Kim & Muthén (2009). Two-Part Factor Mixture Modeling: Application to an Aggressive Behavior Measurement Instrument. Structural Equation Modeling: A Multidisciplinary Journal, 16, 602—624.
I’m encountering difficulty with understanding the modelling of the continuous component of the data.
I notice on page 605 you indicate that “…the first part of the model separated “no-use” from “any sort of use” by creating binary indicator variables that reveal any level of use within the previous time. In the second part of the model, continuous indicator variables represent the amount of the usage if “use” occurred. If the binary variable records ‘no use’ (0) then the continuous indicator variable that represents frequency of use was treated as missing in your article. That much I understand and it must have obviously created a lot of missing data on the continuous indicator variable in your study. I note that you use ML estimation, so this must mean that when the binary variable was zero, and hence the continuous indicator variable was missing, then this type of missingness on the continuious variable must have been handled via "FIML" under MAR even though the binary variable indicated that there was no frequency of the behavior – in effect you were estimating what the frequency of the behavior might have been had the behavior occurred?