Exploring Integrated Theories Underlying Women’s Pathways to Prison – Using Abductive Inference Methods

• In Event: Intersections of Gender, Race, and Ethnicity in Justice-Involved Populations

Abstract

We first review major phases in identifying and explaining women’s pathways to prison from early qualitative studies to recent Machine Learning (ML) methods. These studies suggest that the broad range of explanatory factors and the complexity of their interactions is linked to multiple theories blended within each pathway. This complexity has required theory-integration methods; we chose Abductive Inference to clarify the internal explanatory structure of each pathway. Abductive Inference, however, specifies several stringent scientific validity requirements, particularly that each pathway is causally homogeneous. A recent multi-state study (Brennan and Jackson, 2022) has met these requirements, identifying four causally homogeneous stable pathways. We found that each pathway can be explained by a specific blend and time ordering of diverse theories such as, but not limited to, Feminist, Control, Social Learning, and Routine Activities. Abduction with person-centered ML methods differs from classic theory integration methods (Hirschi, 1979; Elliott, 2006) since pathway constituents are objectively specified by the density search ML methods. We discuss results and several challenges in this project, as well as next steps.

Authors

• Tim Brennan, Northpointe Institute (retired)

• Emily Salisbury, University of Utah

• Genie Jackson, equivant/Northpointe