A method that can be used to test for heterogeneity of effect estimates between instrumental variables (IVs) being used in a MR analysis. Overidentification tests are applicable where there are more IVs than exposures of interest.
If all IVs are valid, they should have identical effect estimates in an MR context. Substantial differences (i.e., heterogeneity) in effect estimates across multiple IVs may indicate that at least one IV is invalid for MR analyses through a violation of MR assumptions. Overidentification tests (e.g., such as the Sargan test, traditionally used within one-sample MR analyses) can be used to test for the existence of heterogeneity between IVs and, therefore, the presence of horizontal pleiotropy or bias due to confounding from population stratification, for example. However, if all IVs are biased in the same way, over-identification tests will not be able to accurately identify such bias. Over-identification tests also rely on the assumption that all IVs are estimating the same causal effect, which may not be true for complex traits.