What type of Analysis for longitudinal moderation?

[gebaunco]

Hello everyone,

I am currently working on a project where I have a data set with measurements taken place at two time points (longitudinal). I got two variables, lets call them a and b, both being continuous scaled, and i got four more variables, lets call them 1,2,3,4 also all continuous scaled. I am interested in gender effects, so now I I want to check if there is a difference for a & b for gender (male/female). Then I want to check if there is a difference for 1,2,3,4 for gender. In the end, I would like to see if gender is influencing a &b, which in turn is influencing 1,2,3,4.

What I am thinking is, I could do structural equation modeling, including all the variables and then do two groups, one for each gender and compare those two. However I am unsure if that is valid. Moreover, I am super new to SEM so not even sure if its the right approach.

I am thankful for any hints or advice

 

[Karabiner]

This is a bit confusing. There are some ideas which indicate moderator analysis, other ideas refer to mediation.
What is the dependent variable in your model, are these a and b? Were all variables measured twice?
What exactly do you mean by "if there is a difference for a & b for gender (male/female) ... if gender is
influencing a & b" if a and b generally differ by gender? If the difference of a and b between time points
differs by gender? Could you maybe describe your theory and the real variables instead of a b 1 2 3 4 ?

With kind regards

Karabiner

 

[gebaunco]

Yeah I am sorry, I wanted to keep it simple but ended up making it confusing..

So:
All Variables were measured twice.
It is about Networking, it was asked if people do network or not and if they do, certain characterstics of networks of the people were assessed: the size of the network AND the amount of men in the network.
Certain benefits of networking were assessed: amount of promotions, wage, social support, and informational advantage.

So I want to see if the network characteristics (Size&men) of men and women differ. Moreover I want to check if the effectiveness of networking (assessed by the four benefits of networking), compared to people not networking, is moderated by gender, so if men and women differ on those variables.

In generel I would say paired t-test for the question about the network characteristics, but what about the effectiveness?
I hope it is a bit more understandable now

 

[Karabiner]

In generel I would say paired t-test for the question about the network characteristics, but what about the effectiveness?

The paired t-test will deal with whether size & amount changed over time.

Does the repeated measurement play an important role here? Your research questions do not include that aspect.
And, are there cases who were not networking at t1, but at t2?

One approach could be repeated measures analysis of variance with the additional grouping variable "gender".
But I am still not sure if I understand the research questions correctly.

With kind regards

Karabiner