Linking theory and Taxonomy: Using Abductive Inference to elicit underlying generative mechanisms of separate pathways

• In Event: Technology, Theories, and Prosecutorial Discretion

Abstract

This paper addresses theory development in regard to a well replicated quantitative taxonomy of Women’s Pathways to chronic crime and incarceration. The study used two large State Prison samples of women. Classification variables included the Women’s Risk and Needs Assessment (WRNA) as well as several gender-neutral criminogenic variables. Multiple machine learning (ML) methods, cross-sample validations, and bootstrap re-sampling methods had demonstrated the robust stability of the identified pathways. Abductive inference was then applied to each separate pathway to elicit the most plausible theoretical explanations. Each pathway reflected quite different forms of theoretical pluralism with specific blends of theories occurring in each pathway. Thus, each pathway was sufficiently complex so that no single theory could dominate the pathway. However, all the pathways reflected substance abuse at various levels. An interesting finding is that the ML “detection” of these pathway patterns reflects nature’s way of theoretical integration in contrast to the efforts of criminologists to join theories together

Authors

• Tim Brennan, University of Colorado (Retired)

• Emily J. Salisbury, Utah Criminal Justice Center, University of Utah

• Eugenie Jackson, Northpointe, Inc.