Table 4.
Correlation between self‐report and different versions of the racial/ethnic probabilities and percent error reduction of original probabilities during each stage
| MBISG 1.0 | End of phase 1 | MBISG 2.0 | MBISG 2.0‐End of phase 1 | MBISG 2.0‐MBISG 1.0 | % of MBISG 1.0 error reduced by phase 1 | % of phase 1 error reduced by MBISG 2.0 | % of MBISG 1.0 error reduced by MBISG 2.0 | |
|---|---|---|---|---|---|---|---|---|
| White | 0.85507 | 0.88701 | 0.90145 | 0.01444 | 0.04638 | 20.7% | 12.1% | 30.3% |
| Black | 0.94082 | 0.94401 | 0.94640 | 0.00239 | 0.00558 | 5.2% | 4.2% | 9.2% |
| Hispanic | 0.78500 | 0.85276 | 0.87554 | 0.02278 | 0.09054 | 28.9% | 14.4% | 39.2% |
| API | 0.88614 | 0.88780 | 0.91632 | 0.02852 | 0.03018 | 1.4% | 24.3% | 25.3% |
| AI/AN | 0.46683 | 0.46970 | 0.53851 | 0.06881 | 0.07168 | 0.3% | 8.9% | 9.2% |
| Multiracial | 0.04817 | 0.10398 | 0.12283 | 0.01885 | 0.07466 | 0.9% | 0.4% | 1.3% |
Percent error reduced is [(1 − r old 2) − (1 − r new 2)]/(1 − r old 2), where r old and r new are the Pearson correlation between self‐report and the old and new estimates, respectively.
N = 284 627.