Sound psychometric findings are essential. The endophenotype should be reliably measured, ideally with reliability equal to or better than that of the target disorder with which it is associated. An ideal endophenotype would be stable over extended periods of time, allowing for its measurement reliability to be enhanced by exploiting the added information gained from repeated assessments. Twin studies can confirm the heritability of the endophenotype, which should be substantial and higher than that of the associated clinical disorder. In addition, twin studies should show that the phenotypic association between the endophenotype and the disorder is attributable to shared genetic effects. As Dick (Dick, This Issue) notes, genetic influences on a clinical phenotype like alcohol misuse show developmental change. The search for endophenotypes should assign priority to measures that are relatively insensitive to changing genetic influence. Longitudinal studies with twins can determine the degree to which genetic influences are largely the same over extended developmental periods. If this criterion is met and the endophenotype is stable, then the endophenotype is likely to index the nature of genetic risk similarly at different ages, thus enhancing its utility for gene finding.
