SF Wang posted on Thursday, September 05, 2013 - 6:59 am
Dear Linda and Bengt,
I need to model two dependent variables simultaneously, one ranging from 0 to 30, the other from 0 to 85. I was wondering whether we need to transform two outcomes first so that they are in the same range.
If transformation is appropriate, how should I do it? I feel that standardizing the outcomes at each time point is not appropriate since it will change the order of observations over time, potentially. Standardizing them at all time points does not sound right to me too, since observations over time are correlated.
If transformation is not a good idea, do the different ranges of the outcomes have any impact on the parameter estimation? In other words, does the outcome with wider range play a more (or less) important role when identifying the latent classes?
I would leave them in their original scale. This is not a problem.
SF Wang posted on Wednesday, September 18, 2013 - 1:18 pm
I was wondering whether you have any sample code for parallel process GMM, especially how to specify the plot statement to get one trajectory plot for each dependent variable.
currently, I have
PLOT: TYPE IS PLOT3; SERIES = Y11-Y14(*) Y21-Y24(*);
Each dependent variable has 4 time points. I got one trajectory graph with 8 time points. This is not what I want. I would like to see two graphs, one for Y1(Y11-Y14), one for Y2 (Y21-Y24), each has four time points.
See page 767 of the V7 UG showing how to use the | symbol to plot 2 processes.
SF Wang posted on Friday, September 20, 2013 - 7:52 am
I am not sure where to post this question -
Is there a way to ask Mplus to read the data from the second row? - the first row contains headers. Currently I just remove the headers manually. But I was wondering that Mplus may be able to do that automatically. (I searched in the Internet, but didn't find anything on this topic.)