Query : Disjoint 2 stage Approach

Frequently asked questions about PLS path modeling.
Post Reply
navya66
PLS Junior User
Posts: 5
Joined: Thu Jul 09, 2020 1:36 pm
Real name and title: Navya Jain , Research Scholar

Query : Disjoint 2 stage Approach

Post by navya66 »

Dear experts,

Please clarify if the following procedure is correct:

1. I have 4 variables of study. In Stage 1, I have used lower order constructs, established the measurement model, saved the latent variable scores. One set of predictors in my model can be used as a higher order or as separate lower order constructs as they are examining identity but very different aspects of it, the authors of the scale themselves outline that an overall score or summative score is not advisable. All other 3 constructs which are to be predicted by these identity constructs are well established higher order constructs.
2. In Stage 2, I should be using the LVS of the identity scales obtained from Stage1 to either formulate overall identity scale and proceed or can I use the LVS scores obtained to create separate identity scales (in order to stay in line with the authors' recommendation of not using an overall score). This means the identity scales measuring different parts of identity will have 1 item as loading (the LVS score). Since the measurement model was already established-CR,AVE were satisfactory and HTMT (0.90) criterion was met, would be it the right approach?

Regards,

Navya
jmbecker
SmartPLS Developer
Posts: 1129
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: Query : Disjoint 2 stage Approach

Post by jmbecker »

I do not fully understand the problem. It also seems to be more like a decision that has to be made by the researcher than a technical issue.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
navya66
PLS Junior User
Posts: 5
Joined: Thu Jul 09, 2020 1:36 pm
Real name and title: Navya Jain , Research Scholar

Re: Query : Disjoint 2 stage Approach

Post by navya66 »

Dear Sir,
My query is on this point: "can I use the LVS scores obtained to create separate identity scales (in order to stay in line with the authors' recommendation of not using an overall score). This means the identity scales measuring different parts of identity will have 1 item as loading (the LVS score)"

I'm confused about whether or not this is statistically acceptable.

Regards,

Navya
jmbecker
SmartPLS Developer
Posts: 1129
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: Query : Disjoint 2 stage Approach

Post by jmbecker »

If you only want to model them as first-order constructs then there is not need to use the two-stage approach.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
navya66
PLS Junior User
Posts: 5
Joined: Thu Jul 09, 2020 1:36 pm
Real name and title: Navya Jain , Research Scholar

Re: Query : Disjoint 2 stage Approach

Post by navya66 »

Dear Sir,

Apologies for the late response.
The reason I want to use Disjoint Analysis is because all my other variables are higher order constructs.

In the sample model diagram: ID & ED represent dimensions of identity (authors recommend not to use overall identity and each dimension may have a very different effect hence I don't want to model them as a higher order).

Please could you let me know if my thought process is correct in how I can model from Stage 1 to Stage 2 as depicted in the diagram. My focus is on the prediction of Y.
Attachments
Sample_Model_Query_1.jpg
Sample_Model_Query_1.jpg (197.88 KiB) Viewed 6034 times
Post Reply