Table 3.
Revised structural model and fit assessment.
| Panel A. Explained variance | |||
|---|---|---|---|
| Outcome | R2 | Adjusted R2 | 95% percentile CI for R2 |
| EEX | 0.2478 | 0.2443 | [0.1970, 0.3108] |
| EFD | 0.2616 | 0.2570 | [0.2124, 0.3236] |
| PC | 0.3924 | 0.3876 | [0.3424, 0.4540] |
| Panel B. Conditional direct associations | |||||
|---|---|---|---|---|---|
| Association | Beta | Bootstrap SD | p | 95% percentile CI | Inner VIF |
| EEX → EFD | 0.0257 | 0.0375 | 0.494 | [−0.0476, 0.0999] | 1.3294 |
| EEX → PC | 0.2345 | 0.0344 | <0.001 | [0.1681, 0.3024] | 1.3303 |
| EFD → PC | 0.1888 | 0.0372 | <0.001 | [0.1152, 0.2609] | 1.3542 |
| RAC → EEX | 0.0706 | 0.0425 | 0.097 | [−0.0147, 0.1521] | 1.5401 |
| RAC → EFD | 0.1636 | 0.0424 | <0.001 | [0.0797, 0.2464] | 1.5467 |
| RAC → PC | 0.2423 | 0.0387 | <0.001 | [0.1671, 0.3184] | 1.5829 |
| TAC → EEX | 0.0752 | 0.0400 | 0.060 | [−0.0021, 0.1559] | 1.4260 |
| TAC → EFD | 0.2410 | 0.0402 | <0.001 | [0.1621, 0.3175] | 1.4336 |
| TAC → PC | 0.3392 | 0.0372 | <0.001 | [0.2652, 0.4128] | 1.5122 |
| TRC → EEX | 0.4115 | 0.0406 | <0.001 | [0.3314, 0.4890] | 1.6122 |
| TRC → EFD | 0.2045 | 0.0452 | <0.001 | [0.1172, 0.2943] | 1.8373 |
| TRC → PC | −0.1834 | 0.0429 | <0.001 | [−0.2686, −0.1003] | 1.8939 |
| Panel C. Approximate and bootstrap fit evidence | |||||
|---|---|---|---|---|---|
| Model | Criterion | Original value | HI95 | HI99 | Comparison |
| Saturated | SRMR | 0.049243 | 0.041823 | 0.042769 | Above HI99 |
| Estimated | SRMR | 0.049243 | 0.041823 | 0.042769 | Above HI99 |
| Saturated | d_ULS | 0.727447 | 0.524749 | 0.548747 | Above HI99 |
| Estimated | d_ULS | 0.727447 | 0.524749 | 0.548747 | Above HI99 |
| Saturated | d_G | 0.219479 | 0.219742 | 0.225452 | Below HI95 |
| Estimated | d_G | 0.219479 | 0.219742 | 0.225452 | Below HI95 |
Native ordinary PLS, N = 650, 10,000 saved bootstrap parameter draws. Parameter intervals are two-tailed 95% percentile intervals. HI95 and HI99 are one-sided reference quantiles from the adapted Bollen–Stine model-fit bootstrap, not parameter confidence intervals. Approximate SRMR below 0.08 coexists with SRMR and d_ULS values above HI99; the fit diagnostics do not uniformly support the specification. No causal interpretation is assigned to the software path directions.