Dear all,
The scale of one of my measured constructs is unidimensional with a mix of positive and negative items. The construct is reflective and is theoretically conceptualized (by the developer of the scale) as a positive construct; thus, the negative items should be reverse-coded.
My questions are:
(1) After running the PLS algorithm with all items of the construct included, I found that the loadings for reverse-coded items had negative signs. How should I interpret these values?
(2) I tried a different solution after reading several comments that I found in this forum. I ran factor analysis on these items. Results suggest that there are three components to the construct. As such, I modelled the construct in SmartPLS as having three components. That is;
First;
Component 1 --> Construct X
Component 2 --> Construct X
Component 3 --> Construct X
Then;
Construct X (with the three components) --> Construct Y
Interestingly, I found the loadings for all the items were well above the cut-off value of 0.70. Also, there's no negative loading. Although I am happy with the results, is this the right way of solving the issue pertaining to negative loadings of reverse-coded items in a unidimensional construct?
I humbly seek expert advise. Thank you.