Factor scores in Growth Mixture Modeling
Message/Author
 Hanno Petras posted on Thursday, June 19, 2003 - 2:16 pm
Dear Linda & Bengt,

when using saving FScores from a growth mixture model, among other information two sets of individual latent growth paramteres are saved. The second set uses the prefix c_. What is the difference between the two? Thanks.

Best,

Hanno
 Linda K. Muthen posted on Thursday, June 19, 2003 - 2:19 pm
The one with prefix c is for the most likely class. The other is mixed over classes.
 Hanno Petras posted on Friday, June 20, 2003 - 6:42 am
Thanks. In order to get an idea about the individual trajectories within one class, it makes most sense to use the ones for the most likely class. Would you agree with that?

Best,

Hanno
 Linda K. Muthen posted on Friday, June 20, 2003 - 8:48 am
Yes.
 JPower posted on Monday, May 08, 2006 - 6:28 pm
Hello Drs. Muthen,

Thank you so much for the 'Mplus Discussion'.

If I measure the same factor over time (with identical indicators), and want to introduce the change in this factor over time to predict the outcome of interest at each time, how do I do this? Is it possible to subrtact the factor scores between times for each individual when specifying the model?

Thank you.
 Bengt O. Muthen posted on Tuesday, May 09, 2006 - 5:52 am
You should define the change as a latent variable in the model. For example with 2 time points and factors f1 and f2, you define fchange = f2-f1+0 by the input:

fchange on f2@1 f1@-1; fchange@0;
fchange with f1-f2@0;

You can dummy define the fchange factor as

fchange by x@0;

where x is any observed variable.
 Violet Xu posted on Tuesday, April 08, 2014 - 8:48 am
Hi,Dr. Muthen,

I am requesting factor scores of a growth curve model with binary outcomes.

iy sy | y1@0 y2@1 y3@2 y4@3 y5@4 y6@5 y7@6;

savedata:
file is temp4_factor.txt;
save = fscores;

Would you explain some of variables in the saved data set as below? For example, what are IY and IY_SE?

Order and format of variables

IY F10.3
IY_SE F10.3
SY F10.3
SY_SE F10.3

Thank you.
 Linda K. Muthen posted on Tuesday, April 08, 2014 - 9:59 am
IY is the factor score for IY. IY_SE is the standard error of the factor score for IY.
 Violet Xu posted on Tuesday, April 08, 2014 - 1:53 pm
Thank you, Linda. I have a follow-up question. From the output, the variance of IY and SY were 2.673 and 1.606, respectively. However, when I simply ran descriptive statistics, the variances of these two factor scores were 1.44 and 0.27, respectively. Thus, I am wondering whether I missed anything here?

Additionally, what is the relationship between the factor score of IY to the threshold of each individual?
 Linda K. Muthen posted on Wednesday, April 09, 2014 - 10:47 am
The factors in the model are not the same a factor scores based on the estimated model.

I'm not sure what you mean by the threshold of each individual.
 Samuli Helle posted on Wednesday, August 20, 2014 - 2:20 am
Dear Drs Muthén,

We are fitting a LGC model (not mixture) and want to plot slope factor scores (save = fscores). However, all the slope values were negative, which sounds odd given that factor mean should be set to zero. Am I missing something or is there an error in our analysis? In our LGC model we have used free time scores and also estimated an interaction between intercept and slope factors on an outcome (if that has something to do with this problem).