R. Schappin posted on Saturday, April 12, 2008 - 1:59 am
I have a latent growth curve model with two growth curves of respectively cognitive development and motor development. The intercepts and slopes of both growth curves are correlated. My problem is that cognitive development is measured on an interval scale while motor development is a dichotomous variable (whether a child performed a certain movement or not). Is it possible to estimate both growth curves in one model?
Yes, you would use the CATEGORICAL option to specify that the motor development outcome is binary. The default is continuous so the cognitive development will be treated as continuous.
R. Schappin posted on Saturday, April 12, 2008 - 7:49 am
Thanks for your swift reaction. When I've defined the variables as categorical, could I then best use the WLSMV estimator (I assume default would be ML)? My sample size is rather small (74 for each occasion).
The default is WLSMV. You can ask for maximum likelihood using the ESTIMATOR option. You can do a Monte Carlo study to see if your sample size is sufficient. With repeated measures, the sample size can be smaller than with cross-sectional data. See the following paper:
Muthén, L.K. & Muthén, B.O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 4, 599-620.