Latent difference score and XWITH PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Ryan Vogel posted on Friday, November 15, 2013 - 11:32 am
Hello,

Following syntax provided by Selig & Preacher (at http://quantpsy.org/
supp.htm.), I have created a latent difference score representing the change in a measured variable from one time to another.

Now, using the XWITH function, I'm attempting to create a latent interaction term between the latent difference variable and another variable. Then, this interaction term will be used in an equation to predict a third variable. When I attempt to run the model, I receive the error message: "THE ESTIMATED COVARIANCE MATRIX COULD NOT BE INVERTED. COMPUTATION COULD NOT BE COMPLETED IN ITERATION 1. CHANGE YOUR MODEL AND/OR STARTING VALUES.
THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO AN ERROR IN THE COMPUTATION. CHANGE YOUR MODEL AND/OR STARTING VALUES."

I'm wondering if anybody out there has familiarity with running this type of variable/model. I have seen others do this before (e.g., Toker & Biron, 2012, Journal of Applied Psychology); however, I am having problems obtaining their syntax.
 Bengt O. Muthen posted on Friday, November 15, 2013 - 4:48 pm
Stoppage in iteration 1 is often caused by starting values that give an inadmissible covariance matrix.

Others?
 Zhi Li posted on Thursday, November 08, 2018 - 11:22 am
Hello Dr. Muthen,
I have encountered the exact same problem, do you think this is caused by a non-zero mean for the latent difference score? I am wondering how does XWITH work with creating interaction term for latent variables. If we were to manually create the interaction term, would it be possible?
Thanks!
 Bengt O. Muthen posted on Friday, November 09, 2018 - 1:12 pm
Q1: No.

Q2: No.

Send your output to Support along with your license number.
 Daniel Lee posted on Wednesday, December 26, 2018 - 10:29 am
Hi Dr. Muthen, can the mean of latent change factor scores be negative? I ask this simple question because while my descriptive statistics (observed means) show decreases in a certain construct over time, the latent change score factor means are statistically significant and positive. Thank you in advance for your thoughts!
 Bengt O. Muthen posted on Thursday, December 27, 2018 - 9:23 am
Factor means can be different from observed variable means due to the latter also being a function of the measurement intercepts.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: