Table 2 Simulation results: Means, empirical standard errors and 95% coverage probabilities of parameter estimates from 10,000 simulations under observed misclassification rates of race/ethnicity by sampling design and method
From: Survey design and analysis considerations when utilizing misclassified sampling strata
| Â | Full cohort | SRS | Disproportionate stratified sampling | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Â | Â | Â | Design-agnostic | Model-based | Design-based | |||||
Non-differential misclassification | ||||||||||
Intercept | -0.75 (0.01) | 94.7 | -0.75 (0.05) | 95.0 | -0.75 (0.06) | 95.1 | -0.75 (0.10) | 94.5 | -0.75 (0.07) | 94.9 |
Black | -0.25 (0.02) | 95.0 | -0.26 (0.15) | 94.9 | -0.25 (0.11) | 94.8 | -0.26 (0.25) | 95.4 | -0.25 (0.12) | 94.9 |
Asian | -0.50 (0.07) | 95.0 | -0.59 (0.82) | 96.7 | -0.51 (0.21) | 95.2 | -0.51 (0.26) | 95.2 | -0.52 (0.27) | 93.0 |
Other | 1.25 (0.03) | 95.1 | 1.26 (0.21) | 95.2 | 1.25 (0.14) | 95.5 | 1.26 (0.18) | 95.6 | 1.26 (0.21) | 95.1 |
Hispanic | -1.50 (0.07) | 95.0 | -1.63 (0.87) | 96.8 | -1.55 (0.41) | 95.9 | -1.55 (0.42) | 96.0 | -1.56 (0.47) | 95.2 |
Low income | 1.00 (0.02) | 94.5 | 1.00 (0.12) | 95.4 | 1.00 (0.11) | 95.1 | 1.01 (0.11) | 95.1 | 1.00 (0.14) | 95.0 |
Differential misclassification | ||||||||||
Intercept | -0.75 (0.01) | 94.7 | -0.75 (0.05) | 95.1 | -0.57 (0.06) | 13.0 | -0.74 (0.10) | 95.0 | -0.75 (0.07) | 95.2 |
Black | -0.25 (0.02) | 95.0 | -0.25 (0.15) | 95.1 | -0.42 (0.11) | 67.6 | -0.42 (0.24) | 90.1 | -0.26 (0.12) | 94.8 |
Asian | -0.50 (0.07) | 95.0 | -0.58 (0.79) | 96.5 | -0.81 (0.20) | 67.9 | -2.16 (0.27) | 0.0 | -0.52 (0.28) | 91.7 |
Other | 1.25 (0.03) | 95.1 | 1.26 (0.21) | 95.0 | 0.96 (0.13) | 42.4 | 0.02 (0.19) | 0.0 | 1.25 (0.20) | 94.5 |
Hispanic | -1.50 (0.07) | 95.0 | -1.62 (0.87) | 97.0 | -1.63 (0.36) | 96.3 | -2.14 (0.38) | 62.1 | -1.55 (0.42) | 95.2 |
Low income | 1.00 (0.02) | 94.5 | 1.00 (0.12) | 94.7 | 1.00 (0.11) | 95.2 | 1.01 (0.11) | 94.9 | 1.00 (0.15) | 94.8 |