Stability and Change: Dynmic Risk Factors for Sexual Offenders

Sexual offences are among the most disturbing of crimes, and the public has considerable concern about the risk posed by sexual offenders in the community. Approximately 1% to 2% of the male population will eventually be convicted of a sexual offence. Follow-up studies, however, have found that once detected, most sexual offenders are never reconvicted for a new sexual offence, even when the follow-up period extends to 20 years. Some offenders, however, are much higher risk to sexually reoffend than others, with the observed recidivism rates varying between 10% and 50%.

There are different methods for distinguishing between the risk levels of offenders. One of my early supervisors recommended an offender assessment system based on two categories: workable, and “no good”. Most current risk assessments are more complicated. Sexual offender risk assessments typically consider a range of risk and protective factors, with the higher risk offenders having more of the risk factors than the low risk offenders.


Risk factors can be classified as “static” or “dynamic”. Static risk factors are unchangeable, historical characteristics, such as prior offences, early childhood problems and age. Dynamic risk factors are those that are potentially changeable and, when reduced, are associated with reductions in the recidivism risk. Dynamic risk factors can be further divided into “stable” and “acute” risk factors. Stable risk factors, also called “criminogenic needs”, 2003) or “causal psychological risk factors”, are relatively enduring attributes associated with chronic recidivism risk. Stable dynamic risk factors are the most appropriate treatment targets, for changes in these traits should be related to enduring changes in recidivism potential. In contrast, the acute risk factors, also called “triggering events” or “contextual risk factors” are short-term states that signal the timing of reoffending.

The majority of research on sexual offender recidivism has focused on static, historical factors. Static factors are useful for long-term recidivism prediction, but have limited utility in many applied contexts. Static factors cannot be changed and, consequently, cannot be used to evaluate whether an offender is getting better or getting worse.

Why Should We Care About Dynamic Risk Factors?

There are several reasons why dynamic factors are worthy of attention by clinicians and researchers: to make a difference, to improve prediction, and to advance scientific understanding.

To Make a Difference It is useful to know that an offender is at high risk to reoffend, but it is even more useful to know how to reduce that risk. Knowledge of acute dynamic risk factors is required to determine what needs to be restricted and monitored on community supervision. Knowledge of stable risk factors is required for offenders to know what they need to change, and for therapists to know what should be addressed in treatment.

To Improve Prediction

Even for assessments that are only concerned about the probability of reoffending, the consideration of current characteristics improves predictive accuracy over that provided by static risk factors alone. Not infrequently, experts within the criminal justice system are asked whether an offender has made sufficient changes to justify release from some form of sanction. Such assessments can be meaningful only when they are based on stable dynamic risk factors. A long history of bad behaviour may be sufficient to identify an offender as potentially high risk, but evaluators need to consider dynamic risk factors when evaluating changes in risk levels.

The existing actuarial risk scales are moderate to strong predictors of sexual recidivism, but none of them include all relevant risk factors. Thornton, for example, found that sexual recidivism in his sample was strongly predicted by Static-99, an actuarial risk scale containing items such as prior sexual offences, victim characteristics and age. Despite the strong predictive accuracy of static factors, overall prediction was significantly improved by considering stable dynamic factors such as pro-offending attitudes, hostility and problems in self-regulation. Similarly, Hanson and Harris found that dynamic risk factors significantly differentiated sexual recidivists and non-recidivists after considering the static, historical factors contained in the Violence Risk Appraisal Guide, the Rapid Risk Assessment for Sex Offence Recidivism and the Static-99. Evaluators who rely only on static, historical factors are neglecting potentially important information.

For Scientific Understanding

Even if it was possible to predict future behaviour using purely static factors, the advancement of scientific understanding requires the identification of dynamic risk factors. Static factors make unsatisfying scientific explanations. Even the worst offenders are not offending all the time; consequently, explanations of sexual offending must rely on latent potentials that are activated by proximal triggers.

Consider the well established relationship between prior sex offences and future sex offences: in what sense do prior sex offences “cause” future offending? Most scientific explanations require that causal factors are contiguous with their effects. Scientific theories typically do not accept that events that ended years ago cause future events. It would be unacceptable for a fire investigator to conclude that the current fire was caused by a similar fire on the same location ten years ago. An explanation of the association between prior and future sex offences requires an appeal to enduring characteristics or propensities of the individual, such as “habit strength”, “insensitivity to sanctions”, or “deviant sexual interests”. An observed relationship does not specify which combination of conditions is linked to sexual offending, only that these conditions have re-occurred. A prediction based only on static risk factors is an admission of ignorance.

If you ask most people what makes a plant grow, they will talk about sunshine, “good soil” and water. Botanists can provide much more detailed explanations, referring to genes, photosynthesis and phosphorus. Although no explanation is ever complete, good explanations provide answers to common problems and suggest new possibilities. The most satisfying explanations not only solve specific problems, but do so using concepts that are compatible with other domains of knowledge. Just as good explanations for plant growth should conform to the laws of physics, good explanations for sexual recidivism should be consistent with what we know about human nature.

What Are People Like?

The prediction of human behaviour is difficult because we are not entirely predictable. Most conceptions of human nature include a concept of free will or choice. Our choices, however, are not entirely “free”: they are constrained by our environments and our individual characteristics. Humans live in societies, and many of our most meaningful choices concern how we relate to other people.

One useful way of describing individual differences is through the concept of “schema”, defined `a la Piaget, as perceptual-motor sequences. Schemata can be thought of as latent cognitive structures that embody our past experiences and serve to guide ongoing perceptions and behaviours. Our simplest schemata involve physical actions, but we soon develop schemata for all forms of complex social interactions. Most of our perceptions are fused with values and implied actions: “A postal box is where you mail a letter”, “There is a great bar on Elgin Street with cheap drinks and great music.” Thinking is a form of doing, and each time we enact a schema, it becomes increasingly believable and real.We share common schemata based on similar biology and culture. We develop different schemata based on our efforts to understand our own unique experiences, and as a result of the limitation that attention can only be focused on some of the available options.

What Are Sexual Offenders Like?

Sexual offenders, like everybody else, choose their conduct based on their perception of the options available. Sexual offenders differ from many other people, however, by perceiving certain situations as ones in which a sexual crime is a legitimate option. Later, sexual offenders may wonder why they did it but, at the time, the sexual offence was perceived as something they could do. Sexual offenders who are at high risk for recidivism would be those whose sex offence schemata are readily accessible. Many situations would invoke their urge to offend, such as the sight of potential victims, or common internal states, such as frustration and loneliness. By practising their deviant sexual crimes in fantasy, they increase the probability of eventually enacting them. When they are able to question their habitual patterns of thought, the probability of recidivism decreases. An outline of this model is presented in Figure 1.

At the centre of the model are the deviant plans/scripts/schemata, surrounded by the major triggers. When the schema is activated, offenders will be focusing or ruminating on sexually deviant fantasies or behaviour. They may be aware of little else, and it would be difficult for them to perceive other alternatives: perhaps they are rehearsing a specific sexual crime scenario; perhaps they have an intense urge for sexual release; perhaps they are scanning the environment for potential victims. The enactment of the schema will last for a period of time and then stop. The schemata are likely enacted many, many times, and only rarely will the enactment result in an actual sexual offence. Each time it is enacted, however, it becomes stronger and more accessible. Each time it is interrupted or questioned, it becomes weaker. It is unlikely to ever go away completely.

The specific factors that trigger deviant schemata will vary across offenders. Nevertheless, there are some common risk factors, the most obvious of which is the presence of potential victims. Other potential triggers include subjective distress, conflicts in intimate relationships and sexual arousal. In this model, schemata are distinguished from the offender’s explicit attitudes and values.

Figure 1: A model of recidivism risk among sexual offenders.

Attitudes are considered relatively conscious beliefs about what is and what ought to be. For example, some child molesters state that children want to have sex with adults, and that there is nothing really wrong with adult–child sexual contact. For the evaluation of sexual offender recidivism risk, the most important attitudes are permission-giving beliefs that are tolerant of sexual crimes.

Schemata exist within networks such that reciprocal activation is common. Thinking that it is OK to have sex with children may trigger images of sexually provocative children, which may trigger the urge to have sex with a particular child, which may trigger the thought that it is OK to have sex with children. As a reminder of these reciprocal relationships, most of the arrows in Figure 1 are bi-directional. There is one important exception: the arrow going from subjective distress to deviant fantasies is drawn in only one direction. Distress tends to invoke deviant schemata more often than deviant schemata invoke distress. Relapse prevention theory posits that minor lapses, such as deviant fantasies, should invoke subjective distress, but there appears to be little evidence for a classic abstinence violation effect among sexual offenders.

Another important facet to this model is the offender’s capacity for self-regulation and their motivation to change. Habit change is difficult, and offenders vary on the extent to which they are able to self-regulate. Offenders wishing to change need to become familiar with their deviant schemata, the urges that pull them, the situations in which they are invoked, and what they do to disengage them. Much of cognitive-behavioural therapy involves methods for identifying and disengaging deviant schemata. Although selfregulation deficits can directly lead to sexual crimes, self-regulation is particularly important for offenders wishing to change established patterns.

How Can You Tell That Something is a Dynamic Risk Factor?

There are two broad approaches to identifying risk factors: idiographic and nomothetic. The idiographic approach looks for patterns within individuals whereas the nomothetic approach looks for patterns among groups. The idiographic approach typically tries to identify the factors present at the time of offending and absent at the time of non-offending. For example, a rapist may start cruising for victims after having conflicts with his intimate partner. Identifying the significant idiographic factors is difficult because offenders often have many life problems and there are often few known offences from which to infer patterns.

If the schema theory is correct, then the idiographic approach could also productively focus on what offenders do to make themselves “feel like” offending. Although offenders often describe their offences as “just happening”, it is possible to become aware of, and take responsibility for, the thoughts, intentions and emotions that direct behaviour.

The simplest approach to assessing schemata is to ask the offenders. Many are able to clearly articulate what “turns them on” and gives them the urge to offend. Valuable information can also be gained by examining the circumstances of previous offences and the accounts of victims. Specialized testing can also help offenders develop insight into what “hooks” them. Phallometric assessment has long been used to assess sexual arousal to deviant stimuli; more recently, Gene Abel has been working to create a standardized test based on the natural propensity of attractive models to capture our attention. In general, it would be possible to craft a range of scenarios to probe for the unique circumstances that engage deviant schemata.

Recent research has suggested that standardcognitive psychology paradigms can be used to assess the deviant schemata of sexual offenders. The Implicit Attitude Test was used to assess child molester’s latent association between sex and children, a procedure that involves training specific responses to categories of stimuli. For example, subjects practise pushing a button with their right hand when presented with words related to children, and with the left hand when presented with words related to adults. In the next stage, they push the right hand button when they see words related to sex and the left hand button when they see neutral words. After the responses are trained, the instructions are reversed: now they push the left hand button when they see sex words and the right hand button for neutral words. The basis of the test is that reaction times are quicker when there is an implicit association between the concepts. As expected, Gray et al. found that child molester’s reaction times were quickest when they pressed the right hand button for both sex and children. For the comparison group, the reaction times were quickest when sex was paired with the concept of “adult”.

A related cognitive paradigm is the Stroop task. Subjects are presented with a list of words printed in different colours, with the instruction to say out loud the colour of each word. By changing the content of the words, it is possible to increase or decrease the time it takes individuals to say the colour names. The changes in naming speed are explained by interference created when subjects have to inhibit highly accessible schemata in order to respond to the requested task of naming the colours. Research has found, for example, that violent offenders are slow at naming the colour of words related to violence, that smokers are slow at naming the colour of words related to smoking, and that sexual offenders are slow at naming the colour of words related to sexual crimes.

The research on the cognitive paradigms is preliminary, and may never lead to practical applications. This research is important, however, because it demonstrates the types of enduring propensities associated with recidivism risk. Most of the time deviant schemata are latent, but they can rapidly structure perception given even minimal cues. They are sufficiently fast and automatic that they attract little awareness.With training, however, it is possible for offenders to identify the times when they have “lost it” and succumbed to habitual patterns. With continued training, they can spend less time being hooked, and can increasingly take responsibility for their behaviours, thoughts and feelings.

Group Differences Between Recidivists and Non-Recidivists

Idiographic approaches to identifying risk factors have considerable clinical utility but are poorly suited to some assessment questions. For example, evaluators are often tasked with identifying whether an offender is “high” risk to reoffend, with high risk defined relative to other offenders, or as a recidivism probability. Answers to such questions require group data: specifically, they require follow-up studies that compare the recidivism rates of sexual offenders with or without particular characteristics.

We recently summarized the findings of 95 recidivism follow-up studies using meta-analysis. Table 1 displays the dynamic risk factors that have received the strongest research in these follow-up studies. The results are reported in terms of d, the standardized mean difference. The d statistic is a measure of how much the recidivists are different from the non-recidivists, and compares that difference to how much recidivists and non-recidivists differ among themselves. According to Cohen, d values of 0.20 are “small”, 0.50 are “moderate” and 0.80 are “large”.

Stable Risk Factors

Compared to non-recidivists, sexual recidivists are more likely to show signs of deviant sexual interests, particularly sexual interest in children and paraphilias. Although sexual interests tend to be rather stable, it is not unusual to find individuals whose sexual orientation has changed throughout their lifespan. The effectiveness of psychological interventions for changing sexual preferences remains controversial.

Sexual offenders often report high levels of diverse sexual activity, and sexual preoccupations increase the risk of sexual recidivism.

Table 1: Potential Dynamic Predictors of Sexual Recidivism

	Average effect size (d)	Sample size (studies)
Sexual deviancy
Any deviant sexual interest	0.31	2,769 (16)
Sexual interest in children (phallometric)	0.32	1,140 (7)
Paraphilic interests	0.21	477 (4)
Sexual preoccupations	0.39	1,119 (6)
Antisocial orientation
General self-regulation problems	0.37	2,411 (15)
Antisocial personality disorder	0.21	3,267 (12)
Impulsivity, recklessness	0.25	775 (6)
Employment instability	0.22	5,357 (15)
Hostility	0.17	1,960 (9)
Attitudes
Attitudes tolerant of sexual crime	0.22	1,617 (9)
Intimacy deficits
Conflicts in intimate relationship	0.36	298 (4)
Emotional identification with children	0.42	419 (3)

Source: From Hanson & Morton-Bourgon (2004). Predictors of sexual recidivism: An updated meta-analysis. Corrections Policy User Report No. 2004–02. Ottawa: Corrections Policy, Public Safety and Emergency Preparedness Canada.

Such findings suggest that problems with sexual self-regulation form a core deficit associated with sexual offending. We all need to develop strategies for managing our sexual impulses, and sexual offenders would represent individuals whose sexual self-management skills are at the low end of the continuum. It is not unusual for “normal” adult men to have some sexual interest in deviant sexual behaviour, such as voyeurism, young girls or frottage. For non-offenders, however, the attractions are weak and fleeting. In contrast, many sexual offenders developed their deviant urges through rumination and masturbation fantasies, and by creating opportunities to enact their deviant desires.

Problems with sexual self-regulation can be understood within the larger context of general self-regulation problems and antisocial orientation. The association between low self-control and crime is so strong that Gottfredson and Hirschi considered it to be the cause of crime. Individuals who commit crimes tend to change jobs and residences, have unrealistic plans for the future, and engage in a variety of high risk behaviours. In addition to low self-control, the other major indicator of antisocial orientation is hostility—often expressed as a chronic grievance against the world and those in it.

Antisocial orientation may directly result in sexual offences, but it is particularly important for those who have deviant sexual interests. An individual may find young boys sexually attractive, but never act on this attraction given sufficient selfcontrol and good judgement. In contrast, individuals with an antisocial orientation may feel that they cannot control their impulses, and, besides, why should they? The world owes them something.

Attitudes tolerant of sexual crime showed a small, but significant, relationship to sexual recidivism. It is important to distinguish between believing that sexual offending is OK and efforts to minimize culpability. When we are caught doing something wrong, wetypically struggle tofindaccounts that mitigate the negative social consequences of our transgressions. The most common strategies for diverting culpability are to deny that we did the act, or to minimize the consequences; sexual offenders are no different. Evaluators wishing to differentiate between pro-offending attitudes and defensiveness may benefit from considering the offender’s opinions about sexual offences committed by others. Attempting to justify one’s own transgressions is quite different from believing that it is acceptable for others to do the same thing. Offenders who deny their offences are at least admitting that sexual offending is wrong.

Some of the most promising dynamic risk factors involve intimacy deficits. The lack of an intimate partner increases the risk of recidivism as do conflicts within an existing intimate relationship . Such findings are not surprising considering the strong natural links between sex and intimacy. Most people want their sexual partners to be likeable and attractive, and for their sexual partners to find them likeable and attractive. Sex deepens pair bonding, whether or not increased intimacy is even desired. Sexual offences, by definition, involve a disruption in the normal process of mutual sexual attraction.

Given that most sexual offenders have had difficult childhood environments, it is likely that their relationship problems started early and shaped the development of their sexual interests and activities. For rapists, adversarial or impersonal sexual relationships can be an extension of adversarial and impersonal relationships with family and peers. Child molesters, in contrast, may be attracted to immature, childish relationships, and may feel very much like children themselves. Marshall has made an important contribution by focusing attention on the need to address intimacy deficits in the treatment of sexual offenders; further research is required, however, to determine the extent to which improved intimacy is associated with medium- or long-term reductions in recidivism risk.

Acute Risk Factors

The research on acute risk factors is much less developed than the research on static or stable factors. Acute risk factors are harder to assess than stable factors because measurements must be obtained just before the recidivism event. Pithers and colleagues used retrospective interviews with the offenders to identify precursors of sexual offending. In their research, the risk factor most frequently mentioned by the offenders was negative mood. Researchers at Institut Philippe Pinel asked inmates to keep daily records of interpersonal conflicts, negativemoodand deviant sexual fantasies. They, too, found that negative mood increased the likelihood of deviant sexual fantasies and masturbation.

Hanson and Harris used records from community supervision officers to identify the precursors of sexual recidivism. By comparing the factors present in the month prior to recidivism with the factors present six months previously, it was possible to identify the changes associated with recidivism risk. The acute factors suggested by this research including negative mood, hostility, substance use, victim access, sexual preoccupations, problematic social relationships and lack of cooperation with supervision. One weakness of the study was that most of the information was based on interviews with the supervising officers, which raises the possibility of retrospective recall biases. Subsequently, we have initiated a large prospective study of stable and acute predictors of recidivism during community supervision, with results expected in 2006.

Using Dynamic Factors to Predict Recidivism

Although there is sufficient research evidence to indicate that dynamic risk factors contribute information not captured by static risk variables, we have much to learn about how to integrate dynamic risk factors into an overall evaluation. How is it possible to predict future events using characteristics that are inherently changing?

Consider, for example, three offenders with repeated assessment on “defiant hostility”, rated from 1 to 10, with 10 indicating “extremely defiant, hostile”. Tom was very hostile at Time 1, Time 2 and Time 3, but displayed only mild hostility at Time 4; Dick was never very hostile; and Harry was usually agreeable, but showed moderate levels of hostility at Time 4. Which of the three offenders is at highest risk to recidivate at Time 5? Considering only the information at Time 4 suggests that Harry is higher risk than the other two. Considering prior assessments suggests that Tom is the highest risk. How then should evaluators appropriately integrate present and past evaluations?

The short answer is we don’t know. Although a number of potentially changeable factors have been reliably associated with recidivism, we know very little about how changes in such variables should be considered to change estimations of risk. I will, however, provide concepts that are worth considering when broaching this question.

Stability of the Characteristic

Characteristics vary in their stability across time, and in the stability of their relationships with recidivism. Some degree of stability is required in order for there to be an association between the characteristic and subsequent recidivism. If the characteristic is of short duration and risk is only increased when the factor is present at the time of the offence, then the most recent assessment should provide the most information. If the link with recidivism is indirect or the characteristic is highly stable, then predictions are most likely to be accurate when they consider the offender’s baseline functioning.

Table 2: Three Hypothetical Patterns of Change

	Time 1	Time 2	Time 3	Time 4	Time 5
Tom	8	9	8	2	?
Dick	2	3	2	3	?
Harry	2	2	1	5	?

Note: Hypothetical scores on “defiant hostility”, rated from 1 to 10, with 10 indicating “extremely defiant, hostile”.

Baseline Level of Functioning

Offenders will vary as to their “typical” level of functioning. One of the important questions concerns the length of time required in order to establish a baseline, a question that is open to empirical investigation. Consider a study that conducted monthly assessments of “hostility” over a period of ten years. All things being equal, it is likely that the most recent assessment would be the most informative, followed by the previous, then the one before that. Eventually, consideration of the nth prior assessment would provide no new information. For rapidly changing risk factors, the current month’s evaluation would be the most informative. For highly stable factors, the most recent assessment may be less informative than an average of the offender’s functioning in preceding months or years. It is even possible that the first few assessments may provide the most valid information, with all subsequent assessments contaminated by response bias or other artefacts. For highly stable factors, any current deviation from baseline level of functioning is likely due to measurement error or temporary transient factors. If the attribute is highly stable, then an unusual reading is likely to revert back to baseline levels prior to the opportunity to recidivate.

From studies of habit change, we know that relapse is most likely in the first few months following cessation, and that there are relatively few new cases of relapse after two years of abstinence. It is likely that many other behaviours follow a similar pattern in that several years are required to generate a new baseline.

Reversibility of Change

Once an individual has changed, how easy is it for the individual to revert back to previous modes of functioning? The reversibility of change is highly related to the stability of the characteristic, but the concepts are not identical. It is possible for a highly stable feature to change quickly to a new level. For most criminogenic features, however, there are few barriers to reverting back to prior ways of functioning. Consequently, offender’s recent changes are typically viewed with suspicion.

Deviant vs Non-deviant Levels of Functioning

Most problematic behaviour exists along a continuum. For some criminogenic needs, it is possible to specify a non-arbitrary distinction between problematic and acceptable functioning. Change within deviant or within non-deviant levels of functioning may have less significance than changes between levels. For example, consider deviant sexual interests indexed by a ratio of deviant to non-deviant arousal. If the non-deviant arousal remains constant at 30 mm, a reduction in arousal to children from 40 mm to 20 mm would be more meaningful than a reduction from 60 mm to 40 mm.

Implications for Research

One important line of research would be to examine the stability of dynamic risk factors. For evaluators, definitions of change need to include some criteria for establishing the stability of the observed changes. If the observed changes are highly stable, then it would be possible for the recent assessments to completely “write over” previous assessments. Most people, however, are unlikely to completely escape from their past actions, although relapses into deviant levels of functioning can become increasingly rare over time.

One common paradigm for identifying dynamic risk factors is to examine whether post-treatment evaluations predict better than pre-treatment evaluations. Although interesting, such designs are poorly matched to the clinical task. The basic task of prediction is to determine the probability that the characteristics necessary for sexual offending will be present at some point in the future when the opportunities for offending are also present. Clinicians need to consider past and present functioning, but the fundamental task of predicting future sexual crimes requires judgements concerning the offender’s typical or baseline level of functioning in the months and years after treatment has been completed. Estimating the re-emergence or persistence of criminogenic needs is quite different from knowing how the offender is doing in the last treatment session.

In principle, evaluations made post-treatment should be more accurate than pre-treatment evaluations because more information is available. In practice, however, post-treatment evaluations have poor predictive accuracy. It may be that the evaluators conducting the post-treatment assessments would be equally poor at assessing recidivism risk pre-treatment, but it is also possible that evaluators give undue importance to recent within-treatment behaviour. Although most post-treatment risk assessments have little relationship with recidivism, there are examples in which post-treatment evaluations provided moderate levels of predictive accuracy. These studies examined highly structured approaches to risk assessment in which the criteria for improvement were empirically based and objectively recorded. Given that structured, empirically based risk assessments are frequently superior to unguided clinical judgement, it is likely that increasing structure could improve the validity of post-treatment evaluations.

Dynamicfactors provide the direction for advancing both the science and practice of risk assessment. A number of promising dynamic factors have been identified, and more will be discovered. Although we know that changeable factors, such as victim access and sexual preoccupations, contribute information not captured by purely static factors, little is known about how to combine static, stable and acute factors into an overall evaluation. Static and stable factors are correlated, which means that the degree of risk suggested by static factors will typically agree with the degree of risk suggested by the stable factors. Questions arise, however, when static and stable risk factors disagree. Given the solid research base establishing the validity of static risk factors, prudent evaluators will still rely heavily on static risk factors for assessing long-term recidivism potential. For many questions, evaluators must use dynamic factors even if the research evidence is inadequate because only changeable factors can address questions concerning change.

With increased scientific understanding, static factors will become less and less important. When evaluators are able to accurately identify the causes of recidivism, the practise of purely mechanical prediction using static factors will become a historical footnote.

Conclusions

Dynamicfactors provide the direction for advancing both the science and practice of risk assessment. A number of promising dynamic factors have been identified, and more will be discovered. Although we know that changeable factors, such as victim access and sexual preoccupations, contribute information not captured by purely static factors, little is known about how to combine static, stable and acute factors into an overall evaluation. Static and stable factors are correlated, which means that the degree of risk suggested by static factors will typically agree with the degree of risk suggested by the stable factors. Questions arise, however, when static and stable risk factors disagree. Given the solid research base establishing the validity of static risk factors, prudent evaluators will still rely heavily on static risk factors (e.g. age, the number of prior offences) for assessing long-term recidivism potential. For many questions (e.g. conditional release), evaluators must use dynamic factors even if the research evidence is inadequate because only changeable factors can address questions concerning change.

With increased scientific understanding, static factors will become less and less important. When evaluators are able to accurately identify the causes of recidivism (i.e. criminogenic needs, triggers), the practise of purely mechanical prediction using static factors will become a historical footnote.

Author Notes

I would like to thank Jim Bonta and Andrew Harris for shaping my ideas about dynamic risk factors. The views expressed are those of the author and are not necessarily those of Public Safety and Emergency Preparedness Canada.

References

• Abel, G.G., Jordon, A., Hand, C.G., Holland, L.A. & Phipps, A. (2001). Classification models of child molesters utilizing the Abel Assessment for sexual interest. Child Abuse and Neglect, 25, 703–18.
• Andrews, D.A. & Bonta, J. (2003). The Psychology of Criminal Conduct, 3rd edn. Cincinnati: Anderson.
• Beech, A., Erikson, M., Friendship, C.&Ditchfield, J. (2001).A Six-year Follow-up of Men going through Probation-based Sex Offender Treatment Programmes (Findings #144). London, UK: Home Office.
• Beech, A.R., Fisher, D.D.&Thornton, D. (2003). Risk assessment of sex offenders. Professional Psychology: Research and Practice, 34, 339–52.
• Beech, A. & Ward, T. (2004). The integration of etiology and risk in sexual offenders: a theoretical framework. Aggression and Violent Behavior, 10, 31–63.
• Caspi, A., Moffit, T.E., Silva, P.A., Stouthamer-Loeber, M., Krueger, R.F. & Schmutte, P.S. (1994). Are some people crime-prone? Replications of the personality-crime relationship across countries, genders, races, and methods. Criminology, 32, 163–95.
• Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences, 2nd edn. Hillsdale, NJ: Lawrence Erlbaum.
• Cortoni, F. & Marshall, W.L. (2001). Sex as a coping strategy and its relationship to juvenile sexual history and intimacy in sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 13, 27–43.
• Gottfredson, M.R. & Hirschi, T. (1990). A General Theory of Crime. Stanford, CA: Stanford University Press.
• Gray, N.S., MacCulloch, M., Smith, J. & Snowdon, R.J. (2002, September). Abnormal associations between children and sex in paedophiles as measured by the covert association task (CAT). Presentation at the 7th Biennial Conference of the International Association for the Treatment of Sexual Offenders, Vienna.
• Hanson, R.K. (1997). The Development of a Brief Actuarial Risk Scale for Sexual Offense Recidivism. (User Report 97–04). Ottawa: Department of the Solicitor General of Canada.
• Hanson, R.K. & Bussi`ere, M.T. (1998). Predicting relapse: a meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348–62.
• Hanson, R.K. & Harris, A.J.R. (2000). Where should we intervene? Dynamic predictors of sex offense recidivism. Criminal Justice and Behavior, 27, 6–35.
• Hanson, R.K. & Morton-Bourgon, K. (2004). Predictors of sexual recidivism: an updated meta-analysis. Corrections Policy User Report No. 2004-02. Ottawa: Corrections Policy, Public Safety and Emergency Preparedness Canada.
• Hanson, R.K., Scott, H. & Steffy, R.A. (1995). A comparison of child molesters and nonsexual criminals: risk predictors and long-term recidivism. Journal of Research in Crime and Delinquency, 32, 325–37.
• Hanson, R.K., Steffy, R.A. & Gauthier, R. (1992). Long-term Follow-up of Child Molesters: Risk Prediction and Treatment Outcome. (User Report No. 1992-02.) Ottawa: Corrections Branch, Ministry of the Solicitor General of Canada.
• Hanson, R.K. & Thornton, D. (2000). Improving risk assessments for sex offenders: a comparison of three actuarial scales. Law and Human Behavior, 24, 119–36.
• Harris, A.J.R. & Hanson, R.K. (2003). The Dynamic Supervision Project: improving the community supervision of sex offenders. Corrections Today, 65, 60–2, 64.
• Howes, R.J. (2003). Circumferential change scores in phallometric assessment: normative data. Sexual Abuse: A Journal of Research and Treatment, 15, 365–75.
• Kafka, M.P. (2003). Sex offending and sexual appetite: the clinical and theoretical relevance of hypersexual desire. International Journal of Offender Therapy and Comparative Criminology, 47, 439–51.
• Krall, E.A., Garvey, A.J. & Garcia, R.I. (2002). Smoking relapse after 2 years of abstinence: findings from the VA Normative Aging Study. Nicotine and Tobacco Research, 4, 95–100.
• Langan, P.A., Schmitt, E.L. & Durose, M.R. (2003). Recidivism of sex offenders released from prison in 1984. Bureau of Justice Statistics NCJ 198281.Washington, DC: US Department of Justice.
• Marques, J.K., Day, D.M., Wiederanders, M. & Nelson, C. (2002, October). Main effects and beyond: new findings from California's Sex Offender Treatment and Evaluation Project (SOTEP). Paper presented at the 21st annual conference of the Association for the Treatment of Sexual Abusers, Montreal.
• Marshall, P. (1997). The prevalence of convictions for sexual offending. Research Finding No. 55. Research and Statistics Directorate. London: Home Office.
• Marshall, W.L. (1989). Invited essay: Intimacy, loneliness and sexual offenders. Behaviour Research and Therapy, 27, 491–503.
• McKibben, A., Proulx, J. & Lusignan, R. (1994). Relationship between conflict, affect and deviant sexual behaviors in rapists and child molesters. Behaviour Research and Therapy, 32, 571–5.
• Piaget, J. & Inhelder, B. (1969). The Psychology of the Child. New York: Basic Books.
• Pithers,W.D., Kashima, K., Cumming, G.F., Beal, L.S. & Buell, M. (1988). Relapse prevention of sexual aggression. In R. Prentky & V. Quinsey (eds), Human Sexual Aggression: Current Perspectives (pp. 244–60). New York: New York Academy of Sciences.
• Proulx, J., McKibben, A. & Lusignan, R. (1996). Relationship between affective components and sexual behaviors in sexual aggressors. Sexual Abuse:AJournal of Research andTreatment, 8, 279–89.
• Quinsey, V.L., Harris, G.T., Rice, M.E. & Cormier, C.A. (1998). Violent Offenders: Appraising and Managing Risk.Washington, DC: American Psychological Association.
• Quinsey, V.L., Khanna, A. & Malcolm, P.B. (1998). A retrospective evaluation of the Regional Treatment Centre Sex Offender Treatment Program. Journal of Interpersonal Violence, 13, 621–44.
• Smith, P. & Waterman, M. (2003). Processing bias for aggression words in forensic and nonforensic samples. Cognition and Emotion, 17, 681–701.
• Smith, P. & Waterman, M. (2004). Processing bias for sexual material: the emotional Stroop and sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 16, 163–71.
• Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–62.
• Thornton, D. (2002). Constructing and testing a framework for dynamic risk assessment. Sexual Abuse: A Journal of Research and Treatment, 14, 139–53.
• Ward, T., Hudson, S.M. & Marshall, W.L. (1994). The abstinence violation effect in child molesters. Behaviour Research and Therapy, 32, 431–7.
• Wertz, J.M. & Sayette, M.A. (2001). Effects of smoking opportunity on attentional bias in smokers. Psychology of Addictive Behaviors, 15, 268–71.
• Wilson, R.J. (1999). Emotional congruence in sexual offenders against children. Sexual Abuse: A Journal of Research and Treatment, 11, 33–47.