FORUM on Corrections Research

Expanding the Recidivism Inquiry: A Look at Dynamic Factors⁽¹⁾

Everyone knows that the difference between the public service and private enterprise is that when you're working for the government, there's no bottom line to worry about. I don't know who first said this, and although it may originally have been an accurate observation, by now it's a cliche'. In the case of corrections, it's also wrong. In corrections there is a bottom line, one that must never be forgotten or ignored: recidivism.

The ideas in this paper follow from the recognition that the prevention of recidivism is the primary mission of any correctional system. The first part will deal with some questions on the prediction of recidivism generally - what we know, and what we don 't. This leads to a discussion of some ongoing research that aims to reduce our ignorance and perhaps provide some information that can be used when making decisions both for general policies and in individual cases.

The results reported here are the first from a continuing project, which is being conducted jointly with Vern Quinsey of Queen's University. Because the data are incomplete, I would like to caution that this report does not represent our bottom line. I reserve the right to disclaim any results included here, should they not be supportable in the future when we have more information. Our perspective on the truth is evolving, and it is subject to change.

Static Predictors of Recidivism

It is well known that a variety of historical measures can be used to predict recidivism. For example, individuals with a long criminal his-tory or those who have had serious substance abuse problems are more likely to return to prison than others. While none of these measures is an especially strong predictor on its own, they can be combined statistically into scales that yield much better results.

Canadian studies have contributed substantially in this area. One of the better predictive scales, the Statistical Information on Recidivism (SIR) scale,⁽²⁾ is likely familiar to most FORUM readers as it is routinely used by the Correctional Service of Canada in material to support parole decisions.

The available scales are unquestionably useful because they can predict recidivism in the aggregate. That is, given a large number of offenders with a certain set of characteristics, we can reliably predict what percentage will return to prison for a new offence within a few years. As a rough generalization, we can say that the accuracy of the scales falls about halfway between chance and perfect prediction.

This is pretty good for people like myself who enjoy playing with numbers, but it's thin gruel for the front-line workers in corrections who need to make decisions about individuals. When we focus on individuals rather than aggregates, some serious errors can be made, even when using the best available scales. Clearly, we need to improve our techniques in this area.

Following the Path to Recidivism

I believe there are some major limitations, however, that will make it difficult or impossible to improve significantly on the performance of current prediction instruments. One reason is that the scales are simply collections of measures that correlate with recidivism, without considering the psychological or social processes that actually cause the resumption of criminal behaviour. As a result, one can't very easily build on past work to make gradual improvements. The only way to improve prediction using these scales is to re-create the entire scale and hope that the new version will be better than the previous one.

A better approach to improving prediction is to consider what happens when an offender's behaviour changes from lawful to criminal. Psychological causation is not statistical; for individuals it is always personal.⁽³⁾

To construct a theoretical model of what causes recidivism, one must include more than just the historical factors that appear in current predictive scales. Not only have past attempts ignored the need for theoretical integration, but they have also used a restricted range of information - usually whatever was readily available in institutional files.⁽⁴⁾ Thus, researchers have predicted future behaviour exclusively on the basis of a selected range of historical information. Data on more dynamic and current functioning (attitudes, thoughts, behaviour in prison or coping patterns) have not yet appeared in the scales, largely because the data are not universally or easily available in offender files.

This restriction of sources of information has some undesirable consequences. Not only does it limit the predictive usefulness of the scales, but some of the variables used are also tied to social and economic inequities. An unpublished study comparing the predictive ability of several scales⁽⁵⁾ found that the total scale scores were significantly related to such things as social class or amount of education. To base decisions about an offender's future treatment on such measures may therefore be unfair.

Further, to base decisions solely on historical factors denies the possibility of change. Some offenders do change, because of the influence of the system, changed circumstances or something that happens within themselves. If we ignore current behaviour, then we assume that offenders' fates are sealed by their past and deny the possibility of rehabilitation.

These observations on the limitations of previous research lead to the conclusion that we must work toward creating a theoretical model that identifies the psychological events and processes that precede recidivism. In addition to historical predispositions, it must include current behaviour and cognitions, and explain why some individuals reoffend and others do not. In short, we need an explanation of the process of recidivism, not just its predictors, and of how an individual's experiences, cognitions and capacities interact to lead to the reversion to criminal acts.

The Unified Recidivism Model (Mark I)

We chose two models to guide our investigation. The first came from our previous study of how offenders interact with their environment, and especially how they cope with their problems.⁽⁶⁾ Although it was primarily designed to study behaviour in prison, this study also looked at how inmates had coped with problems they experienced before imprisonment. Not only was the general level of coping disastrously poor, but there was also evidence that inability to cope was linked to criminal behaviour.

These and other similar results led to the formulation of a "coping-criminality" hypothesis, which states that new criminal offences result from inadequate or destructive ways of dealing with ordinary life problems (there are some additional stipulations, but we can ignore them for now). The data support this hypothesis in several ways. For example, we found that high scores on measures of coping ability were significantly related to less-extensive criminal histories (retrospective) and vice versa. As well, measures of coping ability and associated behaviour from the original study predicted recidivism (prospective) with the same accuracy as commonly used predictive scales.

If we consider that recidivistic behaviour represents a breakdown of the self-control mechanisms that maintain lawful behaviour, then it is comparable to what happens when a substance abuser returns to drug use. The second model we adopted was developed to explain relapses into addictive behaviours, and sees them as largely triggered by negative emotional states, interpersonal conflict and particular thought patterns, along with such things as social pressure.⁽⁷⁾

Although formulated quite independently and in different contexts, these two theoretical statements are clearly compatible. When the models are considered together, it becomes apparent that each emphasizes a different part of the same process and that the two can be joined at the "risk point." If coping difficulties lead to criminal actions, it is likely that the inability to cope produces emotional distress and certain thoughts, which in turn trigger violent or irrational behaviour or reduce self-monitoring and self-control. Conversely, for relapse theory, poor coping efforts probably lead to the critical sequence of emotional responses and cognitions preceding criminal recidivism. Inadequate coping results in stress, and relapse theory describes what happens then.

The Study

On the basis of this reasoning, we decided to investigate the process of recidivism. The principal study was begun with support from the Research and Statistics Branch of the Correctional Service of Canada, and the research is now being enlarged and extended under a grant from the Social Sciences and Humanities Research Council.

The subjects were federal recidivists in the Ontario region. They were randomly selected and recruited as soon as possible after their return to prison, most of them in the Reception Centre at Millhaven Institution. The results included here are limited to the initial sample, which consisted of 100 men whose new offence was violent. This sample is now being enlarged. For controls and comparisons, we are including a group of offenders reincarcerated for property offences, as well as a group of released inmates who have not reoffended.

We included a variety of measures, focusing on problems, emotions and behaviour in the period immediately preceding the new offence. In an attempt to plot the relapse process, there were a series of questions about critical events in the sequence, from the first passing thought of committing a new offence to the point of inevitability. We also gathered some detailed information on other behaviours, such as coping responses, time use and substance abuse. Finally, we included several measures of personal and criminal history. The various measures were obtained through a structured interview, a series of standardized questionnaires and a file search.

Offence Precursors

Historical data on the first 100 subjects indicated that we had recruited a population of serious offenders. More than one third had a current offence of robbery, another third had been convicted of assault and the rest had committed other types of violent crimes. Their previous criminal histories were extensive, varied and violent, with an average of more than 24 previous convictions. They had been in the community an average of about five months since being released from prison, with a range from 15 months to about 15 minutes before rearrest.

Their lifestyles outside prison conformed, in most respects, to what had been described in previous studies. For example, "hanging out" with friends was the largest single category of time use, and family activities occupied much less time on average. Other measures indicated much instability in their lives, with frequent changes in residences, jobs and living partners.

Their emotional state at the time of the interview was even more unstable than we had seen before in offenders starting a prison term. Many subjects showed substantial evidence of depression. About half the sample had scores above 15, the level usually used to indicate a strong possibility of symptoms at a clinical level on the Beck Depression Inventory, the most widely used index of depression. Measures of other states, such as anxiety, also showed that the subjects were having substantial emotional problems.

One would certainly expect emotional distress in reaction to being back in prison, but questions about life on the outside showed very similar patterns, indicating that the problems preceded incarceration and probably preceded the new offences. Two areas stood out as indicators of difficulties in adjustment and as possible predictors of future trouble. The first was substance abuse, the second was emotional stress.

Concerning the first indicator, our figures are similar to previous findings. For example, the level of alcohol consumption averaged more than eight drinks daily for all subjects. (This average included abstainers.)

We also gathered information on the resumption of drinking after release, which shows the ineffectiveness of current control tactics. While more than three quarters of the sample had been released with the condition that they would abstain from alcohol or other drugs, the majority admitted to violating this condition in the first week after release. Indeed, 44% said they had taken a drink of alcohol on their first day in the community.

Substance abuse was also clearly linked to new offences. Two thirds of the sample said they had been drinking in the 24 hours preceding the (first) new offence, and the average amount was over 11 drinks (even higher than the overall daily average for drinkers). Also, measures of substance abuse seemed to differ according to offence type. For example, robbers consumed less alcohol than other offenders but more of other drugs.

More important for our theoretical model are findings about emotional states in the period before the offence. We asked subjects to name their predominant emotional state during the final month; for the great majority, it was an unstable state (or dysphoria). Just less than one quarter of the subjects indicated depression, with anger and anxiety next in frequency.

Given their unstable state at the time of the interview, there may be some retrospective bias colouring the offenders' descriptions of their emotional states before the offence, but the results are very strong and internally consistent.

We then asked them to describe their mood on the day preceding the new offence. Unstable emotions again predominated, although anger was now the most frequent emotion.

We also asked what they thought had driven them to committing the offence. Although one quarter named financial gain as their primary motive, roughly another quarter pointed to anger or frustration.

Our model asserts that emotional disturbances largely result from difficulties coping with problems outside prison. Other results indicate that most subjects were severely limited in their ability to choose and implement effective solutions to common problems. Using a measure of "coping efficacy," none of the present sample had scores as high as the average score for a random sample of non-offenders (correctional workers).⁽⁸⁾

All of this is consistent with the position that both emotional problems and substance abuse are precursors to the resumption of criminal actions, and the results also support the claim that they follow from poor coping ability. Of course, it is impossible to draw any firm conclusions at this time, in the absence of comparable information from other populations. However, since the results from the first of the non-recidivist control groups show what appear to be considerable differences, it may be tentatively concluded that a number of dynamic factors can lead to recidivism.

Offence Sequence

One other result bears mentioning, even at this early stage in the project. In addition to looking at the precursors to offences, we wanted to describe the progression of events in the offence process itself. This is important for several reasons, among them to determine whether we can intervene before a new offence is committed. Much of the interview concentrated on offenders' actions and thoughts about a possible offence before it occurred.

To determine offence sequence, the subject was given a time line. Points were marked along the line, from several months before the offence until its occurrence. Subjects were asked to mark a series of "offence milestones" on the line, from their first passing thought of possibly committing an offence to the point of inevitability when they had already begun the series of actions leading to the offence.

The results can be seen in the table. Whichever milestone one considers, the lack of anticipation is striking. For almost half the subjects, the entire process - from first passing thought to committing the offence - was collapsed into an hour or so. Only about one quarter reported any real planning more than an hour before they committed the offence.

While the impulsive nature of many criminal offences, and criminal offenders, has been known for some time, these data put it into quantitative terms. Not only was there a lack of forethought, other data indicate that previous sanctions failed. For example, only 7 of the 100 said they had thought about the possible negative consequences before committing their new offence, and even fewer considered the victim, despite their histories of previous imprisonment. The implications for models of rational, considered decision making, on which most criminal law is based, appear to be profound.

Milestones in the Offence Process (Percentage of Subjects)
Time Interval Before Offence	First	Longer	consdr	Plan 1	Plan 2	NoRel
One month or more	26	14	7	4	3	0
One week or more (but < 1 month)	11	11	9	5	3	0
One day or more (but < 1 week)	10	12	13	10	11	6
Hours (but < 1 day)	8	11	13	11	9	11
Within an hour of the offence	6	10	8	12	12	7
At the time of the offence	38	44	50	59	62	76
First - First passing thought of offence Longer - First longer thought of offence (more than one minute) Consdr - First time considered might actually commit offence Plan 1 - First thoughts of planning ofence Plan 2 - First definite or concrete plans NoRel - Point of no return

Conclusion

Clearly, there are some interesting things to be seen when one examines the recidivism process of individuals. Although we do not yet have the conclusive evidence that would convince any properly sceptical person, there are indications that emotions and habits play critical roles in the chain of events leading to a relapse into criminal behaviour.

It also appears that the offence process works something like a ballistic missile. Once it is set off, it runs quickly and inevitably on its course. At the same time, the triggering of the chain of events is affected by certain behaviourial and emotional events, and it may be predictable and preventable. Obviously, much remains to be done in this area.

In the meantime, data gathering is ongoing, and we may branch out into further studies of the process. Once one begins looking at something, there's no telling what one will find.

---

⁽¹⁾This article is based on a talk given at the correctional Service of Canada's Third Annual Research Forum at Whistler, British Columbia, in June 1991.
⁽²⁾J. Nuffield, Parole Decision-Making in Canada: Research Towards Decision Guidelines (Ottawa: Supply and Services Canada, 1982).
⁽³⁾J. Monahan, Predicting Violent Behavior: An Assessment of Clinical Techniques (Beverly Hills: Sage, 1981), p.99.
⁽⁴⁾V.L. Quinsey, "Deinstitutionalization Policy and the Identification of Dangerous Individuals: A Literature Review," Criminologie, J 7, 2 (1984): 53-78.
⁽⁵⁾F. Porporino, E. Zamble and S. Higginbottom, "Assessing Models for Predicting Risk of Criminal Recidivism." Unpublished manuscript, Queen's University, 1988.
⁽⁶⁾E. Zamble and F. Porporino, Coping, Behaviour and Adaptation in Prison Inmates (New York: Springer-Verlag, 1988).
⁽⁷⁾G.A. Marlatt and J.R. Gordon, Relapse Prevention: Maintenance Strategies in the Treatment of Addictive Behaviours (New York: Guilford Press, 1985). And see H.M. Annis and C.S. Davis, "Relapse Prevention Training: A Cognitive-Behavioral Approach Based on Self-Efficacy Theory," Journal of Chemical Dependency Treatment, 2 (1989): 81-103.
⁽⁸⁾G.V. Hughes, Personal Disposition, Coping, and Adaptation in Correctional Workers, Ph.D. thesis, Queen's University, 1990.