R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Anova method for grouped objects

anova.grouped

R Documentation

Anova method for grouped objects

Description

Performs a Likelihood Ratio Test between two nested grouped models.

Usage

## S3 method for class 'grouped'
anova(object, object2, ...)

Arguments

`object`	an object inheriting from class `grouped`, nested to `object2`.
`object2`	an object inheriting from class `grouped`.
`...`	additional arguments; currently none is used.

Value

a list of class aov.grouped with the following components:

`name0`	the name of the null model represented by `object`.
`L0`	the log-likelihood under `object`.
`df0`	the number of parameters in `object`.
`AIC0`	the AIC under `object`.
`BIC0`	the BIC under `object`.
`name1`	the name of the alternative model represented by `object2`.
`L1`	the log-likelihood under `object2`.
`df1`	the number of parameters in `object2`.
`AIC1`	the AIC under `object2`.
`BIC1`	the BIC under `object2`.
`L01`	the value of the likelihood ratio test statistic.
`p.value`	the p-value of the test.

Warning

The function does only partial checking whether the two models are nested; the user is responsible to supply nested models in order to perform a valid test.

Note

anova.grouped performs a likelihood ratio test between two nested models; for simple Wald tests for the estimated parameters use summary.grouped.

Author(s)

Dimitris Rizopoulos d.rizopoulos@erasmusmc.nl

Examples


m1 <- grouped(cbind(lo, up) ~ treat, link = "logit", data = Sdata)
m2 <- grouped(cbind(lo, up) ~ treat * x, link = "logit", data = Sdata)
anova(m1, m2)

m1 <- grouped(equispaced(r, n) ~ x1, link = "logit", data = Seeds)
m2 <- grouped(equispaced(r, n) ~ x1 * x2, link = "logit", data = Seeds)
anova(m1, m2)