The simplest SEM you can create, but its real power lies in expanding on that regression model.  Here I will suggest four ways to do that.

Hi Batoul,

generally, it is not wrong to approach a SEM from an exploratory perspective but I would refrain from using stepwise regression. Further, if you do exploratory work you should not simply treat the variables resulting from the exploration as if they were theoretically based. But an explored possible causal structure can be a very good starting point for future, enhanced analyses.

Perhaps a completely different road would be to use the freeware software TETRAD which explores all models belonging to the same equivalence class (i.e., models with the same data implications)

Here is a nice 2h workshop video by Richard Scheines (one of the founders of TETRAD) and further references.

https://bit.ly/2dyMTAV

Glymour, C. (2004). The automation of discovery. Daedalus, 133(1), 69-77. doi:10.1162/001152604772746710

Glymour, C., Scheines, R., & Spirtes, P. (1988). Exploring causal structure with the TETRAD program. Sociological Methodology, 18, 411-448.

la Bastide-van Gemert, S., Stolk, R. P., van den Heuvel, E. R., & Fidler, V. (2014). Causal inference algorithms can be useful in life course epidemiology. Journal of clinical epidemiology, 67(2), 190-198. doi:10.1016/j.jclinepi.2013.07.019

Landsheer, J. a. (2010). The specification of causal models with Tetrad IV: A review. Structural Equation Modeling: A Multidisciplinary Journal, 17(4), 703-711. doi:10.1080/10705511.2010.510074

Best,

Holger

Ajit kumar Roy Thanks for your suggestion