Table 4.
The estimates, SEs, and 90% and 95% confidence intervals of the BMDs from the normal-based and bootstrap analyses.
| Parameter | Analysis | Estimate (mg/kg-d) | SE (mg/kg-d) | CI Typea | Two-sided 90% CI (mg/kg-d) | Two-sided 95% CI (mg/kg-d) |
|---|---|---|---|---|---|---|
| Level 1 BMD05 | Normal-based | 176 | 38.2b | Unadj. | (123, 251)c | (115, 269)c |
| Bonf. | (115, 269) | (108, 286) | ||||
| Bootstrap | 215d | 91.8d | Unadj. | (116, 425) | (102, 477) | |
| M&B | (111, 428) | (98, 478) | ||||
| Bonf. | (102, 477) | (91, 480) | ||||
| Level 2 BMD01 | Normal-based | 188 | 46.9b | Unadj. | (125, 283)c | (115, 306)c |
| Bonf. | (115, 306) | (107, 329) | ||||
| Bootstrap | 234d | 93.0d | Unadj. | (121, 449) | (105, 478) | |
| M&B | (116, 459) | (101, 479) | ||||
| Bonf. | (105, 478) | (91, 481) |
a
“Unadj.”: CI is not adjusted for multiplicity; “M&B”: Mandel & Betensky multiplicity adjustment is applied; “Bonf.”: Bonferroni multiplicity adjustment applied; the principal confidence intervals used are in boldface.
b
The normal-based SEs were computed using the delta method.
c
The normal-based confidence intervals are t-intervals.
d
The bootstrap estimates and SEs were the means and standard deviations of the BMD estimates from the 10,000 bootstrap runs.