BICn: An alternative BIC for longitudinal models

Description Usage Arguments Details Value Author(s) References Examples

Description

Calculates the BIC as -2 * log-likelihood + log(n) * npar for a longitudinal model where npar is the number of parameters in the fitted-model and n is the number of subjects

Usage

1
BICn(object,...)

Arguments

object

a fitted longitudinal model object

...

some methods for this generic function may require additional arguments

Details

When applying the BIC in a longitudinal context, there is some debate as to whether the sample size should be taken to mean the number of subjects or the total number of observations across all subjects (see Section 7.3 of Hedeker and Gibbons, 2006).

Assuming the default BIC function accounts for the latter case, this generic function can be implemented for longitudinal models where the number of subjects can be extracted in order to calculate the BIC under the alternative definition.

Value

A numeric value with the BIC of the longitudinal model, with the penalty taken as a function of the number of subjects as described.

Author(s)

Maurice Berk [email protected]

References

Berk, M. (2012). Smoothing-splines Mixed-effects Models in R. Preprint

Hedeker, D. & Gibbons, D. R. (2006). Longitudinal Data Analysis. Wiley

Examples

1
2
3
  data(MTB)
  fit <- sme(MTB[MTB$variable==6031,c("y","tme","ind")])
  BICn(fit)

sme documentation built on May 2, 2019, 4:03 a.m.