Moderating effects in structural equation models


New Member

first, I want to apologize if I'm opening this thread in the wrong folder etc. I just registered here and I'm a bit confused still.

I'm having trouble interpreting moderation effects in a structural equation model.
We're regressing antisocial behaviour (four latent factors) on childhood maltreatment (also 4 latent factors) under the assumption that a certain genotype (with two levels) has a moderating effect. To do this, we first estimated the model via mPlus and then took the standardized regression coefficients for the genotype1 vs. genotype2 group and checked for differences using Cohen's q.

Now I've got two questions:
1. In almost alle cases, the standardized regression weights are not significant, but the moderator effect is significant. It confuses me to think that although the single regressions are not significant, their difference is!?

2. (and more important): how to interpret the outcome? Say, I have a standardized regression weight from maltreatment factor 1 on antisocial factor 1 of .32 for genotype1 and -.19 for genotype2 (both not significant) and a significant moderator effect q of .52. Can I say something like: "In the genotype1 group, there was a positive relation between maltreatment and later antisocial behaviour, whereas in the genotype2 group, children who had been abused showed less antisocial behaviour in later life. The large significant moderating effect shows that in this case, the genotype functions as a moderating variable".

I'd be happy if anyone could help me in this matter.
Thank you so much!