Behavior-genetic research on the sources of individual differences in
personality relies on self-report data almost exclusively. As indices of
inter-judge agreement yield the most adequate reliability estimates in
behavior-genetic research on personality, and inter-judge agreement can
not be estimated from self-report data alone, behavior-genetic self-report
studies do not allow for adequate reliability estimates. Therefore, the
reliability problem is usually ignored and the total variance is treated
as true-score variance, resulting in underestimates of genetic and shared
environmental and in overestimates of nonshared environmental influence.
This problem may be overcome by using descriptions of the target persons
by (at least) two independent knowledgeable informants, as we did in a
study on 1,000 twins pairs.
Another problem that is shared by behavior-genetic self-report and peer-report studies concerns possible contrast effects in descriptions of relatives as the relatives may be compared to each other and not to the population mean. This would result in attenuated correlations between relatives and in underestimates of the importance of the shared environment. The only way to overcome this problem lies in observational behavior-genetic studies in which the judges know only one of the relatives whose similarities are compared. We therefore ran an observational study on the similarity of 300 monozygotic and dizygotic adult twins pairs, the German Observational Study of Adult Twins (GOSAT). The study and its most important findings for personality are reported.
Zurück zum Publikationsverzeichnis