Indirect reverse effect

[Olivier]

Hello everyone, I'm new to this forum and with lavaan...

I'd like to calculate an indirect effect that goes in the opposite direction of a regression. In my study, I have A influencing B and C. I have no direct effect of B on C. Does it work (from a statistical point of view and with lavaan) if I look for an indirect effect B-->A-->C. Thank you for your valuable feedback.

[Terrence Jorgensen]

I have A influencing B and C. I have no direct effect of B on C. Does it work (from a statistical point of view and with lavaan) if I look for an indirect effect B-->A-->C.

If that is the model you want to fit, then that is the model your syntax should specify.

https://lavaan.ugent.be/tutorial/mediation.html

Terrence D. Jorgensen    (he, him, his)
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
http://www.uva.nl/profile/t.d.jorgensen

[olirer...@gmail.com]

Thank you for your answer. What I'd like to check is whether there are any indirect effects in the opposite direction. In my model, I start with IMPUTICF, for example, to see if SEPCF plays a mediating role with CALCF. Can I do that? I don't know if I can benefit from your expertise. But do you have the impression that my model is ‘good’ or could something be missing?

 

 

model.SPE <- 'PERF=~DPERFC3+DPERFA3+DPERFB3

              SEPCF=~SEPB+SEPA+SEPC

              IMPUTICF=~IMPUTIA+IMPUTIB+IMPUTIC

              INTCF=~INT01+INT02+INT03

              CALCF=~DCALC3+DCALA3+DCALB3

 

             PERF~p1*CALCF+Duree+SEPCF+INTCF+IMPUTICF

             CALCF~Duree+p2*SEPCF+p3*INTCF+IMPUTICF

             SEPCF~Duree

             INTCF~Duree+p4*SEPCF

             IMPUTICF~Duree+p5*SEPCF

             INTCF~~IMPUTICF

             p1p2 := p1*p2

             p1p3 := p1*p3

             p4p2 := p4*p2

             p5p2 := p5*p2

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/80bfe2d7-e5e0-456a-b6eb-4b99ed4408e3n%40googlegroups.com.

[Edward Rigdon]

I don't see any "reverse direction" in your syntax. As I read it, the relations your propose are:

INTCF and IMPUTICE have residual covariance

Duree predicts SEPCF

Duree and SEPCF both predict INTCF and IMPUTICE

All four variables mentioned all predict CALCF

All five variables mentioned all predict PERF

If there is anything "wrong," it appears that your structural model is saturated. So you will not be able to assess your theoretical model based on the structural model here. Any lack of fit that you encounter will be due to relations between factors and observed variables, not between factors--or due to violations of assumptions, including distributional assumptions.

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/001f01db09a5%2442c6a4d0%24c853ee70%24%40gmail.com.