Page 1 of 1

dynamic autoregressive models

Posted: Thu Oct 26, 2017 9:59 am
by rdmcdowell
Hello. I see from the Lemma course on multilevel modelling of repeated measures (Ch 15) that Model 3 (where u`j' is a factor with loading 1 for time 1 and freely estimated for other time-points) cannot be estimated in Mlwin. I wondered if with the new release of Mlwin whether this is still the case.
I also wished to clarify something regarding the initial conditions problem. Supposing I am analysing an outcome measured annually from 2011-2016, and fitting an AR(1) model. But I also actually have the outcome measured from 2010. Can I just use it as my as my lagged predictor for 2011 ?