Lukas Frei posted on Wednesday, February 03, 2016 - 12:58 am
Exploring data from a quasi-experimental study, we want to investigate effects of general participation in a program (dichotomous) aswell as intensity of participation (continuous). More specifically, we wonder if effects of general participation are moderated by intensity of participation. Since all non-participants hold an intensity-value of zero (3/4th of the sample), we wonder if there is a way to investigate our research-question without splitting the sample. Is it legitimate to simply include both variables as predictors? Or does Mplus offer advanced options similar to two-part modeling but in regard to predictors?
You can try to do a 2-group analysis based on the dichotomous variable. The intensity variable is then a covariate that is missing in one group. Bring this covariate into the model by mentioning its variance so that subjects with missing are not deleted. This is a research question and I haven't tried this out.