Issues in the Measurement and Analysis of Redemption

Scarlet Letters and Recidivism: Does an Old Criminal
Record Predict Future Criminal Behavior?
Megan Kurlychek
University of South Carolina
Robert Brame
University of South Carolina
Shawn Bushway
University of Maryland
Obstacles for Offenders
• Individuals with a criminal record encounter
difficulties when they return to the community.
–
–
–
–
–
–
Voting disenfranchisement.
Restricted access to public housing.
Reduced/terminated parenting and adoption rights.
Ineligibiliy for student loans and grants
Restricted volunteer and civic activities
Obstacles to Employment
• Legal bars (armed robber can’t work in a gun shop).
• Discretionary hiring rejections
Collateral Consequences of Criminal Records Have
Become Increasingly Severe in Recent Years
• Almost all states restrict voting rights for individuals with a felony
conviction (Uggen and Manza 2002). The percentage of states
restricting access to the polls is currently at an all-time high.
• Students convicted of drug-related offenses are barred from
receiving federal financial aid (Higher Education Act of 1998).
• Drug-related felons experience restricted access to public
assistance and food stamps (1996 Federal Welfare Law).
• Federal Adoption and Safe Families Act of 1997 provides for
restrictions on adoption and foster parenting for people with
certain types of convictions.
• In 1996, an estimated 56% of large employers conducted
criminal background checks; By 2004, the percentage of large
companies conducting criminal background checks rose to 80%.
Reasons for Increased Reliance on
Criminal Background Screening
• Protection of property and preservation of a
safe and orderly environment.
• Reduced risk of liability.
• Competitive market for employment or access
to services - obvious reason to reduce size of
applicant pool.
• Because they can: improved technology,
reduced cost.
Risk of New Offenses By Number of Prior Offenses
(1958 Philadelphia Birth Cohort Males, N=13,160)
Figure 1: Risk of New Offenses By Number of Prior Offenses
(1958 Philadelphia Birth Cohort Males, N = 13,160)
Probability of at Least One More
Contact/Arrest
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
1
2
3
4
5
6
Number of Prior Contacts/Arrests
7
8
Constraining Influences
• Fundamental fairness: sentence has been imposed
by criminal justice system. Debt to society has been
paid when sentence is served
• Behavioral outcomes: redemption barriers may
increase likelihood of recidivism and paradoxically
decrease public safety.
• Laws - Wisconsin statute described by Pager
(2003:946): “Employers are cautioned that crimes
may only be considered if they closely relate to the
specific duties required of the job, however ‘shocking’
the crime may have been.”
Unresolved Policy Issues
• What is a criminal record?
• How accurate are databases used for criminal record
screening?
• Should evidence of rehabilitation be considered along
with a criminal record?
• What level of access to juvenile records should be
allowed?
• Consideration of ancillary criminal record information
such as the number of arrests/convictions, the types
of offenses in the record, or the amount of time that
has passed since the last offense in the record.
Five Year Arrest Recidivism Hazard Rate Among Offenders
Arrested for the First Time at Ages 18-20 (N = 805)
Figure 2: 5-Year Arrest Recidivism Hazard Rate Among Offenders Arrested
for the First Time at Ages 18-20 (N =805)
0.045
0.035
0.030
0.025
0.020
0.015
0.010
0.005
Number of Months Since First Arrest
58
55
52
49
46
43
40
37
34
31
28
25
22
19
16
13
7
10
4
0.000
1
New Arrest Hazard Rate
0.040
Research Questions
• Among individuals who have offended in the
past, does the probability of committing new
offenses decline as the time since the last
offense increases?
• If so, does it decline to the point that it
becomes indistinguishable from those who
have no record of past offending?
• Simulates the situation of an employer who
obtains an arrest record of a job candidate.
An Example
• Consider a group of individuals who have a
prior record but the last entry in that record
occurred at age 18 (Group 1).
• These individuals are now age 35. What
proportion of them will accumulate a new entry
this year?
• Now consider another group of individuals who
have no prior record (Group 2).
• These individuals are now age 35. What
proportion of them will accumulate a new entry
this year?
Our objective is to understand whether:
(1) the risk of new criminal record entries
(among those with prior records) drops with
increasing time since the last entry; and, if it
does,
(2) does that risk approach the level found
among those with no record?
• Group 1% > Group 2% - “Persistence”
• Group 1% < Group 2% - “Redemption”
Two Studies
• Study #1 - Data from 13,160 males from
the 1958 Philadelphia birth cohort study
(Tracy et al., 1985) followed through
age 27.
• Study #2 - Data from 670 males from
the 1942 Racine birth cohort study
(Shannon, 1982) followed through age
32.
Study #1 - Data Overview
Sample of N = 13,160 males from the 1958 Philadelphia birth
cohort study.
– Born in Philadelphia in 1958.
– Resided in city between ages 10 and 17.
– Includes history of police contacts for criminal offenses
through age 17.
– Includes history of arrests for criminal offenses between from
age 18 to age 26.
– No requirement that individuals reside in city as adults; some
individuals may have moved away and some may have been
incarcerated for at least part of the period.
Descriptive Information
• Total N = 13,160
• The majority of these individuals had no record of any criminal
activity through age 24 (N = 8,043; 61.1%).
• A smaller but significant group of individuals had experienced
police contact for criminal activities as juveniles (through age
17) but no adult arrests through age 24 (N = 2,197; 16.7%).
•
•
•
•
•
•
•
Last Arrest at Age 18 = 432; 3.3%
Last Arrest at Age 19 = 341; 2.6%
Last Arrest at Age 20 = 292; 2.2%
Last Arrest at Age 21 = 361; 2.7%
Last Arrest at Age 22 = 403; 3.1%
Last Arrest at Age 23 = 497; 3.8%
Last Arrest at Age 24 = 594; 4.5%
Probability of Failure at Age 25-26
Probability Fail at Age 25-26
12,000
0.4000
8,000
0.2500
6,000
0.2000
0.1500
4,000
0.1000
2,000
0.0500
Age at Last Record Entry
24
23
22
21
20
19
<
18
<
&
R
18
d
or
N
Re
c
o
O
nl
y
0.0000
18
0
N
Number of Cases
0.3000
Probability Distribution
0.3500
10,000
Another Perspective
• Let’s assume that juvenile offending
records are unavailable and now
consider two groups of individuals:
– those who offend at least once at age 18.
– those who refrain from offending at age 18.
Hazard Rate
• Now, we follow both groups of individuals
from age 19 through age 26.
• At each age, we ask the following question:
Of those who have “survived” this far without
a new arrest, what proportion acquire a new
arrest at this age?
• The answer to this question is given by the
estimated hazard rate for that age.
• We are interested in whether the hazard rate
declines over time for the age 18 offenders
and whether it approaches the hazard rate for
the age 18 nonoffenders.
Hazard Rate Analysis Using Philadelphia Data
•
•
•
•
N = 13,160 Males
Age 18 Offenders - N = 1,009 (7.7%)
Age 18 Nonoffeners - N = 12,151 (92.3%)
For each group, at age 19, we calculate the
proportion of individuals who fail.
• For each group, at each age after 19, we calculate
the proportion of individuals who fail among those
who have not failed between age 19 and that age.
• To sum up, we are interested in the following
quantity: Given a population of individuals who have
not failed prior to a given age, what proportion will fail
at that age?
Hazard Rate Trend for Age 18 Nonoffenders (N = 12,151)
Arrest Hazard Rate by Age (Age 18 Nonoffenders)
0.018
0.016
0.012
0.010
0.008
0.006
0.004
0.002
0.000
19
.0
19
.3
19
.7
20
.0
20
.3
20
.7
21
.0
21
.3
21
.7
22
.0
22
.3
22
.7
23
.0
23
.3
23
.7
24
.0
24
.3
24
.7
25
.0
25
.3
25
.7
26
.0
26
.3
26
.7
Hazard Rate
0.014
Age
Hazard Rate Trend for Age 18 Offenders (N = 1,009)
Arrest Hazard Rate by Age (Age 18 Offenders)
0.160
0.140
0.100
0.080
0.060
0.040
0.020
Age
26.7
26.3
26.0
25.7
25.3
25.0
24.7
24.3
24.0
23.7
23.3
23.0
22.7
22.3
22.0
21.7
21.3
21.0
20.7
20.3
20.0
19.7
19.3
0.000
19.0
Hazard Rate
0.120
Hazard Rate Trend for Both Groups
Arrest Hazard Rate by Age
0.180
0.160
0.140
Age 18 Offenders
Hazard Rate
0.120
0.100
0.080
0.060
0.040
Age 18 Nonoffenders
0.020
Age
26.7
26.3
26.0
25.7
25.3
25.0
24.7
24.3
24.0
23.7
23.3
23.0
22.7
22.3
22.0
21.7
21.3
21.0
20.7
20.3
20.0
19.7
19.3
19.0
0.000
Adult Hazard Rates for Age-18 Violent
and Non-Violent Offenders
Figure 5. Arrest Hazard Rate by Age Among Age-18 Offenders (N =
1,009)
0.160
0.140
0.120
Age 18 Violent
Offenders
(N = 375)
Age 18 Offenders
(N = 1,009)
0.080
0.060
0.040
Age 18 Nonviolent
Offenders
(N = 634)
0.020
Age
6.
7
2
6.
3
2
6.
0
2
5.
7
2
5.
3
2
5.
0
2
4.
7
2
4.
3
2
4.
0
2
3.
7
2
3.
3
2
3.
0
2
2.
7
2
2.
3
2
2.
0
2
1.
7
2
1.
3
2
1.
0
2
0.
7
2
0.
3
2
0.
0
2
9.
7
1
9.
3
1
9.
0
0.000
1
Hazard Rate
0.100
Study #2 - Data Overview
Sample of N = 670 males from the 1942 Racine birth
cohort study.
– Born in Racine, Wisconsin in 1942.
– Includes history of police contacts for criminal
offenses through age 32.
– Offense prevalence rates are higher in Racine
than in Philadelphia.
– Smaller sample size creates some obstacles:
• Increases uncertainty about the estimated
onset and recidvisim rates.
• Does not allow us to investigate the types of
offenses (e.g., violent vs. non-violent).
Descriptive Information
• About 30% of these
individuals had no contacts
prior to age 25.
• About 26% had escaped
contact with the police through
age 27.
• About 18% (119) of these
individuals had at least one
juvenile contact but then had
no further contacts between
ages 18 and 24.
• About 15% (100) had at least
one juvenile contact but no
further contacts between ages
18 and 27.
Probability of Failure at Age 25-32
Probability of Failure at Age 28-32
Hazard Rate Trend for Juvenile Offenders
and Nonoffenders
Hazard Rate Trend for Age 18 Offenders
and Nonoffenders
Hazard Rate Trend for Age 19 Offenders
and Nonoffenders
Hazard Rate Trend for Age 20 Offenders
and Nonoffenders
Hazard Rate Trend for Young Adult Offenders
and Nonoffenders
Conclusions
• Abundant evidence that access to criminal
records is easier and less expensive than
ever.
• Many unresolved issues about the
boundaries of appropriate use of these
records and the utility of imposing restrictions
based on information obtained from them.
• One key consideration is the time that has
elapsed since the last arrest or contact.
Conclusions (Continued)
• Our analyses of two birth cohort data
sets suggest that older criminal records
are significantly less predictive of future
behavior than new criminal records.
• The Racine analysis actually indicates
that after about 7 years, a criminal
record provides little insight into the risk
of new police contacts.