OpenMx, a free and open source software that allows researchers to perform their analysis quickly and easily, has recently undergone several upgrades along with the addition of a new YouTube channel.
According to Tim Brick, a member of the OpenMx development team and assistant professor of human development and family studies, OpenMx is intended for statistical analysis in the extended Structural Equation Modeling (xSEM) framework. “Structural Equation Modeling (SEM) is broad framework for linear modeling. The big strength of xSEM is that the models can be thought of either as a path model, a box-and-arrow plot that shows how different constructs interrelate, or as the underlying system of mathematical equations underneath. OpenMx lets you move seamlessly between the two representations.”
Since last year, Brick and his team have released two minor version releases to fix bugs and introduce power analysis tools, bootstrapping and jackknifing functions, and easier mixture modeling. “Overall, OpenMx is growing in popularity, with downloads almost doubling from the previous year,” said Brick. “We estimate that about 60 percent of our users are in the field of behavior genetics, with the rest primarily in psychology or related disciplines.”
OpenMx extends the SEM framework to allow scientists to specify a variety of complex models that can’t be easily modeled in other software. For example, it allows people to specify sets, or ‘trees’, of models that interlock. “This is helpful when modeling the differences between single or multi-group models, as it allows you to model each component separately and compare how all of those fit together,” Brick explained.
Additionally, the extensions allow OpenMx users to specify many types of models, ranging from multiple-group (e.g. male/female) models, to behavior genetic ACE models, dynamical systems models, and continuous-time SEMs.
OpenMx works within the R statistical computing language so it is useful for a wide range of people. “R has recently become very popular in the behavioral sciences as a tool for data management, analysis, and visualization,” said Brick. “OpenMx is a library of R functions that integrate seamlessly into R, so it's possible to use R's existing tools to load in and set up your data, use OpenMx to fit some models, and R tools to visualize the results.” Integration with R also makes it a great tool for simulation or large-scale analysis, because it's easy to programmatically create, adjust, and re-fit OpenMx models in R.
Additionally, OpenMx runs on several computer platforms including Mac OS X, Windows (XP, Vista, 7, 8, 10), and Linux. This means the same scripts written in Windows will run in Mac OS X or Linux, which can be very convenient for research teams whose members are on different platforms.
The OpenMx site gives examples of the both path model and matrix model specifications, as well as free downloads of the current software, description of its features, a user’s guide, resources, and a discussion forum. “The forum is for users of all types of software to discuss issues in multivariate statistical modeling and to work together to create open source scripts to for use in a wide variety of biological, medical, epidemiological, genetic, and behavioral sciences,” said Brick.
Additional resources include the newly launched the OpenMx YouTube channel which features video tutorials on basic and intermediate use of OpenMx.
OpenMx is hosted by the Social Science Research Institute and the Quantitative Developmental Systems Methodology Core and is partially funded by a grant from the National Institute of Drug Abuse. It is a multisite collaboration between Penn State, University of Virginia, University of Oklahoma Health Sciences Center, Northeastern University, Virginia Commonwealth University, and University of Edinburgh.
For more information on OpenMx and to get started, visit the OpenMx website.