Multinomial regression with slope as ... PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Anonymous posted on Monday, July 04, 2005 - 2:44 am
I have estimtated a linear growth curve model and am now regressing a nominal variable on to the random slope. How do I interpret the direction of the probit coefficients if the mean of the slope growth factor is negative. That is to say, the mean of the slope paramter is -.04 and the probit regression of observed category 1 on to the slope growth factor is -5.178. Just in terms of the direction of this relationship, should I interpret it to mean that, as scores on the slope variable increase (i.e. become positive), the probability of being in category 1 of the dependent variable decrease? Thanks,

Patrick
 bmuthen posted on Monday, July 04, 2005 - 11:13 am
Yes, that's right.
 j guo posted on Thursday, December 01, 2016 - 2:18 am
Hi Bengt and Linda, I tried to use intercepts and random slopes (linear and quadratic) to predict outcomes based on a bivariate latent curve model. Given that there are several random slopes, it is not manageable to test them directly in the LCM because of computational burden. So I saved the factor scores of intercepts and slopes in the LCM model and then used them for subsequent regression analysis. it is legitimate to do that?

Results showed that quadratic slope but not intercept and linear slope had a significant effect on the outcome. I'm not sure how to interpret this kind of finding. Any thoughts?

Thanks!
 Bengt O. Muthen posted on Thursday, December 01, 2016 - 10:41 am
Using estimated factor scores only gives a rough idea. Estimated factor scores don't behave quite like true scores. See e.g. our FAQ

Factor scores

Predicting from growth factors can be difficult given that they are often highly correlated. See the paper on our website:

Muthén, B., Khoo, S.T., Francis, D. & Kim Boscardin, C. (2003). Analysis of reading skills development from Kindergarten through first grade: An application of growth mixture modeling to sequential processes. Multilevel Modeling: Methodological Advances, Issues, and Applications. S.R. Reise & N. Duan (Eds). Mahaw, NJ: Lawrence Erlbaum Associates, pp.71-89.
download paper contact first author show abstract
 j guo posted on Sunday, December 04, 2016 - 6:13 pm
Thanks for your quick response.

If I wan to present the correlation matrix among different growth factors based on multivariate LCM, which output should I use (unstandardized or standardized)?

Thanks again.
 Bengt O. Muthen posted on Monday, December 05, 2016 - 5:25 pm
Tech4 correlations.
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: