Explaining Criminal Careers: Implications for Justice Policy

Front matter

JOHN F. MACLEOD, PETER G. GROVE, AND DAVID P. FARRINGTON

Clarendon Studies in Criminology

EXPLAINING CRIMINAL CAREERS

Foreword

The construct of a ‘criminal career’ is a powerful approach for accumulating rich knowledge about offenders and using that knowledge for developing rational policies for dealing with crime.

This book makes a major advance in pulling together that knowledge and pointing to directions for improved use and for expanding that knowledge. The basic thrust of criminal career research is to look at the characteristics of individual offenders and to use that information for dealing with a particular individual, but also to look at aggregates of offenders and their collective characteristics for guidance on how criminal justice policies can best respond to their offending patterns.

The predominant structure of an individual criminal career looks first at the envelope of the offending pattern, the initiation and termination of the career, and then at the pattern of offences within that envelope. The characteristics of those offences would be the mix of crime types, their frequency, and the way that mix varies over the course of the career.

There are some clear relationships between these characteristics and policy choices by the criminal justice system. The interval between initiation and expected termination, or the duration of the career, should have an important influence on sentencing policy. In particular, it would make little sense for sentences to be longer than the residual duration, or the time from any point of intervention to expected termination. As one examines different offenders in terms of their threat to public safety, those with a highest offending frequency, especially of the most serious crime types, represent prime candidates for incarceration in order to achieve the largest crime-reduction effect through incapacitation.

The authors of this volume have done an impressive job of analysing a large body of longitudinal UK data on individual criminal careers, have accumulated a prodigious amount of analytic insight into a wide variety of aggregate criminal career characteristics, and have gone well beyond criminal careers in the issues on which they have generated important understanding. Their volume is replete with a rich array of models on many aspects of individual

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com viii Foreword criminal careers and aggregate implications of those careers. The results are strengthened by examining the careers in multiple cohorts drawn at different times, with a striking consistency of the results across the cohorts.

The authors have also done an impressive job of capturing many important facets of individual criminal careers. They find it sufficient to characterize the population of offenders into three categories. There is one low-risk category with a low-offending frequency (frequency is often denoted by the Greek letter lambda, A). There is a high-risk category with considerable heterogeneity in its offending frequency, and that heterogeneity is captured sufficiently by partitioning them into two subcategories: a high-A subcategory and a low-A subcategory.

Where others have tended to describe the envelope of the career in terms of its duration, the authors find that unsatisfactory and instead characterize the termination process by a desistance probability or its complement, a reconviction probability. This approach captures the aggregate decline in criminal participation without requiring any change in the individual offending frequency, which some criminologists have assumed must occur because of the wellestablished aggregate decline in the age-crime curve (a measure of the rate of arrest or conviction as a function of age, which rises rapidly to a peak in the late teen years and then declines more slowly than the teenage rise). Importantly, the authors attribute the aggregate decrease in crime to the declining participation in offending rather than to a slowing down by a roughly constant population of offenders.

With this basic structure, they are able to capture many aspects of careers such as survival distributions (the fraction of an offending cohort still active at any time after initiation), disentangling the contributions of prevalence and Ato the widely known age-crime curve.

They also examine the degree of specialization or generality in a career based on the mix of crime types included (emerging with the intriguing finding that the number of different crime types ina career involving K offences is given by the logarithm of K). Then they go further to develop estimates of the mix of anticipated types of crimes included in a career based on information on types they already know to be included. They also derive estimates of the fraction of the population in each offender category who ever get convicted, a fraction that they find impressively constant across birth cohorts. In their models, since participation is reduced after each conviction,

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com

Foreword ix that highlights the importance of conviction as a means of crime control and downplays any significant effect of incapacitation.

The impressive richness of their analysis of the many facets of criminal careers is exceeded by the extent to which they are able to link information on the criminal career structure to a wide variety of other features about offenders and aspects of criminal justice policy. They obtained psychological information on offenders and find a variety of intriguing associations between those psychological characteristics and the offending category. They also emerge with an intriguing array of policy insights and approaches for the criminal justice system. Then, by identifying a variety of policy approaches they can estimate trends in prison population and emerge with an intriguing variety of recommendations about how to reduce crime and make the criminal justice system more efficient and effective.

This volume is a powerful example of the richness of the variety of models that could be developed by people with strong mathematical skills, multiple sources of large samples of relevant data on individual criminal careers, and strong insights into the dynamics of crime and the criminal justice system. Of course, any such modelling inevitably involves a wide variety of assumptions. Most of the authors’ assumptions seem quite reasonable, but some may have an unanticipated strong influence on their conclusions. For those who would like to challenge their conclusions, the modelling analysis provides the opportunity. Those who would like to draw other conclusions have the challenging task of identifying which assumptions were most influential in generating the conclusion they didn’t like. They could then suggest other assumptions and their implications. Then, others could choose the most reasonable assumptions, preferably by data analysis or experiment. The volume provides rich opportunities for such debates and comparative tests.

While this volume should stimulate criminal career research in all countries, I would find it particularly fascinating to apply some of the modelling perspectives generated in this volume to the wide variety of individual criminal career data available in the US. All the states maintain repositories of individual arrest histories and one could readily generate cohorts of first-time arrestees in different time periods and in different states and pursue very similar analyses, perhaps substituting arrest events, which are well recorded in the US, for the conviction events that are much better recorded in

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com x Foreword the UK (and which are not as well recorded in the US). It would be particularly interesting to contrast criminal career patterns and their implications in three or four different large states characterized by different aggregate levels of criminal activity. For example, California, New York, Illinois, Pennsylvania, and Texas all have excellent criminal record information, and are different in many aspects of their policies and demography. A replication of at least some of the approaches presented by this volume would be intriguing in the results obtained in each state, in contrasting results in different states, and in contrasting results with the UK.

It would be particularly appropriate and desirable for the National Institute of Justice, the research arm of the US Department of Justice, to pursue some of this criminal career research. There was considerable effort along those lines in the 1970s and 1980s, and those early efforts provided some of the initial starting points for the much fuller developments in this volume. Nevertheless, very little of that research has been supported more recently. It is entirely possible that criminal career patterns could well have changed in the US since those earlier results. Has the proportion of serious offenders in the population changed to any significant degree, and are offending frequencies or career durations or other aspects of criminal careers significantly different from 30 or 40 years ago?

And how do they compare to those characteristics in the UK developed in this volume?

A particularly challenging finding in this volume is the observation that incapacitation effects are negligibly small, largely because time spent in prison serves merely to lengthen the criminal career by that amount of time. That observation could well be a consequence of the assumption that there is a fixed recidivism (or, nonrecidivism) probability following each conviction (or arrest in the US case), and the absence of arrests in prison could be what keeps the career going after release. Or it could be a consequence of the criminogenic effects of incarceration resulting from interaction with criminal peers or the difficulty of gaining successful employment after release, all of which might even contribute to lengthening the career or leading to more serious crimes after release.

These are the kinds of issues that could well be taken up by researchers skilled in the kinds of modelling that is so admirably displayed in this volume. Even more interesting would be the degree to which the volume will serve to recruit skilled modellers who are currently much more immersed in finance and business matters.

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com

Foreword xi

They will find the models developed here challenging their skills and thereby serve to enrich the population of criminologists in all countries with the analytic skills demonstrated by MacLeod, Grove, and Farrington. That would be a most useful and desirable outcome of this demonstration of analytic prowess and creativity.

Professor Alfred Blumstein, H. John Heinz III College Carnegie-Mellon University, Pittsburgh, PA

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com

Acknowledgements

First and foremost our thanks must go to the Home Office statisticians who, in the early 1960s, conceived and set up the Offenders Index and to the generations of their successors who have painstakingly updated and maintained the database over a period of nearly 50 years. In particular we would like to thank Chris Kershaw and Julian Prime for their help and assistance in providing us with the data required for the analysis and research reported on in this volume. We would also like to thank David Godfrey, Christine Harper, Gerard Sheldon, and Kathryn Glaser for their contributions to the prison projection modelling project. There are also many former colleagues in the Home Office and the wider research community including Chris Lewis, Pat Dowdswell, Richard Price, Paul Wiles, and Robin Marris who are all due our thanks for their encouragement in this endeavour. We particularly wish to acknowledge the inspiring work of Professor Alfred Blumstein, who developed criminal career research in the 1970s and 1980s. We are also very pleased that he agreed to write a foreword to our volume. David Farrington particularly wishes to thank Alfred and Delores Blumstein for their warm hospitality in accommodating him on his many visits to Pittsburgh. We also thank Alex Piquero and Alan Lindfield for their constructive comments on early drafts of the manuscript, and an anonymous reviewer for his/her helpful suggestions and in particular for pointing out that what seems obvious to us is not necessarily clear to others.

Much of the work reported in this volume was undertaken while two of us, Grove and MacLeod, were employed by the Research, Development and Statistics Directorate of the Home Office. But, while our results may have been influential in several areas of policy and management, the views expressed herein are those of the authors and not necessarily those of the Home Office or Ministry of Justice (nor do they reflect UK Government policy, either now or in the past).

This is an open access version of the publication distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/by-ne-nd/3.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact academic.permissions@oup.com