Status change model PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
 Claire Noel-Miller posted on Monday, May 17, 2010 - 10:56 am
Bengt, I am applying a 2-groups (males and females) version the status change model described in “The Influence of Changes in Marital Status on Developmental
Trajectories of Alcohol Use in Young Adults” (Curran and colleagues, 1998). I have 2 clarification questions:
1-- In your analysis, all of the status change coefficients were significant and of comparable magnitude. How would one interpret model results where only some (one out of 3 in my case) of the status change coefficients are significant?
2-- Regarding interactions: while I understand the intuition behind testing against the grand mean effect model as described in the paper, why not simply test the 2-group unconstrained model (model 2 in the paper) against a model that imposes equality on the 3 intercept terms across groups?
 Bengt O. Muthen posted on Tuesday, May 18, 2010 - 8:01 am
I would say that status change effect is age-specific.

Your approach would be fine.
 Claire Noel-Miller posted on Saturday, May 22, 2010 - 8:13 pm
Thank you for your answer. I have a couple more questions:
1--In discussing possible extensions to this model, the paper states "the status change variables can be cumulative (e.g. once married always married)". What would be the interpretation of the coefficients on such a variable?
2-- I am considering an extension to this model which would use the duration (in months) since the status change occurred rather than the status change variable itself. I am interested in seeing whether the effect of status change on the added intercept diminishes with time since event occurrence. Would this be an appropriate way to do this?
Thank you for your help.
 Bengt O. Muthen posted on Sunday, May 23, 2010 - 11:18 am
1. The degree of permanent level (intercept) change.

2. Seems like this could be handled by letting the loadings of the latent intercept be estimated rather than all fixed at 1.
 Katy Roche posted on Friday, February 07, 2014 - 7:18 am
I am attempting to test a model similar to Curran, Muthen & Harford (1998).

Can you tell me how to develop syntax to create a latent intercept factor representing change in status at each time point?

This is the syntax I'm currently using for this model but I am not sure it gets at what Curran et al. do in their article. I want to know how change in school status at each time point is associated with change in depressive symptoms at each time.

I S |dw1@0 dw2@1 dw3@2;

i on community contagew1 hhsingm hhkin hhoth hhnoad poverty male ;

s on community contagew1 hhsingm hhkin hhoth hhnoad poverty male ;

dw1 on T1SCH;
dw2 on T2SCH T1SCH;
dw3 on T3SCH T2SCH T1SCH ;

!T1sch = time 1 in school vs. not
!T2sch = time 2 in school vs. not
!T3sch = time 3 in school vs. not

!dw1 = depressive time 1
!dw2 = depressive time 2
!dw3 = depressive time 3

Finally - how does one interpret the coefficients for, say, "T3SCH" from the "dw3 on T3sch T2sch T1sch" command.
 Bengt O. Muthen posted on Friday, February 07, 2014 - 2:16 pm
Look at slides 157-159 of the Topic 3 handout from our short courses which is on our website.
 Katy Roche posted on Friday, February 07, 2014 - 2:30 pm
Thank you for the reply.

When I looked at the slides, I was unable to figure out the syntax. I then listened to the video by my syntax for By and On statements that are described are giving me error messages.
 Bengt O. Muthen posted on Friday, February 07, 2014 - 4:09 pm
Please send your output and data to Support.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message