Moderation and Mediation analysis

[shahid]

I need urgent assistance  in calculation of mediation and moderation analysis methods. How to study the effect of  both  moderating  and mediating variables (SEM) in Structural model. Sir  i am not able  to  understand how to calculate the interaction effects  of Predictor X Moderator variables.In  most of the studies Aiken and West (1991) approach  has been followed  but  i am not able to understand that approach. Please suggest me  some method  so that i will proceed.

 

Regards

Shahid Mushtaq

[Jeremy Miles]

If you are using structural equation modeling, this is potentially difficult, but it depends a great deal on the precise nature of the problem, and the software that is being used.

 

Can you tell us more?

[shahid]

Sir i am using SEM capabilities of LISREL 8.50. Moreover in my study nine independent variables, one mediating variable, four moderating variables and one dependent variable is there. Plz suggest me some method  to  proceed.

 

 

 

[Jeremy Miles]

I’m not sufficiently familiar with the capabilities of LISREL to answer this very clearly.  But even if I was, I’d need more information.  A moderated relationship is one where:

 

Y = A + B + A * B

 

So when you have four moderating variables, do you mean that these four moderate each other (as A and B do above).  Or is it more like 

Y = A + B + C + D + E + A * B + A * C + A * D + A * E

 

where A is the independent variable, and B, C, D and E are moderators.

 

i would also suggest that four moderators was far too many.

 

Are these moderators latent or observed? Are they categorical or continuous?

 

Jeremy

 

[shahid]

In simple  terms i will like to answer ur answer.Below mentioned is  the simple model. Now in the same model if i want to test the interaction effect of any moderator variable hw  to study the same.

A (Independent)——————–M (Mediating Variable)……………..B (Dependent)

[Judith Terstriep]

As you are using LISREL I assume you are only working with reflective messurement models.

Based on this assumption the moderating effect can be calculated as a production term of each indicator of the predictor variable and each indicator of the moderator variable.

I hope this information is of help.

[Author]

Posts