
The Analytic Identification of Variance Component Models Common to Behavior Genetics
Author(s) -
Michael D. Hunter,
S. Mason Garrison,
S. Alexandra Burt,
Joseph Lee Rodgers
Publication year - 2021
Publication title -
behavior genetics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.865
H-Index - 93
eISSN - 1573-3297
pISSN - 0001-8244
DOI - 10.1007/s10519-021-10055-x
Subject(s) - identification (biology) , variance (accounting) , component (thermodynamics) , set (abstract data type) , behavioural genetics , quantitative genetics , variance components , computer science , artificial intelligence , econometrics , psychology , statistics , mathematics , genetics , biology , genetic variation , ecology , physics , accounting , gene , business , programming language , thermodynamics
Many behavior genetics models follow the same general structure. We describe this general structure and analytically derive simple criteria for its identification. In particular, we find that variance components can be uniquely estimated whenever the relatedness matrices that define the components are linearly independent (i.e., not confounded). Thus, we emphasize determining which variance components can be identified given a set of genetic and environmental relationships, rather than the estimation procedures. We validate the identification criteria with several well-known models, and further apply them to several less common models. The first model distinguishes child-rearing environment from extended family environment. The second model adds a gene-by-common-environment interaction term in sets of twins reared apart and together. The third model separates measured-genomic relatedness from the scanner site variation in a hypothetical functional magnetic resonance imaging study. The computationally easy analytic identification criteria allow researchers to quickly address model identification issues and define novel variance components, facilitating the development of new research questions.