|
Message/Author |
|
|
As i understand, in mixture modeling with continuous variables, the variances are by default constrained to be equal across classes (and can be freed by mentioning them in the model). Hence, the classes should be equally homogenous with each other. Is there any similar type of default assumption when doing mixture modeling with ordered categorical indicators (using the 'CATEGORICAL = command')? |
|
|
There are no variances with categorocal outcomes so this is not relevant. Thresholds are not held equal across class as the default. |
|
Back to top |
|
|