Including Control variables / Covariates in SEM

[lavaanrookie99]

Hi there,

I am planning to conduct a SEM analysis, where instructional quality will serve as the predictor, and different dimensions of well-being will be the outcomes. Since evidence suggests that socioeconomic status (SES), age, and gender are associated with both instructional quality and well-being, I plan to include these variables as covariates in the model.

My Question:

When including control variables in SEM, should these covariates be regressed on both the predictor variables (instructional quality) and the outcomes (well-being), or only on the outcomes?

Thanks in advance!