Risk Assessment of Sex Offenders

• Author: Howard Barbaree, Calvin M. Langton, Andres Gopnik-Lewinski, and Craig A. Beach
• Pages: 783-805



CHAPTER 37

Risk Assessment of Sex Of‌fenders

Howard Barbaree, Calvin M. Langton, Andres Gopnik-Lewinski, and Craig A. Beach

I. INTRODUCTION

Over the past thirty-ve years, forensic psychiatry and psychology have devoted much energy and atten-

tion to the task of assisting courts and other decision-making bodies (e.g., review or parole boards) in

evaluating risk for recidivism in sex oenders. ese assessments have addressed the oender’s suit-

ability for release from custody, or whether or not they meet criteria contained in legislation governing

dangerous or long-term oender status (or sexually violent predator legislation in the United States) (see

Chapter 28: Dangerous and Long-Term Oenders). During this same period, a large volume of empirical

research, conducted primarily by forensic psychologists, has led to signicant advances in the assessment

of the sex oender and to the development and promulgation of numerous assessment instruments that

are demonstrably predictive of recidivism among adult male sexual oenders (Doren, 2002; Hanson &

Morton-Bourgon, 2007). e present chapter provides an introduction to evidence-based risk assess-

ment of the sexual oender.

A. Statistical Prediction

Monahan and Steadman (1994) formalized the notion of statistical prediction of recidivism by dening

three closely related component parts: risk factors, the variables that predict violence; harm, the outcome

being predicted, including the nature and seriousness of the sexual violence to be predicted; and risk level,

the probability that the individual will go on to engage in crimina l sexual behaviour. Monahan and Stead-

man (1994) argued that both harm and risk level should be construed as continuous rather than dichot-

omous variables. is seminal argument changed the emphasis in assessment work from a dichotomous

determination of dangerousness (i.e., is he or isn’t he) to an estimate of risk as a probability or a likelihood

of violence and its potential severity based on an evaluation of the presence of absence of risk factors.

i. Risk factors

Risk factors are those characteristics of the oender and his physical/interpersonal environments that

increase and decrease the likelihood of future violence (see also Hodgins, 1997; Monahan, 1981; Mulvey

& Lidz, 1995). According to Monahan and Appelbaum (2000), there are two necessary characteristics

that a risk factor must possess. First, there must be a statistical correlation between the risk factor and

the criterion outcome (recidivism). Second, the risk factor must be present before the criterion outcome

occurs. In the context of the present discussion, risk factors are quantiable features of sex oenders and

their situations that predict future sexual oences.

Hanson (1998) identied two general types of risk factor: static predictors and dynamic predictors.

Static predictors are variables statistically related to recidivism that are either historical in nature (and

thus cannot themselves be changed with interventions) or else are features of the oender that are very

Howard Barbaree, Calvin M. Langton, Andres Gopnik-Lewinski, and Craig A. Beach

unlikely to change over time. Examples of static predictors for violent reoence would include a history

of criminal convictions and a history of juvenile delinquency and antisocial behaviour (Quinsey et al.,

1998). Dynamic predictors are also variables that are statistically related to recidivism but the value of

these variables can change over time. More specically, changes in the value of these dynamic variables

are statistically related to changes in the likelihood of violence. Examples of dynamic predictors for

sex oenders include attitudes supportive of sex with children, association with criminal peers, and

substance use (Hanson & Harris, 2000). A further distinction can be drawn between stable and acute

dynamic predictors. According to Hanson (1998), stable dynamic factors have the potential to change

but the changes are slow and particular values of these variables can endure for long periods (e.g., devi-

ant sexual preferences or alcoholism). In contrast, acute dynamic factors are rapidly changing states (e.g.,

anger, sexual arousal, or intoxication) that immediately precede sexual crimes (Hanson, 1998).

Beech and Ward (2004) have argued quite convincingly that the distinction between static and dy-

namic risk factors is an articial one, an argument that seems to have been recently accepted by Han-

son and his colleagues (Mann et al., 2010, p. 194). According to this argument, an oender engaging in

anti-social behaviour may create a record that cannot be changed (e.g., criminal convictions, behaviour

problems in elementary school) resulting in a static risk factor, or they may express antisocial attitudes

and beliefs recorded by the probation ocer as a dynamic risk factor. However, the underlying antisocial

predisposition is common to both factors. According to this argument, the distinction between static

and dynamic risk factors depends more on the nature of the recording or method of assessment rather

than any substantive dierences among risk factors.

A distinction that is more important to make, relating to the dierences between static and dynamic

risk factors, is between long-term and short-term prediction. Dierent research designs have been used

to investigate two fundamentally dierent approaches to risk assessment. Static risk factors have been

shown to predict violence in research designs involving long-term (ve to twenty years) follow-up be-

tween the clinical assessment and recording of the risk factor and the recidivism outcome being pre-

dicted (e.g., Quinsey et al., 1998). Most of the scientic research on risk assessment has focused on the

identication of risk factors using research designs involving long-term follow-up, and empirically, these

investigations have found static factors to be superior to dynamic factors. e evaluation of long-term

risk due to static risk factors tells the clinician “who” among their oenders is at highest is risk for recid-

ivism. Such a risk determination allows clinicians to allocate more intensive preventive resources and

interventions to the individuals most in need. In contrast, dynamic factors have been shown to predict

violence in research designs that involve much shorter follow-up times. e values of the risk factors are

monitored continuously over several weeks or months and changes in these values are detected prior to a

violent incident (Hanson et al., 2007). Dynamic risk factors, particularly acute dynamic risk factors, are

potentially useful indicators of “when” a reoence might occur. For example, a change in an oender’s

mood, perhaps to one in which hostility is markedly increased or substantial interpersonal distress is noted,

may indicate that a violent act is imminent and crisis intervention is warranted.

e scientic literature describing empirical studies of long-term risk for sexual violence is vast and

a full review of this literature is beyond the scope of the present chapter. By contrast, the scientic study

of dynamic risk is a much smaller literature. e majority of this chapter will focus on key developments

regarding the identication and use of static risk factors with a view to assessing risk for future sexual

violence over the longer term. Toward the end of the chapter, we will summarize some of what is known

about dynamic risk assessment.