Chunk #48 — Reasons to be Concerned about the Published cGxE Literature — Problems with the Recipe: Statistical Concerns in cGxE Research — Power to detect and characterize different types of interactions
Further, even if an interaction is detected, discerning the true pattern of an interaction from observed results is even more tenuous. In recent years, there has been great interest in determining the form of the observed interaction in cGxE research since the interpretation of disordinal (i.e., “cross-over”) interactions theoretically differs from ordinal interactions. Specifically, cross-over interactions lend themselves to a differential susceptibility interpretation where a given “risk” or “malleable” allele is associated with both poorer outcomes in a “bad” environment but better outcomes in a “good” environment; ordinal interactions lend themselves to a diathesis-stress interpretation where it is the combination of a risk-conferring allele and a “bad” environment that exacerbates the likelihood of manifesting the outcome (e.g., Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2007; Belsky et al., 2009). However, simulations demonstrate that, conditional upon a Type 1 error, the form of an ostensibly “significant” interaction is usually of a “cross-over” (i.e., disordinal) nature, especially when samples sizes are small (Sher & Steinley, 2013). Boardman et al. (2014) recently made a similar observation in reference to the emerging genome-wide gene-by-environment (GWGEI) approach
