using data analysis to explain the state’s criminal
TRANSCRIPT
Using data analysis to explain the State’s criminal
justice system
Presented
Criminal Justice Cross-Training Conference
August 2010
CJPAC – Research Workgroup
• A network of staff from
State criminal justice
agencies
• Monthly meetings to
discuss on-going
research and
evaluation projects
• A venue to assist inter-
agency efforts to share
information, data and
expertise.
• The workgroup
bridges the executive
and judicial branches
OPM/CJPPD Products
• Annual reports:
• Recidivism
• Prison Population Forecasting
• Monthly Indicator Reports
• On-going research and analysis on the CJ system
CJPAC – Monthly Indicators Report
• This four-page monthly
tracks and analyzes
operational flows through
the entire criminal justice
system.
• Data is provided by state
police, the courts, adult
probation, corrections and
parole.
• The report is read by
administrators, legislators,
policy makers,
researchers the media.
Monthly Indicators – inputs & outputs
• Tracks changes in arrests,
incarcerations, VOP rates,
DOC admits, releases and
discharges, re-entry
caseloads, etc.
• The data is used for
• Planning
• Forecasting
• Benchmarking, and
• Training & education
• OPM provides research to
CJ agencies.
The system: inputs & outputs
POPTODAY = POPYESTERDAY + {ADMITS – (RELEASES+DISCHARGES)}
The system: inputs & outputs
POPTODAY = POPYESTERDAY + {ADMITS – (RELEASES+DISCHARGES)}
The murders in Cheshire in July 2007 had a dramatic
impact on the operation and administration of the
criminal justice system in Connecticut.
17,500
18,000
18,500
19,000
19,500
20,000
JAN
'06
FEB
MA
RA
PR
MA
YJU
NJU
LA
UG
SEP
OC
TN
OV
DEC
JAN
'07
FEB
MA
RA
PR
MA
YJU
NJU
LA
UG
SEP
OC
TN
OV
DEC
JAN
'08
FEB
MA
RA
PR
MA
YJU
NJU
LA
UG
SEP
OC
TN
OV
DEC
JAN
'09
FEB
MA
RA
PR
MA
YJU
NJU
LA
UG
SEP
OC
TN
OV
DEC
JAN
'10
Connecticut's prison population, Jan 2006 - Jan 2010
2006 2007 2008 2009
The impact of the crimes in Cheshire
• September 2007, Governor Rell suspends paroles for all violent
offenders. By December 2007, the Board of Pardons and
Paroles reported that releases for 300 non-violent offenders
had been delayed.
• PA-008-01 - An Act Concerning Criminal Justice Reform and
Public Act 08-51 An Act Concerning Persistent Dangerous
Felony Offenders and Providing Additional Resources to the
Criminal Justice System.
• Restructured Board of Pardons and Paroles
• Changes to staff and case review processes
• Video-conferencing
• Expanded electronic monitoring of offenders
• Mandated increased use of risk-assessment methods
• Improved record-sharing
• Expanded victim services
• Established more re-entry services
• New criminal statutes – home invasion and burglary
July 2007 – Dec. 2009 – The post-Cheshire period
Changes to the parole process
0
50
100
150
200
250
300
350
400
450
JA
N …
FE
B …
MA
R …
AP
R …
MA
Y …
JU
N …
JU
L …
AU
G …
SE
P …
OC
T …
NO
V …
DE
C …
JA
N …
FE
B …
MA
R …
AP
R …
MA
Y …
JU
N …
JU
L …
AU
G …
SE
P …
OC
T …
NO
V …
DE
C …
JA
N …
FE
B …
MA
R …
AP
R …
MA
Y …
JU
N …
JU
L …
AU
G …
SE
P …
OC
T …
NO
V …
DE
C …
JA
N …
FE
B …
MA
R …
AP
R …
MA
Y …
JU
N …
JU
L …
AU
G …
SE
P …
OC
T …
NO
V …
DE
C …
JA
N …
FE
B …
MA
R …
AP
R …
MA
Y …
Admin Reviews Hearings
• July 2007 - The Cheshire murders
• September 2007 - Governor Rell suspends paroles for
all violent offenders.
• December 2007 - the Board of Pardons and Paroles
reports that releases for 300 non-violent offenders had
been delayed.
The creation of a population backlog
18875
19714
4869
4078
3500
4000
4500
5000
5500
6000
6500
7000
16500
17000
17500
18000
18500
19000
19500
20000
FE
B 2
006
AP
R 2
006
JU
N 2
00
6
AU
G 2
006
OC
T 2
006
DE
C 2
006
FE
B 2
007
AP
R 2
007
JU
N 2
00
7
AU
G 2
007
OC
T 2
007
DE
C 2
007
FE
B 2
008
AP
R 2
008
JU
N 2
00
8
AU
G 2
008
OC
T 2
008
DE
C 2
008
FE
B 2
009
AP
R 2
009
JU
NE
2009
AU
G 2
009
OC
T 2
009
DE
C 2
009
Offe
nd
ers
in c
om
mu
nity
pro
gra
ms
Inm
ate
po
pu
lati
on
Prison Population
Community Supervision
Incarceration and community supervision in CT
The return to normal community supervision levels.
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
JA
N '0
8
FE
B
MA
R
AP
R
MA
Y
JU
NE
JU
LY
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '0
9
FE
B
MA
R
AP
R
MA
Y
JU
NE
JU
LY
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '1
0
FE
B
MA
R
AP
R
MA
Y
Special Parole
TS
Parole Parcom
Parole
Forecasting or fortune-telling
17,500
17,700
17,900
18,100
18,300
18,500
18,700
18,900
19,100
19,300JA
N '08
FE
B
MA
R
AP
R
MA
Y
JU
N
JU
L
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '10
FE
B
MA
R
AP
R
MA
Y
JU
N
JU
L
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '11
FE
B
Actual
Feb '09 Projection
Sept '09 Projection
Feb '10 Forecast
• OPM had produced three prison-population forecasts since February 2009
• February 2009 forecast – to pessimistic
• September 2009 forecast – to pessimistic
• February 2010 forecast – to optimistic
• Our forecasting is based:
• on historical data
• system knowledge, and
• operational expertise
Forecasting or fortune-telling
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
20,000
22,000JA
N '08
FE
B
MA
R
AP
R
MA
Y
JU
N
JU
L
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '10
FE
B
MA
R
AP
R
MA
Y
JU
N
JU
L
AU
G
SE
P
OC
T
NO
V
DE
C
JA
N '11
FE
B
Actual
Feb '09 Projection
Sept '09 Projection
Feb '10 Forecast
• A matter of scale – from this view it doesn’t look too bad.
• When the prison population was 19,800 the fluctuation of 50 offenders
was significant . At 18,200, a swing of 50 is not so significant.
• Accuracy is important but....
• the ability to explain variations from the projection is equally critical.
Problems with the latest forecast
17,600
17,700
17,800
17,900
18,000
18,100
18,200
18,300
18,400
18,500
18,600
JAN '10
FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN '11
FEB
Actual
Projected
196
289
320
• Unsentenced prison admits:
• are responsible for most of the populations growth in summer, and
• it is hard to model/predict.
• Since May – the number of sentenced inmates had fallen but the number of
unsentenced inmates has surged.
Annual unsentenced admits w/seasonal variations
3800
3900
4000
4100
4200
4300
4400
4500
4600
4700
4800
JAN
1
JAN
9
JAN
17
JAN
25
FEB
2
FEB
10
FEB
18
FEB
26
MA
R 6
MA
R 1
4
MA
R 2
2
MA
R 3
0
AP
R 7
AP
R 1
5
AP
R 2
3
MA
Y 1
MA
Y 9
MA
Y 1
7
MA
Y 2
5
JUN
2
JUN
10
JUN
18
JUN
26
JUL
4
JUL
12
JUL
20
JUL
28
AU
G 5
AU
G 1
3
AU
G 2
1
AU
G 2
9
SEP
6
SEP
14
SEP
22
SEP
30
OC
T 8
OC
T 1
6
OC
T 2
4
NO
V 1
NO
V 9
NO
V 1
7
NO
V 2
5
DEC
3
DEC
11
DEC
19
DEC
27
Unsentenced DOC admits
2005 2006 2007 2008 2009
5 years of unsentenced admit data and the model years
(2005, 2009 and early-2007)
Unsentenced admits with 2010 overlay
3800
3900
4000
4100
4200
4300
4400
4500
4600
4700
4800
JAN
1
JAN
9
JAN
17
JAN
25
FEB
2
FEB
10
FEB
18
FEB
26
MA
R 6
MA
R 1
4
MA
R 2
2
MA
R 3
0
AP
R 7
AP
R 1
5
AP
R 2
3
MA
Y 1
MA
Y 9
MA
Y 1
7
MA
Y 2
5
JUN
2
JUN
10
JUN
18
JUN
26
JUL
4
JUL
12
JUL
20
JUL
28
AU
G 5
AU
G 1
3
AU
G 2
1
AU
G 2
9
SEP
6
SEP
14
SEP
22
SEP
30
OC
T 8
OC
T 1
6
OC
T 2
4
NO
V 1
NO
V 9
NO
V 1
7
NO
V 2
5
DEC
3
DEC
11
DEC
19
DEC
27
Unsentenced DOC admits
2005 2006 2007 2008 2009 2010
OPM anticipates the number of unsentenced offenders
will peak in early September.
Other research
• Parolees – recidivism and post-release pathways
• Halfway house study – bed turn-over rates
• Risk scoring – TPAI and domestic violence
• Recidivism among weapons violators
• Who is a non-violent offender
• The economy for re-entering prisoners
Where do parolees go? (2005) The first movement
New Crimes
and charges
Absconds and
Technical
violations
EOS - discharge
sentence Other Total
462 640 1370 50 2522
18% 25% 54% 2% 100%
Where do parolees go? (2005) Technical violators
movemen
Where do parolees go? (2005) Technical violators
movemen
Where do parolees go? (2005) Reparolees
Where do parolees go? (2005) Reparolees
Halfway house churn rates
13%8% 11% 11% 13% 12% 11%
6% 5% 6% 3% 1%0%
20%
1 2 3 4 5 6 7 8 9 10 11 12
Cohort: May 2008
6%
17%10%
16% 14% 12% 10%6% 3% 2% 1% 0%
0%
20%
1 2 3 4 5 6 7 8 9 10 11 12
Cohort: November 2008
11%
21%12%
19% 17%8% 5% 5% 2% 1% 1% 0%
0%
20%
1 2 3 4 5 6 7 8 9 10 11 12
Cohort: May 2009
11%
25% 25% 26%
12%2% 0% 0% 0% 0% 0% 0%
0%
20%
1 2 3 4 5 6 7 8 9 10 11 12
Cohort: November 2009
Percent of cohort over 5-months in first halfway house placement
May 2008 Nov. 2008 May 2009 Nov. 2009
45% 34% 22% 2%
TPAI-risk scores
TPAI is an scale 8-point scale that is used to assess offender risk with
respect to recidivism. There are other more complex risk instruments but
this is relatively simple and has been in use since 2009.
Recidivism among weapons offenders
• How do we define a non-violent offender?
• By statute (Assault 3?)
• By arrest, conviction or sentence
• The effect of plea negotiation
• Weapons possession?
Sentence mix for 1,191 offenders who served a prison sentence for illegal
weapons use or possession.
Felony violence: 41%
Drugs: 65%
Property: 44%
Recidivism among weapons offenders
Gun-charge
arrest
Gun-charge
conviction
Gun-charge
sentence
Male offenders 4,785 2,286 1,862
Any return to prison
12 months 41% 41% 42%
24 months 57% 58% 60%
36 months 65% 66% 67%
Table 4. Recidivsm among males with weapons
• How do we define a non-violent offender?
• By statute (Assault 2?)
• By arrest, conviction or sentence
• The effect of plea negotiation
• Weapons possession?
The economy for re-entering offenders
From 1998 to 2005, Connecticut
alone has lost 52,000
manufacturing jobs....Meanwhile,
jobs in Connecticut's growing
industries aren't as well
compensated as the jobs in
Connecticut's shrinking
industries. Average wages in the
state's growing industries are 32.3
percent lower -- $18,021 per year
less-- than those in Connecticut's
shrinking industries.
JOHN W. OLSEN, Connecticut AFL-
CIO , The Hartford Courant, September 4, 2006.
-
50,000
100,000
150,000
200,000
250,000
2002 2004 2006 2008 2010
Manufacturing
Construction
Retail
Food Service
Transportaton
Admin. & Waste
Blue-collar employment in CT
John Forbes
Asst. Director, OPM, CJPPD
Ivan Kuzyk, Director
CT Statistical Analysis Center
www.ct.gov/opm/CriminalJustice/Research