Factor-based CFA & Model Fit and Quality Indices

[Joa H]

Dear Professor, 

I have a scale consisting of three sub-scales to validate. Among others I would like to perform a factor-based CFA using the factor-based algorithms in WarpPLS. I consider my scale to be a 2nd order reflective-reflective construct. 

1. I was thinking of using the two-step approach. But I don't know if I should stop at first step and report for internal consistency reliability, convergent validity and discriminant validity. Should I also report additional Model Fit and Quality Indices at this step? May I compare different models based on additional Model Fit and Quality Indices at this stage?
2. Or should I go to the second step (but I have no direct links here as my model is only one construct)? If so, what should be reported then? May I also report additional Model Fit and Quality Indices at this step?
3. Alternatively, I can use the repeated indicator approach, but I'm not sure how to technically do this in WarpPls, what parameters should I report and whether it would be a better approach? What about additional Model Fit and Quality Indices in this case? Do you have an example/video of how to run and apply the repeated indicators in WarpPls? 

Could you please help me with my questions.

Br,

Joanna

[Ned Kock]

Hi Joanna. Have you seen this one?

http://youtu.be/bkO6YoRK8Zg

[Joa H]

Thank you Professor, 

I know this video and the procedure but my questions still remain. In particular because I am testing only one construct which is 2nd order reflective-reflective construct (I have no direct links here as my model is only one construct). In addition to factor-based CFA may I also report Model Fit and Quality Indices at first step (first order)? May I compare different models based on additional Model Fit and Quality Indices at this first step? Or should I go to the second step (second order)?

Joanna

[Ned Kock]

It really depends on what the committee, review panel, etc. is asking you to do. I would start with doing tests and reporting them in a very targeted way (see pubs. below), based on the theory you are testing, and then take advantage of the many features of WarpPLS to address follow up requests. You can always use fit indices to make a case for one model or another, at any level, or to justify the use of a particular algorithm. WarpPLS allows for so many tests and generates so many coefficients and graphs, arguably more than any other existing SEM software, that you can address virtually any request.

Kock, N. (2020). Using indicator correlation fit indices in PLS-SEM: Selecting the algorithm with the best fit. Data Analysis Perspectives Journal, 1(4), 1-4.

Kock, N. (2014). Advanced mediating effects tests, multi-group analyses, and measurement model assessments in PLS-based SEM. International Journal of e-Collaboration, 10(3), 1-13.

[Joa H]

Thank you! This is very helpful. 

Joanna