Hi Raditiya,

I guess you have an identification issue due to the inclusion of the formative variable (with no indicators).

Unfortunately, I cannot really offer a solution other that I would avoid creating such models. Either you have theoretical assumption with regard to a causally operationg latent variable, then create measures (see Howell et al.) or you want to include a composite of the three latent variables, than create a composite or (third option), you regard the formative variable as a kind of umbrella term of the three variables than skip the latent varialbe as the effects emerge from the three singular variables.

Finally, you have many indicators per latent. I would recommend taking a close look to the measurement models and its fit. In my experience, running with 3-4 crystall clear items was the maximum what was realistic and possible.

Howell, R. D., Breivik, E., & Wilcox, J. B. (2007). Reconsidering formative measurement. Psychological Methods, 12(2), 205-218.

Best,

Holger

Hello Raditya,

Iteration limit adjustment for AMOS: http://amosdevelopment.com/webhelp/iterationlimit1.htm

This YouTube video offers advice on troubleshooting iteration limit issues in AMOS: https://www.youtube.com/watch?v=B7YOv7hSohY

As well, there may be an issue with your model. See this related RGate thread:

https://www.researchgate.net/post/Does_anyone_understand_AMOS_regarding_iteration_limit_reached_problem

Good luck with your work!

I think Holger has identified the potential problem, the formative latent variable. Ken Bollen has written extensively on this and in this paper doi:  [10.1037/a0024448] he shows the different ways that formative variables can be specified, and his 'composite' may be easier in this situation as it has no residual (but you then have to live with that assumption). I suspect that a 2nd order latent variable would be easier to model.

Mark

Thankyou all for the answer ! It seems that my problem have been solved ! Thanks