Risk Assessment Flashcards
What statistic is used to evaluate how well something predicts violence?
Recieve Operating Characteristic
How is the result expressed? What classes a result as weak, moderate or strong?
AUC (Area Under the Curve)
.56 = Weak
.65 = Moderate
.71 = Strong
.9 = Best you can possibly do
What is a pro and con of ROC?
Pro - immune to baseline changes
Con - Loses the quality of event eg how long to reoffend
What are the 4 risk assessment techniques
Unstructured Clinical Interview
Actuarial measures
Structured Professional Judgement
Formulation based approach
What 3 signs of future violence were used by professionals for years before others methods came into place?
Arson, bed wetting and cruelty to animals
What happened after Johnnie Baxtrom’s appeal to be detained in a hospital for the criminally insane got upheld?
966 other ‘dangerous’ patients were released and only 20 of them were later arrested for a violent crime
What was found in Odeh et al’s study when they gave professionals information about a patient and make a risk judgement?
The cues they used from the information were unrelated to the violence of the patient
Interrater reliability was poor
What are the 4 main reasons why people are bad at making risk judgements?
Too many variables
Blind to their outcomes
Tendency to favour strange or irrelevant factors heavily
Tend to make judgements quickly and then seek support for these
What is an actuarial assessment?
Preordained way of coming to a decision e.g computer formula
What are the pros and cons to actuarial assessments
Pros - Fast, Avoids individual bias, don’t need clinical skills to formulate
Cons - Often lacks ideographic info, does not easily suggest risk management
Name as many of the 12 items on the VRAG as possible
Schizophrenia, lived with both parents before age of 16, Elementary school maladjustment, history of alcohol problem, Marital status, failure on conditional release, victim injury, sex of victim, personality disorder, psychopathy checklist score
What to the scores range from on the VRAG?
-26 to +38
What did Snowden et al (2007) find when using the VRAG on a sample of males in a medium secure prison?
Has good validity as predicting violent crime
The absolute rates of violence in the UK sample are lower than the Canadian sample
Why might the absolute rates of violence in the UK sample be lower than the Canadian sample
More supervision/management
use of formal reconviction may miss many violent crimes
not able to correct for any further time spent incarcerated
less violent than Canadians
less competent detection/prosecution of incidents
What arguments did Litwack make to say making comparisons of actuarials to clinicians is unfair
Clinicians tend to concentrate on the short term outcome
Have different information available
Clinicians now do better due to knowing the variables that predict violence
Focus on the really dangerous not some single act that may occur
Some clinicians are bad - letting the side down
What were the results when clinicians were compared to the VRAG
AUC for clinicians = .59
AUC for VRAG = .80
What other examples of actuarials are there?
Offender Group Reconviction Scale (OGRS)
Risk Matrix 2000 (RM2000)
Classification of Violence Risk (COVR)
Why don’t clinicians use actuarials?
‘de-individuates’ the person
clinicians feel status is being eroded
Clinicians feel risk assessment isn’t their job
What are some actual limitations of actuarials
person has to fit with the sample
tend to use static measures - does it change if dangerousness changes
tend to be based on most common violence
A number isn’t enough - need more info
What are the 3 categories in the HCR20?
Historical, Clinical, Risk management
What did impulsivity get changed to in the HCR20
trigger reactions - how quickly they get angry
What does the HCR20 do that the VRAG doesn’t
Tells you where the risk is coming from
What were the findings for the HCR20 and women?
some predictive validity in women
DeVogel and De Ruiter (2006) - Men’s AUCs were higher (.75-.88) than women (.52-63)
How did the HCR20 fair with those with disabilities?
Better than other offenders
How did the HCR20 do with ethnic minorities?
Similar properties in UK ‘black’ population as the ‘white’ sample
How did the HCR20 so with mental illness diagnosis?
Showed those with personality disorders are the most predictive of violence and that those with diagnosis in general are a good predictor.
Good predictor of those with schizophrenia but not with PD
What are the limitations of the HCR20
Need more evidence of utility in other populations and specific forms of violence
Need to see if treatments causes changes in the HCR20 and if it is indicative of less dangerousness
Homocide is rare so few studies of its prediction
