In SmartPLS, I used a reflective-reflective model. In the first order model assessment, the reliability is fine. And the validity is fine. My validity assessment included cross loadings, Fornell-Larker ctireria, and HTMT. Even HTMT at this stage was below 0.85. But in the second order model assessment, some of HTMT values were above 0.9. How can I solve this issue, please?