Hi, I am running a few GSEM regressions at the firm-level, wanting to check whether X1, X2, X3, X4 have a positive effect on Y, a binary variable.
Both Xs and Y are likely to be vary with firm sector. Therefore, I believe I have two options:
1. using sector as a control variable,
2. using a latent control variable which varies at the sector level to add a random effect for sector to both Y and collaboration Xs (as suggested in http://www.gllamm.org/PMmsem_04.pdf)
I am not exactly sure about how "option 2" works: the coefficients are set to 1 by STATA, so you cannot discuss how different sectors impact on X and Y, and I am not sure on how this latent variable impacts on the relationship between X and Y (standard errors?).
Can you please give me a piece of advice?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling