Hello,
I am carrying out SEM analysis with a model including two Independent variable (A,B) one mediator (C) and two dependent variable (D,E).
For reasons of statistical control and incremental validity I would like to understand if my variables' statistical contribution remains statistically significant after controlling for other predictors. In plain words my dependent variables (D,E) can be predicted also by other variables (F, G, H, I) which however are not the main focus of my study but that according to theory and research have been found to predict significantly my dependent variable. May I consider those other predictors (F, G, H, I) as control variables since they are not he focus of my study? In the model, should they point directly to the dependent variable?
Thank you!
Eleonora,
you are correct in implying (at least in my reading) that the difference between independent and control variables is purely conceptual. Statistically, they are the same thing, such that the path coefficient from one variable (say, A) refers to the residual variance of the other variable (say, D) after controlling for the control variables.
Maybe these papers will be helpful:
Article Statistical control in correlational studies: 10 essential r...
Article A Critical Review and Best-Practice Recommendations for Cont...
Hi Eleonora,
It's preferred to support your assumptions from previous research. Further, if you're to follow an SEM approach, you could test the model invariance regarding those variables.
See: Article When empathy hurts: Modelling university students’ word of m...
Article Exploring Personal Experience of Wartime Crisis Effects on J...
Thanks,
A
Hello both,
I would llike to thank you for your answers and suggested alticles.
E
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling