I am using PLS-SEM for my studies. My questionnaire involve some adapted questions and few newly established questions as well. I have checked reliability and validity, factor loading, cross loading and discriminant validity, chronbach's alpha using HTMT: And have got all the values within required range.
Still multicollinearity is there in around 10 items from all 35. Is it okay to proceed ahead as other results are good ?
I have also conducted EFA before checking reliability and validity, then also not much difference in the results. For EFA and KMO test -Jamovi is used and then smart PLS for further analysis.
Kindly guide
It's probably not a bad idea to conduct EFA (or CFA) given that you have added some new or modified questions to your scale. Multicollinearity should not be a problem for factor analysis--in fact, the higher the observed correlations, the stronger the factors.
One question that I would have as a reviewer is why you'd want to switch to PLS after doing covariance-based (EFA) analysis in the first step. The more natural thing would be to run covariance-based SEM.
