In addition to the potential for low power and low prior probabilities associated with the study of candidate genes (Munafo, 2009), it is also likely that there is insufficient correction for and/or underreporting of multiple testing. Publication bias is also probable, whereby authors are more likely to submit, and editors are more likely to accept, cGxE findings that are statistically significant. A recent report notes that problems contributing to and a consequence of such bias are rampant in the cognitive sciences (J.P. Ioannidis, Munafo, Fusar-Poll, Nosek, & David, 2014) and there is similar evidence in the cGxE literature (Duncan & Keller, 2011). Specifically, the vast majority of first-reports of a given cGxE finding were positive, whereas a much lower proportion of attempted replications were positive. Ioannidis and colleagues (2014) refer to this as the “Proteus phenomenon” whereby the rapid publication of positive findings might temporarily create a halo in which negative findings might be more readily entertained by journals for a short period of time. Furthermore, and inconsistent with expectations based on statistical power, Duncan and Keller found that the
