Abstract:
|
Mechanism interaction concerns why the effect occurs and why it is the magnitude. When investigating mechanisms, synergism is a widely discussed topic in the fields of genetic study and pharmacy. Synergism under the framework of sufficient component cause is difficult to observe, so the estimation focuses on sufficient cause interaction (SCI). SCI has received much attention to investigating the mechanism of causality since it is all that required for synergism. Under the counterfactual framework, VanderWeele and Robins (2007, 2008) provided empirical tests for SCIs. However, the previous studies only assess the lower bound of SCIs rather than estimate SCIs directly due to the lack of the degree of freedom. Moreover, such empirical tests for the lower bound of SCIs are less powerful. To address this issue, in this study, we propose a novel method to estimate the probability of individual with SCI by introducing a new factor named quasi-instrumental variable, which is necessary for the background condition of SCI. We also develop a corresponding hypothesis test and show that it is more powerful than the empirical test.
|