Learning progressions (LPs) serve as frameworks for understanding the level of complexity of students' ideas within a domain. Multifaceted LPs contain multiple interrelated constructs where increased knowledge in one construct may increase knowledge of a separate construct. While it is challenging to identify and test these contingencies between constructs, our research group has used Bayesian structure learning methods to determine how mental models of genetics change with instruction. In this study, we follow a cohort of high school students for 3 years, measuring their genetics knowledge prior to instruction in 10th grade, immediately after genetics instruction in 10th grade, and at the end of 12th grade. We investigated how their idea networks change as a result of instruction and how they stabilize 2 years after instruction ended. We found students move from a phenotypic to a more molecular/mechanistic understanding of genetics immediately after instruction; students idea networks 2 years after instruction stabilized grounded within molecular explanations of diversity of life and how genetic diversity drives evolution. This study provides a robust mechanism for researchers to test connections between constructs in multifaceted LPs, illustrated by a genetics framework unable to be detected by previous methods.