Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

FORUM on Corrections Research

Warning This Web page has been archived on the Web.

So You Want to Know the Recidivism Rate

What's the overall recidivism rate in Canada? It's a common question, but one that's difficult to answer. In this article, we try to explain why this question is so tough to answer and try to answer it as best we can.

What Is Failure?

What constitutes recidivism? Is it a new offence? Is it any return to the criminal justice system or to correctional custody? Does it include any breach of release conditions or does it only include breaches of the most serious conditions?

...Returns to Custody

If we define recidivism as any return to correctional custody, our recidivism group will include a broad range of individuals and circumstances. For example, an individual released on day parole to take part in a three-week work project could fall into this group because, at the end of the three weeks, he or she would be returned to federal custody. Is this person a recidivist? (Actually, this would be a "positive" return to custody because the offender successfully completed the work program - something that doesn't always happen.)

Okay, so what if we say we will include only those cases where conditional release is revoked for technical violations of release conditions or for new offences, or where someone is reincarcerated for a new offence committed after the original sentence has ended.

...Technical Violations

Even with the above modifications to our definition, it may still be too broad. For example, a fraud offender who was told to abstain from alcohol and drugs while on release decides to celebrate the new-found freedom by getting drunk at a party. The police are called by angry neighbours and find out that the offender is on parole. Is this recidivism?

Okay, okay, you say, let's focus on the most serious technical violations. Consider the case of a sex offender whose conditional release conditions stipulate that he or she stay away from school yards where he or she was known to scout for potential victims. If we find out that this sex offender is in fact loitering in school yards, is this recidivism? It is serious, but is it recidivism?

Enough nit-picking, you say, let's focus on other, more serious types of recidivism - new offences for example.

...New Offences

How do we define and measure new offences? Should we use officially reported offences, such as arrests or convictions? Well, we know we can't say an individual committed an offence just because he or she was arrested for it. More important, perhaps, we also can't say that an individual did not commit an offence just because he or she was not arrested for it.

But if we only look at convictions, we run into the problem of underreporting. We know, for example, that people who commit an offence are not always convicted of that offence. In addition, plea bargaining in court - whereby some charges are dropped in return for a guilty plea to another charge - often occurs. Thus, while someone may have committed four separate offences, only one or two convictions will be registered.

Furthermore, charges may be reduced in exchange for a guilty plea. So, while someone may have committed second-degree murder, he or she will only be convicted of manslaughter. This causes problems when we try to evaluate the seriousness of new offences committed by offenders, because the seriousness of the convictions may not reflect the seriousness of the offences actually committed.

Instead of looking at "officially reported" recidivism, we could ask offenders whether they committed a new offence and when. But what if they don't tell us everything, because they either don't want to or can t remember? What if they can't remember exactly when they committed a particular offence? Finding out when recidivism occurs is important so we know whether offenders were under correctional supervision when they committed new offences.

How long should we track offenders to see if they reoffend? Some people prefer shorter follow-up periods because the findings are more current. That is, if you keep track of offenders for three years and then calculate a recidivism rate, some may argue that the findings are not relevant because "we do things a lot differently now than we did three years ago." But, if we only track offenders for six months, our findings may be incomplete because many people commit a new offence years after release. Only when an individual dies do we know for certain that he or she has stopped reoffending. Should we and can we, with our limited research budgets, track people for a lifetime? Either way, length of follow-up will be a compromise, and it won't satisfy everyone.

The Problems of "Overall" Recidivism Rates

Let's pretend we agree on what we mean by recidivism and how we're going to measure it. Do we also agree that the possibility of someone reoffending is affected by such things as their age and criminal history?

If so, then we should also recognize that asking for an allencompassing recidivism rate for the entire offender population is not so simple. While the recidivism rate for one group of offenders may be 20%, it may be 60% for another group, and combining the two rates would reflect neither group accurately.

We run into this problem even when looking at just one group of offenders. Suppose, for example, we want to know the recidivism rate for sex offenders. Well, past research (see, for example, Vol.3, No.4 of FORUM) has shown that the recidivism rate for incest offenders is substantially lower than for pedophiles. While we could calculate a recidivism rate for all sex offenders as a group, it would probably not be very accurate nor useful.

What Is Success?

We run into more problems when we try to evaluate the success of particular correctional programs or correctional systems as a whole.

Is a program successful if offenders who participated in it no longer commit offences related to the problem addressed by the program? Or should correctional programs be evaluated in terms of their ability to discourage offenders from all types of criminal behaviour?

Suppose we are evaluating a program for sex offenders. The program focuses on attitudes and behaviours related to sexual offending. If none of the program participants commit further sex offences, can we say that the program was successful? What if 50% of the program participants commit other offences instead? Do we then say that the program was unsuccessful? Or do we say that the program was successful in what it set out to do, but other factors and needs came into play which were left untreated?

The same holds true for evaluations of correctional systems. If we lived in a perfect world and none of the offenders who went through the system ever committed another offence, could we say that this correctional system was successful? Maybe, but we still wouldn't know what in the system was successful. Was it every aspect of the system? Was it that each offender had a case management officer who made a real difference in his or her life? Was it because broader social systems, such as the economy or social programs, had a positive impact on offenders' lives? Was it that the food in our institutions was so poor that offenders never wanted to come back?

The point is, even if we know the recidivism rate, we still can't be sure what it means and what accounts for it.

An Analogy

To illustrate some of the difficulties we have been examining, let's use a university as an analogy. People go to university for about three or four years (which, incidentally, is the average length of sentence served by federal offenders). We want to look at how successful a particular university is. How would we define success? How do you define success in your own life?

Success could be defined as the number of students who graduate from the university, but then again some programs within the university (such as psychology or sociology) might have fewer graduates than other programs for various reasons (for example, stricter selection criteria or more rigorous examinations), so an overall graduation rate for the university might not tell us much about particular programs.

Success could be defined as the number of graduates who find employment in their career of choice, but what if their career of choice has nothing to do with the education they received at university? What if they graduated in economics and then decided that graphic design offered them better opportunities or made them happier? What if the economy was so poor that they could not get a job, but this had nothing to do with their education? What if the follow-up period in our study was only three years, but many people found a job four years after graduation? What if a lot of graduates from one university quickly found a job in their field, but many of them developed substance abuse problems while at university: could we still say that the university was successful?

Enough Stalling Already

So far this article has been a rather lengthy attempt to show that although "What is the recidivism rate?" is a popular and valid question, it really is a difficult one to answer, and to emphasize that an answer can be misleading if we don't recognize its limitations.

Keeping this in mind, we are now going to look at some recidivism rates and at an overview of how a sample of 1,000 federal offenders did while on release.

Short-term Recidivism

We defined the recidivism rate as the percentage of released offenders readmitted to federal custody during a particular period of study. Readmission was defined as violations of release conditions (i.e., technical violations) and convictions for new offences.

To see how well we have been managing offenders over a certain period of time, we compared the recidivism rates for a number of consecutive short-term (three-month) periods.

We looked at the recidivism rates for each quarter during the last three years, from 1 April 1990 to 31 March 1993. We found that the average recidivism rate for each of the 12 quarters studied has remained fairly constant at 4.8%. This is impressive, given that the number of offenders on conditional release has increased by nearly 10% during the same period, from 8,937 in April 1990 to 9,793 by March 1993.

Looking at the recidivism rates for offenders on different types of release, we found that offenders on full parole did much better than those on mandatory supervision (now called statutory release). The average quarterly recidivism rate for offenders released on full parole was 1.9%; it was 10.8% for those released on mandatory supervision. In other words, there was approximately one parole failure for every five mandatory supervision failures. This ratio remained fairly stable over the 12 quarters.

The proportion of the supervised population readmitted for technical violations during each quarter averaged about 2.8%. This percentage decreased slightly (by less than a percentage point) during the last half of the study period, except for the last quarter.

On the other hand, the proportion of the supervised population readmitted with a new offence increased slightly during the first part of the study, from about 1.6% in the first quarter of 1990-1991 to a high of about 2.9% in the last quarter of 1991-1992. It then stabilized during the last year. On average, 2% of offenders in the community were readmitted for new offences over the last 12 quarters.

Both trends suggest that changes have been modest. We have not gained much ground in the struggle for improved reintegration, but we haven't lost ground either.

Finally, for those who were readmitted, we calculated how much was left of their sentence at the time of release and examined how much of this "potential release time" they spent in the community before being readmitted. If offenders spend more time in the community, we can say we are having some success in keeping offenders in the community longer, even if they ultimately fail.

Unfortunately there is no clear trend regarding time spent in the community. In the middle quarters, offenders spent less time in the community, then more, then less again. The long-term average shows that recidivists were serving half (50%) their potential time in the community on the outside. This average dipped as low as 45% and climbed as high as 53%. This at least supports the view that it can be influenced.

Long-term Recidivism

We also followed up more than 42,000 offenders who were released between 1 April 1975 and 31 March 1985. All offenders were followed up from their date of release until 31 March 1992, for a minimum follow-up period of seven years.

During the 10-year period, 15,418 offenders were released on full parole. Figure 1 shows their outcomes. Of those released on full parole, almost three quarters (72%) completed their sentence without being returned to federal custody. In addition, 3% had been under supervision for at least seven years without being readmitted, a period long enough to be considered successful.

Of the one quarter (25%) who were unsuccessful, half had their release revoked for technical violations of release conditions and the other half were readmitted for a new offence.

During the release and follow-up period, 11,704 offenders successfully completed their parole; that is, their sentence expired while they were still in the community. After their sentence ended, these offenders no longer had to abide by release conditions or meet with parole authorities - they were free. About 1 in 10(11%) of these offenders committed a new offence after their sentence had expired and were returned to the federal correctional system.

During the same period, 27,124 offenders were released on mandatory supervision. Figure 2 shows their outcomes.

Of those released on mandatory supervision, 57% completed their sentence without returning to federal custody. Almost one quarter (24%) had their release revoked for technical violations of release conditions, and about one fifth (19%) were readmitted for a new offence.

We also looked at those released on mandatory supervision who finished their sentence while on release in the community. During the time period under study, about one in three (34%) committed a new offence and were returned to the federal prison system after their sentence had ended.

Tracking Offenders Through the System

Recidivism can also be measured according to the type of conditional release an offender is first released on. In this study, recidivism was defined as any return to federal custody. (This, of course, may understate the actual number of offenders reentering the criminal justice system because some offenders may receive provincial sentences.)

Offenders are often released on day parole before they are released on mandatory supervision or full parole to test their ability to cope on the outside. If offenders on day parole were returned to custody before receiving other forms of conditional release, looking only at offenders released on mandatory supervision and full parole would not reflect the true numbers of those returning to custody.

Thus, when calculating recidivism rates based on conditional release, we must take day parole into consideration. But this is easier said than done.

Because the automated data systems of the Correctional Service of Canada are limited in their ability to track this type of release, studies often have to leave offenders on day parole out of recidivism calculations. For our study we were fortunate to have access to a recent national review of the day parole program for information on offenders on day parole.

We followed up a random sample of 1,000 offenders admitted to federal custody for a new offence between 1 April 1988 and 31 March 1989. Since then, almost 92% of the offenders were released on some form of conditional release. We followed up these offenders until 30 June 1993 to see what type of conditional release they received and how well they did.

We wanted to present a kind of "flow" of conditional release. So, we divided the sample into the following groups: day parole only (received no other type of release), full parole and mandatory supervision.

Note that the full parole and mandatory supervision groups could include offenders who had previously been on day parole and were not returned to custody; that is, they had successfully moved from day parole on to other forms of release. (In fact, for the 406 day parolees on which information was available, more than one third (36.4%) went on to full parole and about 1 in 10 (9.6%) went on to mandatory supervision.)

All in all, we found that about half (47.8%) of the offenders were released on day parole, a little more than a quarter (28.3%) were released on mandatory supervision and 1 in 10 were released on full parole (10.3%).

As shown in Figure 3, the overall readmission rate was 37.1%. The readmission rate for offenders released on mandatory supervision was almost twice the rate for those released on full parole (46.6% versus 25.1%), while the rate for those released on day parole was somewhere in between (41.6%).

Almost one quarter of those who went from day parole to full parole or mandatory supervision were returned to federal custody.

We also looked at whether offenders were readmitted before or after their sentence had ended. The vast majority (95%) were readmitted before their sentence had ended.



Figure 1
Figure 1


Figure 2
Figure 2


Figure 3
Figure 3

Conclusion

We have presented many reasons why it is difficult to come up with one all-encompassing recidivism rate that pleases everyone. Recidivism is a critical issue in corrections, but it is also one of the most difficult issues to address in an easy-to-understand manner.

We hope this article has answered some questions, although it has probably raised more questions than it has answered. If this is the case, please send us your questions and we will do our best to answer them in future issues of FORUM or by direct correspondence. Send your letters to: Frank J. Porporino, Ph.D.
Director General
Research and Statistics Branch
Correctional Service of Canada
340 Laurier Avenue West
Ottawa, Ontario K1A 0P9
Fax: (613) 941-8477.


(1) Authors may be contacted at: Research and Statistics Branch, Correctional Service of Canada, 340 Laurier Avenue West, Ottawa, Ontario K1A 0P9.