umx: Twin and Path-based Structural Equation Modeling in OpenMx

Structural equation modeling (SEM) is an important research tool, especially in the social sciences. Two major uses of SEM include path-based model specification, where ease of use and graphical and table-based reporting are important for researcher productivity, and complex multi-group and multi-level models developed in fields such as behavior genetics where the model specification might often involve dozens or hundreds of lines of matrix-based code. Often these models are experienced as difficult to implement, and may present a steep learning curve before results can be obtained. This paper describes the umx library, which is available on CRAN. It is ideal for both beginners and for those working on complex models. umx supports rapid development, modification, and comparison of models, as well as both graphical and tabular reporting. High-level support is provided for several standard multi-group genetic models for twin data that involve additive genetic, common environment and specific environment variance components. These support raw and covariance data, including joint ordinal, and give solutions for ACE, ACE with covariates, Common Pathway, Independent Pathway, and Gene × Environment interaction models implemented as “one-line” functions. A tutorial site and question forum are also available.