Hello !
This question is related to lme(). I have a complex fixed-effects
structure in a linear mixed effects model (5 covariates and all possible
interactions). I want to select the best model according to AIC (ML
estimations), given a constant random structure.
Does it make sense to use the following strategy:
1. Use an OLS regression and a stepwise algorithm based on AIC (such as
S+ stepAIC(), MASS library), to discard the unimportant
covariates/interactions,
2. Test all the possible combinations of the remaining fixed effects
with mixed-effects models,
3. Choose the one with the lowest AIC ?
I have little doubt for 2 and 3, but is it safe / sound to assume that
all the unimportant covariates or interactions with OLS will also be
found to be unimportant with a mixed model ?
Thanks in advance and best regards,
Renaud
--
Renaud Lancelot
ISRA-LNERV
BP 2057 Dakar-Hann
Senegal
tel (221) 832 49 02
fax (221) 821 18 79
email renaud.lancelot@cirad.fr
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|