Dear PLS experts,
I have a question regarding discriminant validity.After checking my factor loadings (and deleting those that did not exceed 0.55) I came up with positive reliability results (composite reliability >0.7) and adequate convergent validity (AVE>0.5).
One problem left: discriminant validity => 2 of my LV's seem to correlate a bit too much (r= 0,77 < √Ave = 0,72 & 0,75).
Since it's only a slight difference, I decided to delete two indicators whose cross loadings are higher on the wrong LV.
I ended up with sufficient discriminant validity, but is this an accepted procedure?
Thanks in advance