Good morning,
based on a qualitative research approach I have developed a critical success factor (CSF) model, operationalized it via a questionnaire based on 1...5 Likert scale for formative indicators and reflective result estimation, and administered a survey which resulted in 77 entries to relate CSF to the success of the respective projects.
Based on PLS I estimated the model and verified its quality indicators. Can the results (either factor loadings or weights) be used to predict results of additional projects (According to literature this is one of the advantages of PLS) or is this approach fundamentally flawed? I have not seen this in other (ISR) papers.
As for the use of weights or loading. Using the weights from a standardized PLS calculation and applying it to raw Likert data results end up in about the Likert range for success, but some CSF seem to have a negative impact (which cannot be accepted based on the qualitative research). Using the (all positive) loadings on raw Likert scale data results leads to success estimates out of range (too high in case of weighted sum and too low in case of weighted average). Where is my approach flawed?
Any comments are appreciated!
Rgds.
GS