Yuelin Li posted on Thursday, January 11, 2007 - 8:05 am
I am new to Mplus.
I'm trying to teach myself how to fit the 2 models in Muthen, Brown, and colleagues (2002, in Biostatistics). The app11.html example, found on www.statmodel.com/examples/penn/app11.html), appears to fit model 1. But I can't find the Mplus syntax for model 2, in which the 4-class profiles are fitted to the binary distal outcome of juvenile court record. Help is greatly appreciated.
I'm also trying to learn how to run GGMM, and I've benefited a lot from Chapter 8 of Duncan, Duncan, & Styker (2006). But in their example input 8.3 and in UG ex 8.6, I don't understand how including (say, from 8.6) the "categorical = u" statement results in regression of U on C as shown in UG Figure 8.6, puts U in the model, where U is a distal outcome. Why isn't there a model statement, like "U on C#1"
Probably it would take too much time to answer this way, but is there some place in the manual (I have looked but not found it) or another of your publications that would explain the syntax and/or defaults for distal outcomes in GGMM?
Follow-up question because my earlier question was pretty convoluted: How does the program "know" what to do with a variable (categorical or quantitative) in a model, just because it is listed (in the case of UG ex8.6) in the VARIABLE: NAMES ARE statemtents? Or in other cases, perhaps, in the USEVARIABLES statement? It is not in the MODEL: statement anywhere.
Does that mean that any (all?) variables included in the VARIABLES: NAMES ARE statement or USEVARIABLES statements will be treated as distal outcomes in a mixture model?
All variables on the USEVARIABLES or the NAMES statement if there is no USEVARIABLES statement are included in the analysis. If there are no BY, ON, or WITH statements associated with them, means/thresholds and variances if appropriate are estimated for the.