All models based on normal distributions can be represented as Structural Equation Models (SEMs). Ωnyx is a program to generate SEMs as graphical path diagrams. The path diagram can be fitted to data directly in Ωnyx, or quickly translated into a script to run in different systems like OpenMx, lavaan, or MPlusor as a java program using the Ωnyx package. Ωnyx uses a unique multi-agent optimization approach that, unlike most other SEM programs, continues to estimate even after it has converged on a solution. This increases both the chances of finding a global optimum and not getting stuck in a local optimum, and results in a fit that is largely independent of initial values. Both Bayesian and frequentist tests can be run in Ωnyx.
On the left is a graphical example of a structural equation model as it can be created and displayed in Ωnyx.
At the Thomas Bayes Institute, our research equips SEM programs in general and Ωnyx in particular to meet the challenges of modern data analysis. In particular, we are exploring approaches to (1) ease the transition to a modern Bayesian SEM framework by allowing Ωnyx to import a researchers’ older path models, (2) simplify the use of control and auxiliary variables, (3) integrate new path diagram elements such as, product-of-variable nodes, (4) apply clustering methods to path diagrams, such as Dirichlet clustering or model forests, (5) make multi-level structures easily applicable in Ωnyx, and (6) to investigate LLMs for the input of path diagrams