How each question affect the latent variable

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
dlassen89
PLS Junior User
Posts: 2
Joined: Mon Jun 17, 2013 9:28 am
Real name and title:

How each question affect the latent variable

Post by dlassen89 »

Hi,

I have a question concerning SmartPLS.
'Satisfaction' and 'Loyalty' are my latent variables in my model.

I have made a model with these effects:
Image > Loyalty
Service > Satisfaction
Service > Loyalty
and so on..

But..
I have 3 questions concerning the category "Image".
E.g. "Do you find xxx innovative?" (The respondent uses a 7 point scale to answer this).
And I have another question concerning how creative the xxx is.

I want to know how much effect 'creative' and 'innovative' are concerning 'Image'. Which of the two dimensions (creative and innovative) should be my focus?
Image is to wide so I need to split the dimension in some parts so I can concentrate my focus on the section of Image that have the best effect on e.g. loyalty or satisfaction.

Please help.
What do I have to do? :)
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi Dorthe,

Usually our focus is on latent variables and path coefficients.

If you are using reflective indicators:
They could be replaced by others.
For this reason, you should think more in LV terms than indicators.

If you are using formative indicators, they could not be replaced by others.
And we could use the “outer weights” as measures of importance of the indicator.

But, if there are some multicollinearity between formative indicators, these outer weights are not interpretable (as importance). What is the case with reflective indicators.

Best regards,
Bido
Post Reply