If the reason for missingness is related to covariates being tested, then SCM results could potentially vary across MI datasets, and borderline significant/non‐significant results could lead to varied selection/non‐selection across the 100 SCM runs.
← all excerpts
Incorporating Multiple Imputation Into Tanezumab Dose-Response Modeling of WOMAC Pain CFB Across Six Randomized Placebo-Controlled Phase 3 Trials.
1
—
—