Interpretation of our findings is subject to limitations. In particular, a recent paper describing an attempt to replicate studies of candidate genes influencing subjective responses to amphetamine (Hart et al., 2013) indicates a high probability of false positives, due to factors including insufficient power due to small numbers of subjects (particularly when a rare homozygote is present); publication bias; failure to correct for multiple testing (especially correction for non-published, non-significant tests) and use of intermediate phenotypes, like subjective responses, that are themselves highly variable within and among subjects. Our use of principal components to combine multiple subjective measures into a single component reduced the number of statistical tests necessary; and may have provided us with a less variable indicator of subjective phenotype, but more work is necessary to evaluate the merits of this approach.
