It is difficult to speculate why our novel, population-based heritability estimate is so much lower than those from previous family- and studies. Common environment may partly explain a familial effect that may have been misattributed to genetics. Like all studies, ours has a number of limitations. The EPQ-N was assessed from self-administered questionnaires posted back to the researchers. Only 66% of the eligible subjects sent back the questionnaire, which may have generated a response bias, although we have shown that there are only minor differences in sex and age between responders and non-responders. Our study has a cross-sectional design, and individuals suffering a neurological condition such as dementia may be less likely to complete and return the EPQ. However, all participants in the PsyCoLaus had a psychiatric assessment and completed a Mini Mental Stage Examination [49]. If a locus has pleiotropic effects, affecting both neuroticism and dementia in old age, this could generate an age-dependent response-bias towards protective alleles, leading to a genotype by age interaction such as we observed at GPC6.
