PSYCH 3CC3 Lecture Notes - Lecture 8: Pharmacotherapy, Sex Offender, Voyeurism
6 Jun 2018
School
Department
Course
Professor
Assessment of Violence Risk
Lecture 8
Baxtrom vs. Herold (1966)
• Baxtrom’s rights were violated when he was hospitalized beyond prison term without
dangerousness assessment
• Many prisoners are not released from prison b/c they develop mental disorders in prison
• He then got released from prison mental health hospital to a low security mental health
hospitals
• ~1000 prisoners were moved to low security hospitals as well. The general public
expected them to be crazy & violent in hospital, but this was not entirely the case
o 20% subsequently violent
o 18% discharged within a year into community
o 1% later readmitted back into secure hospital
• 4.5 years later, 50% of patients were released into the community and < 3% returned to
secure hospital
Thorberry & Jacoby (1979)
• 586 patients released from Pennsylvania institution
• 3-year return/recidivism rate = 23.7%
• 4-year violent arrest record = 15%
Quinsey & Ambtman (1979)
• Compared judgement of 4 forensic psychiatrists to 9 high school teachers to see how
good they are at determining violent reoffending risk
• They evaluated data of 30 patients including property offenders, child molesters, &
murderers
• Given 3 types of patient data: offence description, patient history, psychiatric assessments
(IQ, MMPI, Rorschach, etc.)
• Conclusions:
o Inter-rater reliabilities low for both groups – there wasn’t a lot of agreement
among the psychiatrists, nor was there agreement between high school teachers
o Inter-rater reliability similar for both groups – both groups overestimated the
offender’s tendency to re-offend, but they didn’t differ in their accuracy. Teachers
are just as likely to be right in terms of what actually happened
o Psychiatrists rate offenders more likely to commit offense – compared to teachers,
psychiatrists were stricter and found offenders more likely to re-offend
o Little use of psychiatric assessments made by either group – both groups used the
offence descriptions and patient history but ignored the psychiatric assessments
▪ Recommended that psychiatric assessments should be used to determine
treatability rather than dangerousness
Monahan (1981) and Barefoot v. Estelle (1983)
• Found that psychiatrists and psychologists are only right 1/3 predictions of violent
behaviour; they are wrong almost 70% of the time
find more resources at oneclass.com
find more resources at oneclass.com
• Barefoot – “neither petitioner nor the Association suggests that psychiatrists are always
wrong w/ respect to future dangerousness, only most of the time”
Violence Risk Assessment Models
• It is known that psychiatric assessments are often wrong but are still used to predict
violence risk. Below are ways to increase accuracy
• Actuarial methods – evidence is based on statistical data (e.g. characteristics of offender,
his crime behaviours, etc.) and whether those things are predictive of future violence
behaviours. Ignores the inaccurate clinical psychiatric assessment
• Structured professional judgement – doesn’t give up on professional assessments but
makes it more accurate by making it structured. The assessment process is structured so
clinicians are all doing the same thing and it can be tested & changed until it’s accurate
Measures of Accuracy (Actuarial Methods)
• Follows the same hit, miss, false alarm, and correct rejection as other lectures. This is
saying someone will violently re-offend or not, and whether they do/don’t
• This is an actuarial approach as it tests offender variables using statistics to determine
how likely they’ll reoffend
o People are grouped by common characteristics and it is used to predict violent
reoffending but each individual is different. Actuarial method groups people
together
• Positive predictive power (PPP) – H / (H+FA) the amount of people you say will be
dangerous, are actually be dangerous
o Problem: strongly influenced by the base rate of the condition you are testing for.
The less frequent the condition is in the population → worse PPP
• Negative predictive power (NPP) – R / (R+M) the amount of people you say won’t be
dangerous, are actually not dangerous
• Sensitivity – S = H / (H+M)
• Specificity – Sp = R / (R+FA)
• Overall accuracy = R + H / (R + H + FA + M). All correct judgements over all
judgements
• Base Rate – the prevalence of a disease/condition in the population
o Ex. If we assume the breast cancer screening test is 95% accurate, prevalence of
breast cancer in the population is 50%, working w/ 10,000 people
▪ 50% prevalence = 5000 people will actually have the disorder while 5000
people won’t
▪ Hits & correct rejections = 95% → 4750 people each
▪ False alarms & misses = 5% → 250 people each
▪ PPP = 4750 / (4750+250) = 95%
o Ex. If the test accuracy remains the same (95%) but prevalence goes down to 5%
▪ 5% prevalence = only 500 people have the disorder while 9500 people
don’t
▪ Hits = 500 x 95% = 475
▪ Misses = 25
▪ False alarms = 475
▪ Correct rejections = 9025
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Thorberry & jacoby (1979: 586 patients released from pennsylvania institution, 3-year return/recidivism rate = 23. 7, 4-year violent arrest record = 15% It is known that psychiatric assessments are often wrong but are still used to predict violence risk. Below are ways to increase accuracy: actuarial methods evidence is based on statistical data (e. g. characteristics of offender, his crime behaviours, etc. ) and whether those things are predictive of future violence behaviours. Ignores the inaccurate clinical psychiatric assessment: structured professional judgement doesn"t give up on professional assessments but makes it more accurate by making it structured. The assessment process is structured so clinicians are all doing the same thing and it can be tested & changed until it"s accurate. Measures of accuracy (actuarial methods: follows the same hit, miss, false alarm, and correct rejection as other lectures. All correct judgements over all judgements: base rate the prevalence of a disease/condition in the population, ex.
