A. García Pérez
In Mendelian Randomization (MR), the estimation of the effect beta_R_j of an Exposure X on an Outcome Y, by using genetic variants (usually single nucleotide polymorphisms, SNPs) as instrumental variables Z_j, is done with the classical Two-stage Least Squares Estimator.
MR is used to avoid possible biases in the regression of Y on X due to lack of complete randomization in data, or the presence of reverse causation, or confounders.
The combination of these classical estimators of beta_R_j, for different instrumental variables Z_j, is done with the classical inverse-variance weighted (IVW) estimator, that has a 0 breakdown point.
In this contributed paper we propose, first, a new robust estimator of the beta_R_j effect,
using the Median of the Distribution of the Mean estimator, MdM_j, and, second, a new robust way in which these effects are combined, with a robust IVW, using a new dispersion estimator in the combination of effects estimators.
Palabras clave: Robust statistics, von Mises expansions, Mendelian Randomization
Programado
GT03.AMC2 Métodos Robustos
8 de noviembre de 2023 16:00
HC3: Sala Canónigos 3