Table 4.
Censored data simulation: Percent of LRT statistics under the null hypothesis exceeding critical value for pairs of nested models based on 2000 replicates of N=1000.
| % Type I error rates | |||||||||
|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||
| DGM | LRT df | Censored on M | Censored on P | ||||||
| Model for HA | Model for H0 | Condition of H0 | 10 | 5 | 1 | 10 | 5 | 1 | |
| Cholesky with G×M (3) | NL main effects with G×M (4) | large β2 | 4 | 4.9 | 2.2 | 0.3 | 9.3 | 4.8 | 0.9 |
| small β2 | 4 | 4.9 | 1.6 | 0.3 | 5.4 | 2.2 | 0.5 | ||
| NL main effects (4*) | large β2 | 7 | 5.7 | 2.8 | 0.5 | 82.5 | 82.4 | 81.7 | |
| small β2 | 7 | 5.5 | 2.5 | 0.5 | 100 | 100 | 100 | ||
| Cholesky (2) | high rAM | 6 | 23.0 | 15.1 | 4.2 | 100 | 100 | 100 | |
| low rAM | 6 | 50.2 | 37.5 | 16.2 | 100 | 100 | 100 | ||
|
| |||||||||
| NL main effects with G×M (4) | NL main effects (4*) | large β2 | 3 | 8.7 | 4.5 | 0.9 | 83.7 | 83.1 | 81.7 |
| small β2 | 3 | 8.8 | 4.2 | 0.7 | 100 | 100 | 100 | ||
|
| |||||||||
| NL main effects (4*) | Lin main effectsa (4†) | 1 | 10.4 | 5.3 | 1.1 | 67.3 | 54.0 | 30.8 | |
|
| |||||||||
| Cholesky (2) | Lin Main effects (4†) | 2 | 8.0 | 4.3 | 0.7 | 8.8 | 4.5 | 0.5 | |
Note: P refers to a phenotype of interest, M refers to a putative moderator. LRT refers to the likelihood ratio test.
a
DGM does not contain G×M effects.