On Quantifying Bias in Causal Effects When Data Are Non-IID

Traditional causal inference techniques assume data are independent and identically distributed (IID) and thus ignore interactions among units. In this paper, we analyze the bias of causal identifi-cation techniques in linear models if IID is falsely assumed. Specifically, we discuss 1) when it is safe to apply traditional IID methods on non-IID data, 2) how large the bias is if IID methods are blindly applied, and 3) how to correct the bias. We present the results through a real-world example of vaccine efficacy.