Across cohorts, personality scores were largely comparable; that is, the extent of measurement variance was overall not large. We did, however, observe measurement variance for a few cohorts and for some items. Differences in item parameters across cohorts seem largest in cohorts with different spoken languages, suggesting cultural and/or language effects on some of the items. As a consequence, the estimated latent scores across cohorts are not based on completely identical scales. Again, for individual scoring this has consequences (e.g., a person’s ranking within a population), but these imperfections have little effect on results for population effects, because the correlation of two scores based on different calibrations was generally very high. Overall, the conclusion is that data pooling within cohorts and subsequently pooling results across cohorts in a meta-analysis is meaningful for Neuroticism and Extraversion and these inventories. As the power study showed, such pooling of data within cohorts can lead to a potentially large increase in statistical power. Such increase in power is largely due to the increase in sample size, but also of using more phenotypic information per individual.
