ABSTRACT
In this chapter, we consider the effect of weak instruments on instrumental variable (IV) analyses. Weak instruments, which were introduced in Section 4.5.2, are those that do not explain a large proportion of the variation in the exposure, and so the statistical association between the IV and the exposure is not strong. This is of particular relevance in Mendelian randomization studies since the associations of genetic variants with exposures of interest are often weak. This chapter focuses on the impact of weak instruments on the bias and coverage of IV estimates.
