|  |  |  | SNR (\( {e}^{\beta_{HI}} \) = 1) | High MNR (\( {e}^{\beta_{HI}} \) = 0.05) |
---|
ω
0
|
ω
1
|
ω
2
| LRT-1 | LRT-2 | LRT-3 | LRT-1 | LRT-2 | LRT-3 |
---|
Null scenarios
|
False positives
|
3a | 0.05 | 1.0 | 1.0 | 0.55 | 0.02 | 0.03 | 0.0 | 0.06* | 0.10* |
Alternative scenarios
|
True positives
|
3b | 0.05 | 0.5 | 2.0 | 0.95 | 0.87 | 0.92 | 0.01 | 0.44 | 0.26 |
3c | 0.05 | 1.0 | 2.0 | 0.99 | 0.47 | 0.46 | 0.0 | 0.27 | 0.18 |
- LRT-1 compares M1a to M2a (under-fit models). LRT-2 compares G1ax to G2ax (perfect-fit models). LRT-3 compares G1a13 to G2a13 (over-fit models). The asterisk symbol (*) indicates that the results are based on < 100 replicates due to convergence problems with some datasets when there was high MNRs. For LRT-2 case 3a is based on 99 replicates. LRT-3 case 3a is based on 91 replicates