Since R is a full programming language and the OpenMx specification structure is flexible, there are many styles of model specification that could be used to create identical statistical models. We expect several styles will emerge as users become acquainted with the possibilities. One style that has become common among the core programming team members is to specify each of the MxMatrices separately, assigning them to R variables early in a script. Later, these predefined matrices can be combined into different model configurations somewhat like using Lego blocks. This method results in scripts that bear little resemblance to traditional SEM scripts, but these Lego–style scripts can be easier to write, debug, and maintain.
