|
|
GGMM with categorical indicators of c |
|
Message/Author |
|
Anonymous posted on Thursday, May 24, 2001 - 8:54 am
|
|
|
I am investigating the development and heterogeneity of seven binary indicators. While the three class solution runs very well and also makes some sense, I have not been succesful to run a four or higher class solution. Here are my questions: 1) Do the same rules as in LCA analysis apply to determine the maximum number of classes? 2) On what scale are the mean and slope of the individual classes? 3) What does a negative mean of 10 mean for a specific class? 4) In order to draw the development in the different classes, does the same computational rule apply: mean + (slope * timescore) by class? Thanks for any input. |
|
|
1) Do the same rules as in LCA analysis apply to determine the maximum number of classes? The same rules would apply as a lower limit. You may be able to get more classes with a growth model. I'm not sure and don't know of any references to this. 2) On what scale are the mean and slope of the individual classes? Logit scale. 3) What does a negative mean of 10 mean for a specific class? A very low probability. 4) In order to draw the development in the different classes, does the same computational rule apply: mean + (slope * timescore) by class? Yes. |
|
Back to top |
|
|