From theory to conceptualization, through operationalization: Comparing indicators of desistance from crime

Appreciating the development of the literature on desistance and producing scientific research in the context of conceptual and operational instability is a complex exercise. Several constraints affect the choice of operationalization, including the characteristics of available data. Beyond the search for a perfect or consensual measure, it becomes imperative to understand how the definitional and conceptual choices shape and limit studies. The current analysis contributes to this literature by comparing three operationalization strategies for desistance from crime in quantitative and longitudinal designs: a binary measure of participation in delinquency, a scale measure of the versatility of offending, and a scale measure of the intensity of offending (original measure combining versatility and frequency of offending). Taxonomies of multilevel models for change were conducted with a subsample of data collected as part of the Montreal Two Samples Four Generations Cross-sectional and Longitudinal Studies to predict desistance as a function of age and various sets of common time-varying (level 1) and time-stable (level 2) independent variables. Results show differences in the capacity of every measure to capture desistance, and predictors vary in nature and number according to the predicted outcome (participation, versatility or intensity). Overall, results tend to indicate that the various measures are complementary; they provide a more complete picture of desistance as they capture different aspects or phases of the phenomenon. The strengths and limitations of each operationalization strategy are discussed. Relevant direction for further research considering the conceptual and operational diversity in this field of study are suggested.