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Warning This Web page has been archived on the Web.

Offender Assessment: General issues and considerations

by James Bonta1
Corrections Research, Department of the Solicitor General Canada

There are few activities in corrections as important as the assessment of offenders. An accurate assessment facilitates the fair, efficient and ethical classification of offenders. We are currently seeing a convergence in thinking about offender assessment that bridges the traditional concerns of safety, security and offender rehabilitation.

This article presents an overview of what we know about offender risk assessment. Our understanding of criminal behaviour influences our approach to offender assessments. Most theories of criminal behaviour can be grouped into three broad perspectives of criminal conduct, and each of these perspectives suggests different approaches to offender assessment. These three perspectives are: Sociological, Psychopathological, and General Personality and Social Psychological. This article presents evidence which leads to the conclusion that the General Personality and Social Psychological perspective encompass factors (dynamic) which best predict criminal behaviour.

Sociological perspectives

The sociological perspectives proposes that social, political and economic factors are responsible for crime. For example, poverty, lack of employment and educational opportunities, and systemic bias toward minority groups cause frustrations and motivations to engage in crime. These perspectives, in one form or another, say that society creates crime. That is, society is largely responsible for crime and the solution to crime rests in altering the social, political and economic situations of society’s members.

Psychopathological perspectives

Within the psychopathological perspective, people commit criminal acts because there is something psychologically or emotionally wrong with them. Individuals disobey the laws and norms of society because of a neurosis, or they are following the commands of internal voices. They may have too much testosterone that drives them to commit sexual crimes or they have a neurological disorder that results in uncontrolled, violent behaviour. For the psychopathological theories, it does not matter if one is poor or not, from an ethnic minority or a politically powerless group.

General personality and social psychological perspectives

The general personality and social psychological perspectives emphasise the learning of attitudes, emotions and behaviours that lead to criminal conduct. The focus is the individual (like the psychopathological theories) but it is the learning experiences of the person that account for crime. It is not so much that the offender is “sick”, but that the offender was exposed to situations that rewarded and encouraged antisocial behaviour. For example, a child who grows up in a home where the parent(s) allow aggressive and hostile behaviour, model antisocial attitudes and fail to direct the child in prosocial activities (e.g., school) and appropriate friendships, learns antisocial behaviour.

Each of the three perspectives directs our attention to different factors for understanding criminal behaviour. As a consequence, they suggest what should be assessed when dealing with offenders. Table 1 illustrates how the various perspectives forward certain variables as candidates for assessment.

Table 1

The Relationship Between Theory and Offender Assessment
Theoretical Perspective
Example Characteristics Assessed
Sociological
Social status ( e. g., age, gender)
Race and ethnicity
Financial status
Psychopathological
Psychological discomfort ( e. g., anxiety)
Self-esteem
Bizarre thoughts
General Personality& Social Psychological
Peer support for behaviour
Employment instability
Antisocial attitudes
Antisocial personality
Substance abuse
Antisocial behavioural history
High crime neighbourhood

What theory should be chosen to direct offender assessment activities? An evaluation of the evidence in support of a theoretical position is the key for selection among competing theories. A simple and straightforward way of evaluating a theoretical perspective is to see if the factors identified by theory are actually related to criminal behaviour. For example, are financial earnings, ethnicity, “nervousness”, and having criminal friends associated with an individual’s criminal conduct? The research finds that the variables derived from a General Personality and Social Psychological perspective are better predictors of criminal behaviour than the variables suggested by the other theories.

Technical challenges of offender assessment

There are two general approaches to making decisions about the future criminal behaviour of offenders (i.e., recidivism). One approach, often referred to as the clinical method, uses subjective and professional judgements to assess the variables deemed important by theory. The other approach is more objective and leaves less room for subjective interpretation. This second approach is referred to as the structured, actuarial method because it involves statistical, evidence-based estimates of risk.

To illustrate the distinction in approaches, let us use the variable antisocial attitudes. Antisocial attitudes can be assessed in different ways. One can search for evidence of antisocial attitudes during a conversation with the offender (clinical method) or one can administer a paper-and-pencil test of antisocial attitudes (structured, actuarial). In the first case, professional skills and experience are required to elicit and note expressions of antisocial attitudes. The interviewer may vary the questions asked from offender to offender. The problem with this is that the way information is gathered may potentially influence responses and therefore, the reliability of the assessment of antisocial attitudes. In the administration of a paper-and-pencil test, the assessment is conducted in a standard manner. Offenders are asked exactly the same questions and their responses are recorded in the same way for everyone.

In the real world, both approaches are frequently used together. Studies comparing clinical and actuarial methods in the prediction of criminal behaviour, or any behaviour for that manner, usually show that assessments based upon the objective approach tend to be more accurate.2 What do we mean by “more accurate”? In any prediction task, there are four possible outcomes. You can predict that something will happen, and it does. For example, a parole board may predict that an offender is dangerous and the offender actually goes on to commit a violent crime. Or, the board may predict that the offender is no risk to the public and it turns out that the offender makes a successful reintegration into society. You can make mistakes. For example, parole is denied to someone who, on follow-up, commits no new crimes or parole is granted to an offender who re-offends violently.

A problem occurs when people assign different importance to the types of predictions and errors. A prediction strategy that would minimise this type of error would be to predict that all offenders will commit another crime. But, at what cost? Studies suggest that there are large numbers of offenders who will not re-offend. For some, this is a minor problem (“saving one victim is enough”). For others (e.g., civil libertarians, financial managers) it represents a serious social and economic issue.

In general, it is best to think of predictive accuracy in terms of the overall proportions of correct predictions and errors. That is, we need to know how the numbers are distributed across all four possibilities to gain a true appreciation of our predictions. We must also accept the reality that no prediction instrument will be perfect. From our discussion of clinical and actuarial approaches to offender assessment, our starting point for improving predictive accuracy is to use actuarial methods in the measurement of offender characteristics and their situations.

Measuring theoretically relevant factors in an objective, actuarial manner, unfortunately, is not as easy as it sounds. Any measuring instrument will have some error associated with it. Even the trusty ruler that you had since grade school is not 100% accurate. When it comes to the assessment of human factors, the range of error is considerably greater than errors associated with mechanical instruments such as rulers, weigh scales, etc. This is one reason why we can never achieve perfect prediction.

One approach of limiting measurement error is to use different methods for assessing the same factor. Returning to the example of antisocial attitudes, we can measure this variable with a paper-and-pencil test and by way of a structured personal interview. Structured interviews are not open-ended clinical interviews. The structured interview involves an observable and clear method for asking questions and recording the answers. Furthermore, the results from structured interviews can be quantified and evaluated as to their validity.

By using more than one method of assessment, the problems associated with one method of assessment are counter-balanced by another method. For example, a potential problem with a paper-and-pencil measure is that one may not be certain that the offender understood the questions or if he/she was motivated to answer truthfully. In an interview approach, the interviewer can verify whether the questions are understood and gauge the offender’s interest and motivation. Research has shown that when more than one method is used to assess a certain offender characteristic, the overall predictive accuracy improves significantly. These research findings are easily translated into practice and the best correctional practices are seen shown when we use multiple methods (e.g., questionnaires, interviews, direct more behavioural observations).

The objective, multi-method measurement to of theoretically relevant factors is the first step in improving predictive accuracy. certain The second step to improving predictive accuracy is to combine the individual factors to form more comprehensive the offender assessment measures. The combining of factors are usually done in one of two ways. The simplest, called the Burgess method, is to assign a score of 1 if the factor is present and 0 if the factor is absent. Thus, you can have a number of items/factors in a scale that are simply scored (0 or 1) and then summated to give an overall score. The other method uses advanced statistical techniques to assign different weights to the factors. The Level of Service Inventory Revised is an example of an offender assessment instrument using the Burgess method and an example using the weighting method is the Wisconsin Risk-Needs scales. The research evidence however, does not favour one approach of assigning scores over the other.

If we apply multi-method assessments to the different domains or factors related to criminal behaviour and then combine these domains, the prediction estimates increase substantially. Previous research has demonstrated impressive evidence on how multi-method and multi-domain sampling can improve prediction.3 Adult probationers were given an assessment battery that measured different domains and used different methods of measurement. They found that the correlation (r) for antisocial attitudes and recidivism was .46 when a paper-and-pencil measure was used and .63 when it was combined with a structured interview. When this information was combined with other domains (e.g., antisocial personality, criminal history, age), the correlation (Canonical correlation to be precise) increased to a value of .74.

Purposeful use of relevant factors

Risk assessment

Although risk assessment is obviously important for release and security decisions, it also has implications for treatment planning. Appropriate decisions concerning who to place into treatment are informed by offender risk. The risk principle is especially informative for clinicians and treatment staff who have been schooled and trained in therapeutic techniques that are suited to clients who are verbally skilled, reflective and socially skilled.

Although the “talking” and relationship oriented therapies can be helpful to many people, they are not very effective with the typical offender client. Many offenders lack the verbal and thinking skills required by these counselling techniques. Consequently, when therapists practising relationship, verbal therapies meet failure with the offender client, they tend to blame the failure on the client’s “resistance” and “lack of motivation” rather than the technique.

Some observers have long admonished correctional and forensic therapists for preferring to counsel the low risk young, attractive, verbal, intelligent and socially skilled (YAVIS) client rather than higher risk client who really needs the service. Low risk offenders are certainly more pleasant to counsel. Moreover, some of our ideas about criminals make it relatively easy to dismiss efforts for dealing with higher risk offenders (“he’s a psychopath”, “he’s too hard core to change”). The research evidence however, suggests that it is the higher risk client that can benefit from treatment more so than the lower risk offender. Fortunately, the importance of targeting higher risk offenders is permeating throughout the field as more treatment effort is being directed to higher risk offenders.

Needs assessment

One of the important derivations from a general personality and social psychological perspective of criminal conduct is that many of the factors identified as important are dynamic or changeable. An individual can change their attitudes and friends, he/she can find or lose a job, stop taking drugs or begin to drink heavily, and so on. Even antisocial personality features can be changed if we consider antisocial personality in a very broad sense rather than in the narrow sense of a diagnosis of psychopathy. This view of antisocial personality encourages attempts to change a constellation of dynamic offender attributes such as thrill seeking, impulsiveness and egocentrism.

For offender assessment, the theory highlights the importance of objectively and systematically assessing dynamic risk factors. Reviews of the literature show that dynamic risk factors predict recidivism as well as static risk factors.4 More importantly, changes in dynamic risk factors have been associated with changes in recidivism.5

Dynamic risk factors are also referred to as criminogenic needs. Criminogenic needs are those offender needs that when changed are associated with changes in recidivism. The Need principle of effective rehabilitation calls for the targeting of criminogenic needs in treatment programs. From an assessment perspective, the measurement of criminogenic needs is highly important for directing treatment services and for the active supervision of offenders. The evidence is convincing that interventions that target criminogenic needs are associated with reductions in recidivism.6 At present, there are intervention programs that are reasonably effective and assessment instruments that reliably document changes in dynamic risk factors. Some of the assessment instruments are quite specific to a particular criminogenic need (e.g., measures of substance abuse or antisocial attitudes) and other instruments provide more general assessments of offender risk and needs (e.g., the Level of Service Inventory Revised).7

For offenders under community supervision, the monitoring of dynamic risk factors assumes an additional significance. Probation and parole officers need to be attentive to both improvement and deterioration in the offender’s situation. Community supervisors easily note dramatic changes in the offender’s situation. However, more subtle and gradual changes are not so easy to detect. Reliance on subjective, professional judgements of change is difficult to defend when objective, empirically based assessment measures are available. This is especially true when correctional staff can administer many of these measures after brief training. That is, psychologists and psychiatrists are not required to administer risk-need assessment instruments or many of the paper-and-pencil measures of criminogenic needs that are available.

Assessment of Responsivity Factors

How people learn from life’s experiences depends, in part, on certain cognitive, personality and social-personal factors. These factors may, or may not be, offender risk factors or criminogenic needs. They do however, influence the individual’s responsiveness to efforts to help them to change their attitudes, thoughts and behaviours. These responsivity factors play an important role in choosing the type and style of treatment that would be most effective in bringing about a change. A few illustrations of responsivity factors are helpful in understanding this concept.

Our first example is taken from the cognitive domain. Individuals vary in their thinking styles (e.g., concrete vs. abstract, impulsive vs. reflective) and general intelligence. In terms of risk, neither of these two factors are particularly strong risk factors. However, these cognitive factors are very important with respect to learning new thoughts and behaviours. They influence how an individual best profits from instruction and the ease of learning. Two offenders may be of equal risk to re-offend and have the same criminogenic needs but they can differ in their cognitive level and style. One may be more verbally skilled and quicker to grasp complex ideas while the other is less cognitively skilled. The goals of treatment are the same but how one reaches that goal will be influenced by the client’s cognitive responsivity factors. For the more cognitively skilled client, a program that is highly verbal and that requires abstract reasoning skills may be effective. However, this same approach would present a serious challenge for the less cognitively sophisticated offender.

Another example is taken from the personality domain, the trait of anxiety. Once again, a responsivity factor without risk or criminogenic need qualities. Levels of anxiety are poor predictors of recidivism and decreases in anxiety are not associated with reductions in recidivism. Yet, the anxiety levels of offenders could impact on the choice of treatment. For example, an anger management program may work well in a group format consisting of relatively non anxious individuals. For clients who are extremely anxious in social situations however, the program would be more effective if delivered on an individual basis.

Some risk and criminogenic need factors may have responsivity characteristics. For example, offenders described as having an antisocial personality are not only higher risk offenders with many criminogenic needs, but their lack of empathy and anxiety require an intervention approach that is highly structured. Their energetic and restless nature calls for a treatment style that is active and stimulating. Classroom discussions and quiet readings are not the preferred mode of intervention for these types of offenders.

Objective measures of antisocial personality are available with one of the best validated instruments being Hare’s Psychopathy Checklist. Unfortunately, because the Psychopathy Checklist is often used to form a diagnosis of psychopathy, the instrument is not conducive to treatment planning. A diagnosis of psychopathy is often seen as a sign of untreatability. As a result, efforts to treat “psychopathic” offenders is minimal despite the fact that there is no convincing evidence that theoretically relevant interventions will not “work”. In addition, there is no research exploring the role of psychopathy and/or antisocial personality as a responsivity factor.

Psychologists have been extremely successful in developing valid and reliable measures of other responsivity factors. There are many excellent measures of intelligence (e.g., the Wechsler IQ scale), anxiety (e.g., Spielberger’s State-Trait Anxiety Inventory), and interpersonal maturity (e.g., Jesness I-Level). There, is however, a need to develop good measures of impulsiveness, empathy and self-control, to name a few. Clearly, there is much work to be done.

In addition to cognitive and personality characteristics, some personal and demographic characteristics may operate as responsivity factors. Two possible candidates are gender and ethnicity. Female offenders may respond better to a style of intervention that is more women centred. Aboriginal offenders may benefit from a program offered by native counsellors and elders. Although there is no need for assessment measures of personal and demographic characteristics, there is a need for research examining the most effective styles of treatment based on gender and ethnicity factors.

Conclusions

Research in offender assessment holds excitement and promise. Although our prediction instruments will never reach perfection there is still tremendous room to improve predictive accuracy. Research on the assessment of responsivity factors and risk-needs factors specific to certain offender groups (e.g., sex offenders) must become a greater priority. Nevertheless, the momentum exists for continued improvements toward a more effective and humane correctional system.


1. 340 Laurier Avenue West, Ottawa, Ontario K1A 0P8.

2. Grove, W. M. and Meehl, P. E. (1996). “Comparative efficiency of informal (subjective, impressionistic) and formal (mechanical, algorithmic) prediction procedures: The clinical-statistical controversy”.Psychology, Public Policy, and Law, 2, 293-323.

3. Andrews, D. A., Wormith, J. S. and Kiessling, J. J. (1985). Self-reported criminal propensity and criminal behaviour: Threats to the validity of assessment and personality. (Programs Branch User Report). Ottawa, ON: Solicitor General Canada.

4. Gendreau, P., Little, T. and Goggin, C. (1996). “A meta-analysis of the predictors of adult offender recidivism: What works!” Criminology, 34, 575-607.

5. Andrews, D. A. and Wormith, J. S. (1984). Criminal sentiments and criminal behaviour. (Programs Branch User Report). Ottawa, ON: Solicitor General Canada.

6. Andrews, D.A. and Bonta, J. (1998). The psychology of criminal conduct. (2nd ed.). Cincinnati, OH: Anderson.

7. Andrews, D. A. and Bonta, J. (1995). The Level of Service Inventory Revised. Toronto: Multi-Health Systems.