Message/Author 


Hi. I am planning to do some growth mixture modeling with adolescent longitudinal data in Korea. Is there any way to identify a certain subgroup which considers growth in two parallel dependent variables such as depression and delinquency? I am hoping to find a subgroup of adolescents who are both high in depression and deliquency over 2 years of assessment with 4 timepoints. Since I am just beginning to study mplus software, would you provide any tips to write an appropriate syntax for this kind of modeling? Thank you. 


You would need to start with a parallel process growth model like the one shown in Example 6.13 and add TYPE=MIXTURE; and one latent class variable using the CLASSES option of the VARIABLE command. For that see Example 8.1. 


Thank you very much for your kind help. I have one more question about my study. My study follows a cohortsequential design. I have transformed my data set and it looks like the one shown in page 405 of User's guide. For this data set, I would like to apply FIML for missing data. In order to do this, is it correct to simply write "TYPE=MIXTURE missing;"? Following your guidance, is it right to include below commands for my study? ========== VARIABLE : CLASSES = c (1) ; ANALYSIS : TYPE = MIXTURE missing ; MODEL : %OVERALL% i1 s1  y11@0 y12@1 y13@2 y14@3 ; i2 s2  y21@0 y22@1 y23@2 y24@3 ; s1 ON i2 ; s2 ON i1 ; i1 s1 ON X ; i2 s2 ON X ; c ON X ; ========== I really appreciate your guidance and help for this matter. 


This looks correct. I would however start with a model without covariates. 


Again, thank you for your tips! 


Dear Linda, I actually tried to run a growth mixture model with above syntax. And I started with a model without covariates. However, it seems like that I have a problem with convergence. I keep getting the same warning message as following: THE ESTIMATED COVARIANCE MATRIX FOR THE Y VARIABLES IN CLASS 1 COULD NOT BE INVERTED. PROBLEM INVOLVING VARIABLE B7. COMPUTATION COULD NOT BE COMPLETED IN ITERATION 2. 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. Could you please provide any tips to solve this kind of problem? I am VERY VERY new to Mplus, and I am having a tough time studying this on my own. If I want to specify starting values, what kind of procedure is needed? Thank you very much. 


You can read about assigning starting values in the user's guide. I suggest you send your input, data, output, and license number to support@statmodel.com. 


Dear Linda, Is there any way to exclude the default setting of correlation between slope growth factors in Example 6.13 of parallel processes? And can I see the correlation between intercept and slope in each process? (What is the correct syntax for this?) In a longitudinal data, one can usually assume that residuals of indicators are high correlated. In Example 6.13, how can I specify that residuals from y11 to y14 are correlated or same? Thank you very much for your help. 


You can fixed the covariance between the slope growth factors at zero as follows: s1 WITH s2@0; You can specify the covariance between the intercept and slope growth factors as follows: i1 WITH s1; Residual covariances are specified using the WITH option, for example, y11 WITH y12; Equalities of the residual covariances are specified as follows: y11 WITH y12y14 (1); y12 WITH y13y14 (1); y13y14 (1); Please see Chapter 16 of the user's guide for more information. 


Thank you very much for your prompt reply! 

Daniel Lee posted on Sunday, April 09, 2017  6:30 am



Hello, I am modeling a parallel growth model...when I run growth models without regressing intercepts and slopes between two growth constructs (e.g., s2 on i1 s1), the growth processes have significant slopes and intercepts. However, when I include "on" statements between intercepts and slopes (e.g., s2 on i1 s1), the slopes are no longer significant. I was wondering if you can help me understand why this might happen? Thank you. 


If you don't have ON statements, you are estimating means and variances for the growth factors. With ON statements, you estimate intercepts and residual variances. Check the output. I think this is what you will see. 

Back to top 