The adoption study is a “natural experiment” but has methodological difficulties, such as confidentiality, sampling (ascertainment), and selective placement. When performed in nuclear families or combined with twin analysis (of special note are monozygotic twins reared apart), adoption studies have the potential to allow estimation of broad sense heritability and genotype environment interaction. A common method of analysis for continuous traits uses path analysis together with maximum likelihood. In the case of binary traits, logistic regression may be used, or in the framework of structural equation modeling, integration of the multivariate normal distribution is necessary.