Lecture 1
Definition
Probability calculation that a harmful behavior or event will occur, and involves an
assessment about the frequency of the behavior (or event), its likely impact and who it will
affect
The attempt to predict the likelihood of future offending in order to identify individuals in
need of intervention
Risk assessment is also
o Risk management: interventions to manage the risk or to reduce risk
Why?
History
1876 – Lombroso
o Criminals have certain characteristics
Large protruding jaws
Deeper eyes
Low forehead
Large chin
High cheekbones
Hawklike nose
1928 – Burgess: first risk assessment tool
o Developed tools to determine the risk of re-offense of offenders being released from
the Illinois prison system
o Actuarial in nature
o Marital status, criminal and employment history and institutional misconduct
o Based on analyses of data for 3000 individuals paroled in Chicago
o 76% high risk status recidivated within 5 years
1950 – no problem
1960 – Baxstrom
o 967 dangerous psychiatric patients transferred to regular psychiatric hospitals
o 121 patients subsequently released into society
o Followed-up Steadman & Cocozza (1978)
o After 4 years of follow-up
2,7% sent back to forensic psychiatric hospital
Patients in society
17% arrested
7% convicted
1970 – doubt
, o The state of the art regarding predictions of violence is very unsatisfactory. The ability
of psychiatrists or any other professionals to reliably predict future violence is
unproved
o Psychiatry and the presumption of expertise: flipping coins in the courtroom
Expertise of the expert
2 groups of subjects
o Teachers
o Experienced forensic psychiatrists
Questions regarding recidivism based on file information
Results
o IRR teachers low
o But better than IRR psychiatrists
o Averagely large similarity between psychiatrists and teachers
o The more information available, the more conservative in
judgment
1980 – Monahan
o Psychiatrists and psychologists are accurate in no more than 1 out of 3 predictions of
violent behavior
1990 – research RNR model
o Martien Philipse
o Dutch TBS system enjoyed great respect in the world at that time
Praised for their innovative approach to forensic patients
Research: how the Dutch behavioral experts from the forensic field made
their risk assessments in practice
None of the predictors suggested by the clinicians were predictive
There were also no link at all between the clinical assessment of relapse risk
and actual recidivism
2000 – integration in to practice
Risk, need and responsivity model – Andrews & Bonta
Evidence based practice: interventions should be based on results from scientific research
into the effectiveness of those interventions
Risk principle = offenders with a higher risk of recidivism benefit most from more intensive
treatment
o Who to treat?
Match the treatment intensity with the client’s risk of recidivism
o The risk of recidivism is central; not the disorder
o Misalignment of treatment intensity may increase the risk of recidivism
o When you apply the same treatment for high and low risk groups, the treatment
effect is greater in high risk groups
Need principle = only those factors associated with reduction of recidivism should be
addressed during treatment
o What to treat?
Focus on the treatment on the criminogenic needs that are actually related
to criminal behavior and therefore require attention
Each individual has his own combination of factors that led to the crime
These criminogenic needs are the dynamic risk factors related to the crime
Dynamic risk factors are modifiable by treatment interventions, so focus of
treatment is on these high dynamic risks
o Why?
, Adequate treatment of criminogenic needs is associated with a 19%
reduction in recidivism; targeting an intervention to non-criminogenic needs
is associated with a slight increase of 1%
Responsivity principle = interventions should be tailored to the offender’s characteristics,
such as motivational level, personal circumstances and learning style
o How to treat?
Match the treatment and the type of interventions to the client’s learning
style and learning abilities
o External responsivity = type of treatment, characteristics of the therapist, availability
of adequate interventions, possibilities offered by the treatment environment
(stability, quality of life, mobility)
o Internal responsivity = take into account the individual characteristics of a client:
intellect, personality, disorder, treatment motivation, learning style
o Responsive treatment is associated with a 23% reduction in recidivism
Criminogenic needs – central 8
Divided by big four and moderate four
History of antisocial behavior
o Risk: early and ongoing involvement in a number and variety of antisocial acts in a
variety of settings
o Dynamic need: work on non-criminal alternative behaviors in high-risk situations
Antisocial personality pattern
o Risk: thrill seeking, low self-control, restless, aggressive
o Dynamic need: work on problem-solving skills, self-management skills, anger
management and coping skills
Antisocial cognitions
o Risk: attitudes, values, beliefs and rationalization that support criminal behavior;
cognitive-emotional conditions of anger, resentment and pride; criminal vs anti-
criminal identity
o Dynamic need: reduce antisocial cognitions, recognize risky ways of thinking and
feeling, develop alternative and less risky ways of thinking and feeling, adopt a
renewed and/or anti-criminal identity
Antisocial peers
o Risk: close contacts with criminal others and relative social isolation with non-
criminal others; direct social support for crime
o Dynamic need: reduce contacts with criminal others, build contacts/links with non-
criminal others
Family/marital relationships
o Risk: two important elements are: education and/or care, and control and/or
supervision
o Dynamic need: reduce conflict, build positive relationships, improve control and
oversight
School and/or job
o Risk: low levels of achievement and satisfaction at school and/or work
o Dynamic need: improve performance, rewards and satisfaction
Prosocial recreational activities
o Risk: low levels of engagement and satisfaction in non-criminal leisure activities
o Dynamic need: improve engagement, rewards and satisfaction
Substance abuse
o Risk: abuse of alcohol and/or drugs
, o Dynamic need: reduce drug abuse, reduce personal and interpersonal support for
addictive behavior, develop alternatives to drug abuse
Non-criminogenic needs
Less research done and insufficiently proven
Relevant non-criminogenic needs for general offending behavior are
o Self-esteem
o Emotional problems
o Psychiatric disorder
o Health
These dynamic factors are not directly related to recidivism reduction but can indirectly
contribute to reducing dynamic risk
Other (non-?)criminogenic needs that according to Serin et al. (2013) appear to be directly
related to recidivism reduction
o Decrease in negative emotions
o Successful treatment process from the client himself
Lecture 2
Predictive validity
Sensitivity = proportion of recidivists
previously assessed as high risk
o Retrospective measure
Specificity = proportion of non-recidivists
previously assessed as low risk
o Retrospective measure
Area under the curve (AUC)
o Receiver operating
characteristics (ROC)-analysis
o Used to classify future recidivists
vs non recidivists
o AUC = 1,00 = perfect
discrimination = perfect prediction
o Interpretation (Rice & Harris)
.56 - .64 = small
.64 – 71 = moderate
.71 > = strong
o 2 types of information needed
Risk scores
A dichotomous outcome
o Most commonly reported measure in the literature
o Retrospective measure
o Chance that a random recidivist gets a higher score than a random non-recidivist
o An AUC value is therefore the trade-off between
Sensitivity (true positive rate
1 – specificity
