Page 1 of 1

moderator 3 continuos scales

Posted: Thu May 18, 2017 3:36 pm
by kamellia.ch
Hi. I am PhD student with the 300 samples thesis. I have a question appreciate if you guide me :

I have one moderator (Z) which has three continuos dimensions/scales Z1,Z2,Z3 ( each dimension has 8,9,8 indicator respectively) , if I want to see the moderating effect of Z between X and Y , what should I do? In book I read about one continuos moderator , not 3 . should I take them as second order? ( I want to see the moderator effect of z, z1,z2,z3 between x and y) . can anybody please guide me or send me any example of Thesis with Smart PLS on moderating with more than one continuos scale?

appreciate your guidance

kamelia
email :kamellia.ch@gmail.com

Re: moderator 3 continuos scales

Posted: Thu May 18, 2017 3:42 pm
by cringle
In the 2nd edition of the PLS-SEM book
https://www.smartpls.com/documentation/ ... on-pls-sem

We use the switching cost construct with four indicators as a moderator.

Hope that this helps.

Best
CR

Re: moderator 3 continuos scales

Posted: Fri May 19, 2017 6:06 am
by kamellia.ch
thanks Dr.Ringle for kind reply
Actually I read your new edition book.switching cost include only four indicators and used as continuos moderator.
in my model, moderator is "commitment" which has three dimensions ( affective commitment, normative commitment & continuos commitment). each dimension at least have (7-8 indicators).means together around 23 indicators. I want to see the moderating effect of each dimension separately. and also the moderating effect of the whole commitment in general. this is the problem since there are three different continuos commitment with different indicators,I am confused. should I take commitment as second order ? or should I investigate each dimension separately using different path model?

really appreciate you kind guidance
kamelia

Re: moderator 3 continuos scales

Posted: Fri May 19, 2017 8:47 am
by cringle
Hi, the 2nd order approach sounds what you like to do. But what is such a multidimensional construct commitment? Where would a potential moderation come from? I am very careful when it comes to using 2nd order constructs. Personally, I would rather look on the moderation of each 1st order construct.

Best regards
Christian

Re: moderator 3 continuos scales

Posted: Fri May 19, 2017 10:58 am
by kamellia.ch
I will use your kind advice , thanks Prof. Ringle

Re: moderator 3 continuos scales

Posted: Sun May 21, 2017 7:43 am
by kamellia.ch
Dear Prof Ringle,
may I ask for more guidance?
I used your advice and look on the moderation of each 1st order construct instead of making one second order, the question is should I bring all these first order together? I attached the picture for you after bringing all first order construct together and getting bootstrapping result. result showed only one moderator is significant (SN*AC) . but when I do bootstrapping separately one by one for each first order construct the result will be different and non of them is significant. which result is acceptable?
appreciate your kind guidance

Re: moderator 3 continuos scales

Posted: Sun May 21, 2017 7:50 pm
by cringle
It really depends on what you would like to analyze and like to accomplish. If your think that all interactions at the same time make sense, you may go for that analysis. But what does this multiple moderation mean? Does it really make sense? What about a moderation of the moderation (3-way, 4-way, and 5 way interactions)? Would they make sense as well and can you explain them.

We are usually very "conservative" in using moderating effects (e.g., one reasonable one).

Technically things are possible either way.

Best
CR

Re: moderator 3 continuos scales

Posted: Mon May 22, 2017 3:08 am
by kamellia.ch
thank you for kind reply.

about your question, these moderations are three different type of commitment. An individual can show different type of commitment to the organisation at the same time. thats the reason, I couldn't remove any of them and try bringing together at the sam time . I was just worried because the model became so complicated.

your reply solved the confusion.really appreciate it.