Table 7.
Simulations with a common sensitivity and specificity
| I | r | FM model average estimate | GRE model average estimate | % (log LFM ≫ log LGRE)* | ||||
|---|---|---|---|---|---|---|---|---|
| Pd | SENS | SPEC | Pd | SENS | SPEC | |||
| 100 | 0 | .20 (.05) | .75 (.10) | .90 (.03) | .25 (.13) | .64 (.18) | .88 (.04) | 2 |
| .05 | .20 (.05) | .76 (.09) | .90 (.02) | .19 (.09) | .74 (.13) | .88 (.04) | 7 | |
| .10 | .20 (.05) | .75 (.09) | .90 (.02) | .19 (.07) | .75 (.11) | .89 (.03) | 10 | |
| .20 | .20 (.04) | .75 (.07) | .90 (.02) | .20 (.05) | .75 (.08) | .90 (.02) | 12 | |
| .50 | .20 (.04) | .75 (.06) | .90 (.02) | .20 (.04) | .75 (.06) | .90 (.02) | 16 | |
| 1 | .20 (.04) | .75 (.05) | .90 (.02) | .20 (.04) | .75 (.05) | .90 (.02) | 19 | |
| 1,000 | 0 | .19 (.02) | .75 (.04) | .90 (.01) | .32 (.12) | .57 (.16) | .91 (.02) | 12 |
| .05 | .19 (.02) | .75 (.03) | .90 (.01) | .20 (.03) | .73 (.05) | .90 (.01) | 45 | |
| .10 | .20 (.02) | .75 (.03) | .90 (.01) | .20 (.02) | .74 (.03) | .90 (.01) | 55 | |
| .20 | .20 (.01) | .75 (.02) | .90 (.01) | .20 (.02) | .74 (.03) | .90 (.01) | 64 | |
| .50 | .20 (.01) | .75 (.02) | .90 (.01) | .20 (.01) | .75 (.02) | .90 (.01) | 75 | |
| 1 | .20 (.01) | .75 (.02) | .90 (.01) | .20 (.01) | .75 (.02) | .90 (.01) | 79 | |
NOTE: Data were simulated under the finite mixture (FM) model with Pd = .2, η0 = η1 = .2, SENS = .75, SPEC = .90, and J = 5. Results are based on 1,000 simulations. Mean parameter estimate and standard errors are presented in ( ).
*
Proportion of realizations where the log-likelihood under the FM model is more than 1 larger than the log-likelihood under the misspecified GRE model.