Interaction effect - 2-stage approach

[j_terstriep]

Dear all,

in order to calculate latent variable scores for the 1st stage of the procedure to determine moderating effects do I use

(1) the full structural model or
(2) only the latent endogenous, moderting variable and the latent exogenous variable?

Many thanks in advance for your support.

Best regards,
Judith

[Hengkov]

Hi Judith,
Two-Stage Approach for complex model with multiple exogenous and endogenous variabel => stage 1: estimate only IV, MV and DV (main effect moderation). Stage 2 => calculate effect interaction for input full structural model.
Regards,
Hengky

[j_terstriep]

Dear Hengky,

many thanks for your reply.

So is correct to only us the interaction variables in the full structural model, and not the original IV and MV?

Best regards,
Judith

[Hengkov]

Hi Judith,
The moderator variabel in your model just 1 or more?
You construct IV or MV is formative?
For structural model effect interaction two-stage, not original IV and MV but used LVs.
Regards,
Hengky

[j_terstriep]

Hi Judith,
The moderator variabel in your model just 1 or more?
You construct IV or MV is formative?
For structural model effect interaction two-stage, not original IV and MV but used LVs.
Regards,
Hengky

Hi Hengky,

I have to moderator variables the first is absoprtive capacity the second is interaction frequency with external partners. It is assumed that firm's absorptive capacity moderates cluster internal interaction frequency (CLA) -> innovative performance (IPER) and also external interactions (EXA) moderate the this relationship.

Here are the related images of the main model and the model with interaction effects:





Best regards,
Judith
[/img]

[Hengkov]

Hi Judith,
I'm look your model above all construct is reflective. Two-Stage Approach not appropriate procedure for test this interaction, but used Product Indicators Approach. You create interaction terms using SmartPLS moderation function and run algorithm PLS => evaluation outer model, next run bootstrapping procedure => evaluation inner model. For addition, calculate effect size interaction with formula = R2 with moderator - R2 without moderator : 1 - R2 with moderator.
Regards,
Hengky

[j_terstriep]

Hi Hengky,

Many thanks for your help!

Best regards
Judith

[Hengkov]

Ok Judith,
Your welcome and good luck
Regards,
Hengky