Chunk #43 — Reasons to be Concerned about the Published cGxE Literature — Problems with the Recipe: Statistical Concerns in cGxE Research — The selection of model
difference framework, an interaction reflects a deviation from a model in which risk factors add together. In a risk ratio framework, an interaction reflects a deviation from a model in which risk effects multiply. Accordingly, the usage of logistic regression to study cGxE for binary outcomes of interest (such as presence/absence of a disorder) fundamentally changes the nature of the relationship between two variables: multiplication on the original scale of a variable conforms to addition on the logarithmic scale. Thus, an “interaction” on the original scale can ‘disappear’ or even be of opposite sign on the logarithmic scale, and vice-versa. We refer the reader to other sources (K. Kendler & Gardner, 2010; Zammit, Lewis, et al., 2010) for further discussion of these important issues.
