Johan Claas posted on Thursday, July 03, 2014 - 2:26 pm
I would like to model panel survey data with four waves that are not equidistant, e.g. from the years 1996, 2002, 2008, 2011.
I want to do a classic HMM/LTA analysis. For this, do the time points have to be equidistant? If I add two additional years 1999 and 2005 (to make it all 3yr intervals) and fill this with missings the resulting classes are dominated by these missings and not very meaningful.