Since this study employs multiple phenotypic scales, the potential exists for type I errors. For several reasons we have not employed a correction of our significance criteria for multiple testing [91], [92]. The diversity of scales employed in this study should not decrease the sensitivity of analysis. In addition, because of the exploratory nature of our analysis [91], and because the phenotypic variables are often correlated (see Table 1 and Results and Discussion), corrections for multiple tests are too conservative [92]. Nonetheless, to address the risk of Type I errors, we have reduced our significance criteria from the traditional≤.05 to a≤.01 level for the principal analyses (for the correlation matrix in Table 1 a traditional≤.05 criteria is employed). It is important for the reader to bear these cautions in mind as they interpret the results.
