Biased Encouragements and Heterogeneous Effects in an Instrumental Variable Study of Emergency General Surgical Outcomes

by   Colin B. Fogarty, et al.

Instrumental variable (IV) studies seek to emulate randomized encouragement designs where patients are assigned a random nudge towards a particular treatment. Unfortunately, IV studies may fall short of their experimental ideal due to hidden bias affecting both the proposed instrument and the outcomes. While sensitivity analyses have been developed for IV designs most proceed under an assumption of effect homogeneity, unlikely to hold for many applications. In our case study on the efficacy of surgical versus non-surgical management for two gastrointestinal tract conditions, whether or not surgery will be of greatest benefit strongly depends upon patient-specific physiology. The validity of our proposed instrument, surgeons' tendencies to operate, is plausible but not ironclad, necessitating a sensitivity analysis allowing for effect variation. We present a new sensitivity analysis accommodating arbitrary patterns of non-identified effect heterogeneity. With binary outcomes, we prove that the conventional sensitivity analysis based on McNemar's test is itself asymptotically valid with heterogeneous effects. We then demonstrate that the identified components of effect heterogeneity can, when suitably exploited, improve the performance of a sensitivity analysis through providing less conservative standard error estimators. We further highlight the importance of heterogeneity in determining the power of a sensitivity analysis.



There are no comments yet.


page 1

page 2

page 3

page 4


Testing weak nulls in matched observational studies

We develop sensitivity analyses for weak nulls in matched observational ...

Salvaging Falsified Instrumental Variable Models

What should researchers do when their baseline model is refuted? We prov...

Profiling Compliers in Instrumental Variables Designs

Instrumental variable (IV) analyses are becoming common in health servic...

Average causal effect estimation via instrumental variables: the no simultaneous heterogeneity assumption

Instrumental variables (IVs) can be used to provide evidence as to wheth...

Instrumental Variables: to Strengthen or not to Strengthen?

Instrumental variables (IV) are extensively used to estimate treatment e...

Necessary and Probably Sufficient Test for Finding Valid Instrumental Variables

Can instrumental variables be found from data? While instrumental variab...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.