SF Wang posted on Wednesday, September 25, 2013 - 10:59 am
when doing parallel process GMM, I was faced with warning message about PSI matrix. A random slope is involved. In addition, I noticed that the variance of the random slope is not significant. Therefore, I fixed its variance at 0. This gets rid of the warning messages.
But I am not quite sure about the interpretation of the slope parameter. Does it indicate everyone has the same slope? if so, subjects are categorized to different latent classes solely based on their intercepts? - seems not so since four latent classes identified have quite different mean slopes.