In this study, the primary object of the analyses are the effect sizes of the SNPs obtained from the non-parametric regression calculations, although the p-values associated with these SNPs are also considered. The use of 21 SNPs and 18 phenotypes with separate effect sizes for males and females at 131 age centers produces 99036 effect size values. Effect sizes, unlike p-values, provide signed values which enable the direction of effect to be analyzed, and thus to more precisely characterize SNPs which have sex-specific and modality-specific effects. Effect sizes also are appropriately signed and scaled for covariance estimation, a key part of the analysis. The use of effect sizes also enables comparison between the results of this study and other studies which use different methodologies and have different sample sizes. In addition, arithmetic operations on effect sizes are easily interpreted, which is not the case with p-values.
