Imputing missing phenotypes: A new family-based association test

摘要

We define a new test statistic that accommodates missing pheno-typic data in family-based association tests (FBATs). The missing phenotypes are imputed using the conditional mean model (Lange et al. (2003)). When the outcome data are missing at random, FBAT-IMP demonstrates higher power, in both simulations and an Alzheimer study, than the standard quantitative FBAT.