I was interested in conducting some exploratory analyses to identify latent classes within the context of a second order latent change score SEM. This is similar to what was done in McArdle and Prindle (2008; pg's. 714-716). However, I am having trouble producing syntax for this kind of model and haven't been able to find an example in the manual. Is there any example syntax available that might provide a helpful start in building this model?
McArdle, J.J. & Prindle, J.J. (2008). A latent change score analysis of a randomized clinical trial in reasoning training. Psychology and Aging, 32, 702-719.
Thank you Bengt, I will do that. In the mean time I managed to find a simplified version of an LDS input file by Selig & Preacher (2009) which I used since I'm interested in mediation too. The steps of the model testing seemed to develop fine until I entered two interaction terms (both between a regular latent and a change latent variable). I have tried to increase the STARTS several times but I always get the following error:
THE ESTIMATED COVARIANCE MATRIX COULD NOT BE INVERTED. COMPUTATION COULD NOT BE COMPLETED IN ITERATION 1. CHANGE YOUR MODEL AND/OR STARTING VALUES.
THE BEST LOGLIKELIHOOD VALUE HAS BEEN REPLICATED. RERUN WITH AT LEAST TWICE THE RANDOM STARTS TO CHECK THAT THE BEST LOGLIKELIHOOD IS STILL OBTAINED AND REPLICATED.
THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO AN ERROR IN THE COMPUTATION. CHANGE YOUR MODEL AND/OR STARTING VALUES.
This is a post in relation to the Latent Change Score Model I have previously posted about (see above). I have been able to test my model eventually and everything runs fine until I enter gender as a control variable (1= man; 2=woman) to all my dependent (change score) variables. Then I receive the following error:
ERROR Invalid symbol in data file: "ï»¿1" at record #: 1, field #: 1
I checked my data file (.dat) and everything looks ok. Controlling for age also works fine. Do you know what could be the problem?
One more question about my older post on latent difference score analysis (McArdle).
I know you do not give consultation on this method but there is a quite strange result in my output. Based on three time measurements (SR_1, SR_2, SR_3) I have created two change scores from T1 to T2 (DSR1) and from T2 to T3 (DSR2). The mean scores of these latent difference scores are small decimals (0.03 and 0.08 -which agrees with SPSS output). Then according to this method, I need to create a slope and an intercept for the change process as follows:
While the mean of DSRi seems correct in the output, the mean of DSRs is 2.9, which doesn't make sense since it should be the average of DSR1 and DSR2, so somewhere between .03 and .08. Do you know if I am modelling something wrong?