion.
`priorEffects`	Used if `method == "BayesFactor"`. A matrix of prior effects with columns ordered the same as `postEffects`.
`equalBF`	Used if `method == "BayesFactor"`. If the Bayes factor in favor of the null is greater than equalBF, the effects are considered to be equivalent.

Value

Invisibly, a list with the following elements:

grp: A matrix containing grouping information. All effects that share a letter are Indistinguishable. This is what is printed.
eqm: A boolean matrix contining information about which effects were equal (indistinguishable). The value TRUE indicates equality.
bf01: If method == "BayesFactor", a matrix containing Bayes factors in favor of the hypothesis that the two effects were equal.