Aims: We are pursuing methodological research on the concept of 'interdependencies' in causal inference.

Methods: The concept of an interdependency in causal inference is most similar to that of an interaction in standard statistical analysis. There are a series of papers by Vanderweele and Robins from 2007 onwards which define interdependence between causes to produce an effect, as opposed to statistical interaction. They also derive observable constraints to detect interdependence. Our work seeks to extend their constraints and defined interdependence without any deterministic assumptions, which they use with counterfactuals.

Example: In our example we would like to investigate whether there are interdepencies between adiposity genotypes on their effect on measures of adiposity such as BMI and waist-hip ratio. Specifically, the application investigates the interdependence between the FTO (X_1) and MC4R (X_2) SNPs in causing obesity (Y).