1

My task is to estimate the impact of job change and its characteristics (e.g. voluntariness) on well-being.

Well-being is measured at two time points, T1 and T2. Some respondents have changed jobs between T1 and T2, others have not. For those who have, additional information, such as job change voluntariness is available. It can be hypothesized that both job change itself and the degree of its voluntariness have an effect on well-being.

I wonder how to configure a model for this research question.Ideally, I would have a mediated SEM model:

Well-being at T1 -> Mobility (dummy) and, if mobility = 1, then a voluntariness measure -> Well-being at T1.

If it were a mere multiple regression of the well-being score at T2, I'd use a dummy for having changed jobs, and fill in a reference value (e.g. zero) of voluntariness for non-changers (as described here: Time spent in an activity as an independent variable), using both variables as predictors.

The difficulty is that both the change dummy and the voluntariness variable are influenced themselves by well-being at T1. People change jobs where they are not satisfied, after all. Omitting the link between well-being at T1 and job mobility is not feasible theoretically. Yet there is a difficulty of dealing with structural zeros for voluntariness, in cases with no job mobility.

What would be the best strategy to model this situation? Preferably within the SEM/MIMIC framework, but other suggestions are welcome as well.

Many thanks in advance for you help.

Kind regards, Maxim

Community
  • 1
Maxim.K
  • 560
  • 7
  • 20
  • I'd be tempted to try to do this as a latent class analysis. One class has zeroes - zero mean and variance, the other class is Poisson. – Jeremy Miles Nov 17 '14 at 16:40
  • I am not sure what you mean. Class membership is observed here and known a priori. Would you elaborate? – Maxim.K Nov 17 '14 at 16:56
  • In Mplus, you can use a knownclass statement. – Jeremy Miles Nov 17 '14 at 20:47
  • Would you care to post an answer on this one, preferably with an example of such model? The MODEL and ANALYSIS part in Mplus syntax would suffice. In my own understanding, running TYPE=MIXED with KNOWNCLASS will give me two separate models, which would preclude me from directly comparing the effects of having a job transition with its absence, as there is simply no data in the no-transition group in regard to that variable. – Maxim.K Nov 18 '14 at 08:52

0 Answers0