R McDowell posted on Tuesday, December 12, 2017 - 10:34 am
I see that a wide variety of outcomes can be used when fitting a latent growth curve to data eg. continuous, binomial, count. I wondered if theoretically it is possible to fit these models with grouped binomial data (i.e. at each point in time there would be recorded the number of successes and total number of events per individual), or whether theoretically it could be possible but from a practical perspective this could be difficult to estimate?
R McDowell posted on Thursday, December 14, 2017 - 11:57 pm
Thanks very much for those links. The idea of treating the outcome as akin to poisson rates was useful. I have found in some of my analyses that given the range of values in the numerators of the outcome Mplus sometimes tries to cap these. However at least I know now that this is possible. Many thanks.