Treat each sub variable as a moderator on the same x in successive regressions assuming the composite variable is a moderator. This should work I believe. D. Booth

You should consider your objectives based on the theoretical model that supports the moderation hypotheses, although mostly you use the general score of the variable if you have support for the variables also integrate it as part of the objectives of your research and report in a general way and by dimensions.

You need to find both. best use PLS. which will give you a moderating effect with all independent and dependent variable.

Dear Cristian Ramos-Vera,

Thanks for answering. Could you please elaborate further on the general score of the variable? It isn't clear to me.

If I have 5 separate variables and each one has a different score, how should I look for a general score and its demonstration in tabular form? Please guide!

Regards,

Dear Prof. David Eugene Booth,

Thanks for your answer. I hope it works.

Further, considering each one as separate, will I use a graph/slope for each moderator to demonstrate the effect from high to low on (X-Y)? Please guide!

Regards,

Dear Mahadi Hasan Miraz,

Thanks for the new recommendation. Could be please elaborate further about graphs in SmartPLS? I guess PLS does not provide results for moderation levels (Low-Mean-High) as Process Macro does. Correct me if I am wrong.

Regards,

Remember moderation and interaction are the same things. Look at 3-way interactions in a factorial design here (in regression form): Design and Analysis of Experiments, 9th Edition | Douglas C. Montgomery (Author) | download (b-ok.cc)

Best D. Booth

In addition, See this link: https://www.google.com/search?client=firefox-b-1-d&ei=SarOX_jRGsSGtQbu-rPgBA&q=moderation+mediation+virginia+research+R&oq=moderation+mediation+virginia+research+R&gs_lcp=CgZwc3ktYWIQAzIHCCEQChCgATIHCCEQChCgATIHCCEQChCgATIFCCEQqwIyBQghEKsCMgUIIRCrAjoECAAQRzoICCEQFhAdEB5QrZ4CWPyyAmDauwJoAHACeACAAacBiAGoApIBAzAuMpgBAKABAaoBB2d3cy13aXrIAQjAAQE&sclient=psy-ab&ved=0ahUKEwi4mvyu87ztAhVEQ80KHW79DEwQ4dUDCAw&uact=5

David Booth

Use PLS, it will help.

Before proceeding to mediation analysis, you may try to correlate your IV (moderator variable) in general stage if it correlates significantly with the DV. If it does not, try to correlate each individual dimensions to the DV - for to have a lead which dimensions are worthy as moderator variables to farther analysis.