U.S. flag

An official website of the United States government, Department of Justice.

But What Does It Mean? Defining, Measuring, and Analyzing Desistance From Crime in Criminal Justice

Executive Summary
Download the full chapter
Date Published
October 4, 2021

Attend a Webinar

NIJ is hosting a webinar on November 9, 2021, from 1:30-3pm ET. During this webinar, the authors will present and discuss key themes from each of their chapters. 

Register to attend the webinar.

Research on crime over the course of an individual’s life has increased in the last 30 years both in scope and specificity. One focus area that has emerged from this work is what scholars call “desistance from crime.” Generally, desistance is understood to mean the reduction in criminal behavior that occurs after a person reaches adulthood.

But exactly what desistance is remains unclear, as varying definitions and measurement strategies have evolved over time. Early scholarship tended to view desistance as an event — that is, the termination of offending or end of a criminal career. More recent definitions suggest that desistance is instead a process by which criminality declines over time. Because inconsistent definitions lead to varying measurement strategies, it is difficult to come to conclusions about desistance.

The overall goal of this white paper is to provide grounded recommendations for policy and practice. To do that, the paper reviews definitions of desistance used in the literature and then offers an updated, theoretically grounded definition as a foundation for future work: Desistance is “the process by which criminality, or the individual risk for antisocial conduct, declines over the life-course, generally after adolescence.”

The paper discusses how researchers have measured and modeled desistance and explores the implications of these strategies. Which ways of measuring desistance get closest to the phenomenon of interest? Which are most likely to advance our understanding of why people exit a criminal life and how we can facilitate that process? These guiding questions provide a framework for the paper.

Finally, this white paper provides an overview of unresolved issues — such as the choice between surveys and official records, quantitative and qualitative methods, types of samples, and various modeling techniques — and offers detailed recommendations for policymakers, practitioners, and scholars who are seeking to examine and promote desistance from crime. The recommendations are as follows:

  1. The source of data — from surveys or official records — is consequential for the conclusions that can be drawn from research. When possible, researchers should use surveys to measure desistance. If only official data are available, they are suitable as indirect measures of desistance and termination (the ending of a criminal career).
  1. Both qualitative and quantitative data are useful for studying desistance and can provide unique information on the process. If both qualitative and quantitative data are available, researchers will be able to capture a more complete picture of desistance.
  1. The goal of a study of desistance should guide whether researchers use general samples or samples of persons convicted of a crime. If the study aims to generate knowledge on correlates of normative desistance, general (or community) samples will suffice. However, if the goal is to evaluate criminal justice interventions, it is necessary to use persons convicted of a crime for the sample.
  1. Follow-up periods should be as long as possible. Follow-up periods of less than nine or 10 years may not be able to capture the entire process of desistance, but they may help identify early stages of desistance.
  1. Measures and modeling of desistance should move beyond a binary, “committed a crime or not” approach. Scholarship has tended to show that desistance is a process that must be modeled over time using multiple indicators.
  1. Indicators of desistance ideally should be those relating to criminality, such as antisocial attitudes, self-control, or even common risk assessments. However, criminal behavior can be used as an indirect measure or to capture termination.

Date Created: September 30, 2021