To conclude, the IRT results show that the Neuroticism and Extraversion item data from different inventories in different cohorts can be harmonized (for general recommendations and an example R analysis script, see Supplementary Materials Online). The harmonized phenotypes can now also be confidently correlated with brain measures or used in a GWA study. The IRT analysis is not only useful for harmonizing phenotypes, it is also informative regarding the power to find significant genetic variants of various allele frequencies. The TICs show where in the distribution of Neuroticism and Extraversion scores there is most phenotypic information. Relating these TICs to the power a phenotypic test might give in a GWAS (van den Berg and Service 2012), we conclude that there generally is more power to detect low frequency genetic variants associated with scoring at the low end of the Extraversion distribution than towards the high end of the distribution. Similarly, there is more power to detect low-frequency genetic variants associated with scoring above-average on Neuroticism, compared to scoring below-average. Overall, the phenotypic information content is higher for Neuroticism than for
