Anonymous posted on Wednesday, February 08, 2006 - 1:56 am
Dear Bengt & Linda,
Could you help me with the following guestion (I have examined the Mplus manual for this but I'm unable to produce the figures I need).
I have specified and multiple indicator growth model (three time points), and include in the overall model also two observed variables measured at T1 and T3. I would like to display regression curves from the level factor to the observed variables at T3, while controlling for the intial level of the observed variables. There is also a multigoup situation here, the variable used in the grouping obtion moderates these relations. I would produce the figures separately for the two outcome variables, both of which would contain regression curves in two groups of individuals.
The multigroup situation is not a problem (except that the program aborts when I try to view estimated means), but I don't know how to specify the plot -command correctly. Also, I'm guessing that the regression relation among the level and the outome variable should be included in the syntax while preparing the output to view the graphics - what exactly should be included in the syntax? Also, what would be the correct option in the graphical display for this situation? So far, I have been able to display a scatter plot of the level factor as a function of the T3 outcome variable in the two groups, but I would need the reggression curves instead (i.e., utilize the Beta values of the level factor in predicting the T3 outcome variable).
I'd like to display the figures in Mplus, because I'm using the missing data option as well.
I'm have a question about the visual presentation of plots when using a latent class growth analysis that models two processes (English and Spanish vocabulary measured at four time points during children's kindergarten school year) and specifying the following:
Plot; TYPE IS PLOT3; series is EOW1 (1) EOW2 (2) EOW3 (3) EOW4 (4); series is EOWS1 (1) EOWS2 (2) EOWS3 (3) EOWS4 (4);
Whenever I limit the syntax above and specify one series, and ask for a plot of estimated means and observed individual values, I get a growth trajectory for each latent class (which is what I expect). However, when I specify both series above, I get the two expected growth trajectories, but I also get a line connecting EOWS1 with EOW4 for latent class 1 and a line connecting EOWS1 with EOW4 for latent class 2. Latent class 3 turned out as expected, two lines going across the four time points. So my question is, why is there a third line connecting the two series for latent classes 1 and 2? I've looked through the mplus manual, the mplus discussion board, and consulted colleagues, but I haven't been able to figure out what's going on. I would greatly appreciate your feedback when you have a moment.