Hi, community. I have read a lot of threads of the forum and benefit a lot from your ideas; however, i am still confusing about this question.
If i have 3 1st order LV (a1, a2, a3) comprising a 2nd order LV (b), and this 2nd order LV has causal relationship with endogenous LV (c).
a1--\
a2---b-->c
a3--/
Is it correct to,
(1) use repeated indicator approach to test measurement model?
(2) treat the principle component analysis score for each 1st order LV as indicator for 2nd order LV and then test the structural model? (i.e., find a1--b; a2--b; a3--b; b--c; are all with significant path coefficient?)
Thanks in advance:)
Dayu