The LSI-R and its Predictive Validity for Female Inmates Presented at the Undergraduate Research Opportunities Program Symposium by Maro Siranosian

1
The LSI-R and its Predictive Validity for Female
InmatesPresented at the Undergraduate Research
Opportunities Program Symposium by Maro
Siranosian
May 13, 2006
2
Some Definitions

• Recidivism any criminal act (parole/probation
  violation or new crime) that results in re-arrest
  and/or reconviction of a person previously
  released from prison
• Recidivism risk the likelihood that a person
  will recidivate, given the presence or absence of
  certain risk variables
• Risk assessment the attempt to predict risk of
  recidivism typically by using standardized tools
  that consist of variables related to recidivism.

3
Violence risk assessment and women
why should we care?

• Post-release decision making
• Female prisoner population is small, but growing
  rapidly
• Female inmates are a distinct population with
  different needs

4
Source BJS, 1999
5
Past Research

• Clinicians underestimate womens violence, (Skeem
  et al., 2005)
• The LSI-R is predictively valid with male
  samples, (Loza Simourd, 1994 Loza, 2003
  Simourd Hoge, 2000 Kroner Mills, 2001)
• The LSI-R is predictively valid with female
  samples (Rettinger, 1998 Coulson, Ilacqua,
  Nutbrown, Guilekas, Cudjoe, 1996)

6
LSI-R
LSI-R and gender differences
Current Study
Gender and risk assessment
7
Level of Service Inventory-Revised (Andrews
Bonta, 1995)

• 54 items
• Grouped into ten risk/needs factors
• Education/employment
• Financial
• Family/marital
• Accommodation
• Leisure/recreation
• Companions
• Alcohol/drug problems
• Emotional/personal
• Attitudes/orientation
• Criminal history

8
Level of Service Inventory-Revised (Andrews
Bonta, 1995)

• Mix of dynamic and static factors
• Criminal History - static
• Education/Employment - 50/50
• All others dynamic

Dynamic factors that change over an offenders
lifetime, such as employment status or
accommodations Static unchangeable factors, such
as age at release or prior misdemeanors
9
Central Questions

• Does the LSI-R predict general recidivism for
  female inmates?
• Do the factors that predict risk of general
  recidivism differ by gender?

10
Data

• Females (n70)
• Mean LSI-R 23.79
• Mean years in prison 7.76
• Mean age at release 37.04

• Males (n1144)
• Mean LSI-R 26.27
• Mean years in prison 8.94
• Mean age at release 39.35

11
Data
12
Minimum 0-12 Medium 13-23 High/medium
24-36 Maximum 37-54
13
Methodology

• LSI-R
• Total score
• Four risk categories (minimum, medium,
• high/medium, maximum)
• Risk trichotomy (low, med, high)
• Covariates
• Months in prison
• Age at release
• Total misdemeanors and felonies
• Survival Analyses and Stepwise regression

14
Results
15

• Question 1
• Does the LSI-R predict general
• recidivism for female inmates?

16

• LSI-R Total Score
• Trichotomy of Risk
• 4 Risk Categories

• ?2 (1, N70)
  
• 7.31, plt.05

• ?2 (2,
  N70)
• 6.35, plt.05

?2 (3, N70)
6.08, not significant
17
Months to recidivate
18
Question 2 Do the factors that predict risk
of general recidivism differ by gender?
19

• Yes, the factors which most contribute to the
  prediction of recidivism risk differ by gender
• Males criminal history, financial, and
  alcohol/drug problems
• Females criminal history and financial

20
Conclusions

• LSI-R is significantly predictive of risk of
  recidivism for female inmates
• The factors which are most predictive of general
  recidivism risk for males and females differ
  somewhat substance abuse was found to be more
  important for males

21
Implications

• Predictively valid risk assessment tool
  originally tailored to males will predict for
  females
• Utility of a risk assessment tool tailored to
  females
• Supervision of men v. women on parole

22
Future Research

• Larger sample of female inmates
• Longer follow-up period
• Qualitative analysis of why the factors that
  predict differ between males and females

23
Acknowledgements

• Professor Jennifer Skeem, Ph. D.
• Professor Valerie Jenness, Ph. D.
• Sarah Manchak, MA
• UROP
• Robert Barnoski and the Washington State
  Institute for Public Policy

24
Contact information

• Maro Siranosian
• School of Social Ecology
• University of California,
• Irvine
• msiranos_at_uci.edu