bootstraping a mixed-model (lme)
The sampling of complete clusters works better, at least for normal data. See Davison and Hinkley "Bootstrap Methods and their Application" section 3.8. If the clusters are large then there will be no difference if sampling within clusters is performed as well, but there doesn't seem to be any point.
On 8 April 2015 at 13:25, Roslyn Dakin <roslyn.dakin at gmail.com> wrote:
This question is in response to Ken's point about bootstrapping mixed
models (that you have to resample whole clusters/subjects for the
nonparametric bootstrap). Why can't you nonparametric bootstrap a mixed
model by stratified bootstrapping? i.e., resampling within each level of
the random effect, assuming you have a decent number of observations for
each level
Many thanks,
Roz
--
Roslyn Dakin, PhD
Department of Zoology
University of British Columbia
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
*Ken Beath* Lecturer Statistics Department MACQUARIE UNIVERSITY NSW 2109, Australia Phone: +61 (0)2 9850 8516 Building E4A, room 526 http://stat.mq.edu.au/our_staff/staff_-_alphabetical/staff/beath,_ken/ CRICOS Provider No 00002J This message is intended for the addressee named and may...{{dropped:9}}