Classically derived estimates of heritability from twin models have been plagued by the possibility of genetic-environmental covariance. Survey questions that attempt to measure directly the extent to which more genetically similar kin (such as monozygotic twins) also share more similar environmental conditions represent poor attempts to gauge a complex underlying phenomenon of GE-covariance. The present study exploits a natural experiment to address this issue: Self-misperception of twin zygosity in the National Longitudinal Survey of Adolescent Health (Add Health). Such twins were reared under one "environmental regime of similarity" while genetically belonging to another group, reversing the typical GE-covariance and allowing bounded estimates of heritability for a range of outcomes. In addition, we examine twins who were initially misclassified by survey assignment--a stricter standard--in three datasets: Add Health, the Minnesota Twin Family Study and the Child and Adolescent Twin Study in Sweden. Results are similar across approaches and datasets and largely support the validity of the equal environments assumption.