ABSTRACT
101In this part of the book, we describe more advanced methods for Mendelian randomization. We begin by considering robust methods, which do not require all genetic variants to be valid instrumental variables (IVs) to give consistent estimates of a causal parameter.
