Could I use maximum likelihood method or any other frequenist method to estimate parameters for 3-level hierarchical model? Is there any references help me in this case? Thank you
-
$\begingroup$ What's the problem? You mean in theory? Or in practice? Did you try anything? $\endgroup$ – Gijs Nov 13 '17 at 17:25
-
$\begingroup$ I mean in practice. Yes I tried to use the maximum likelihood to estimate the parameters in the first level, then I use these estimates as data and I use it to estimate parameters in second level ( I use rma function from metafor package in r), then I consider estimates from second level as data and I use it to estimate parameters in third level also by using rma function. $\endgroup$ – Zainab Nov 13 '17 at 17:39
You can fit a mixed model with three levels, or more, with lme4
. There's not a lot mentioned in their vignette. You need the :
operator. Here is a link to a tutorial: http://rpsychologist.com/r-guide-longitudinal-lme-lmer.
-
$\begingroup$ Thank you, I am going to look at this tutorial and I hope that will help me. $\endgroup$ – Zainab Nov 14 '17 at 16:39