Vous êtes sur la page 1sur 68

!X"#$$!%#&!'()*X+&#,*"'X!-#.

/X'0*X
"*.'*&X1#&X*11*"'()* +2%$("X+#$("3X
+&*'&(!$X,24'("*X(.4'('2'*X
,24'("*X-!.!/*-*.'X(.4'('2'*X
!.5X
'0*X"!&*3X/&#2+
Aprll 16, 2010
CenLer for LffecLlve ubllc ollcy
A Framew!rk f!r Evidence-Based
Decisi!n Making in L!cal Criminal
Justice Systems
Aoololtlotlveoofotbe
naLlonal lnsLlLuLe of CorrecLlons
A W!rk in Pr!gress
Third Editi!n
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
2
FOREWORD
As we sLand aL Lhe beglnnlng of a new decade, [usLlce sysLem professlonals are challenged by
Lhe rlslng cosLs of crlmlnal [usLlce, Lhe sLorles of vlcLlms harmed by crlme, and Lhe fallure of Loo
many offenders who pass Lhrough our gaLes and doors. We aL nlC, llke our colleagues across
Lhe counLry, are keenly aware of Lhe new opporLunlLles recenL research offers regardlng clear
and speclflc sLraLegles LhaL wlll reduce crlme, ease rlslng cosLs, and, mosL lmporLanLly, prevenL
fuLure vlcLlms.
ln 2008, we launched Lhe Lvldence-8ased ueclslon Maklng ln Local Crlmlnal !usLlce SysLems
lnlLlaLlve. ln dolng so, we soughL Lo encourage and faclllLaLe advancemenLs ln Lhe crlmlnal
[usLlce fleld ln Lhls new decade-Lo bulld upon Lhe experlences of Lhose who have worked hard
Lo use new skllls, approaches, and research Lo englneer sysLems LhaL are vlslon-drlven, efflclenL,
and effecLlve. 8uL even more, we soughL Lo draw upon and draw LogeLher Lhe sLrongesL of Lhe
research flndlngs and Lhe besL of Lhe pracLlces, and consLrucL new ways of worklng LogeLher
Lowards Lhe goal we all share-fewer vlcLlms, safer communlLles.
Cur underlylng bellef ls LhaL we can lmprove ouLcomes lf crlmlnal [usLlce declslons are lnformed
by research. We called for Lhe consLrucLlon of a framework" for evldence-based declslon
maklng aL Lhe sysLem level. 8ecause lL does noL aLLempL Lo answer all quesLlons, provlde all
deLalls, or call for lmplemenLaLlon ln preclsely Lhe same way ln every communlLy, lL ls noL a
model. lL ls lnsLead lnLended Lo frame a purpose and a process for declslon maklng LhaL can be
applled Lo Lhe sysLem as a whole-Lo all Lhose enLerlng Lhe sysLem, regardless of Lhelr [usLlce
sysLem sLaLus, Lo all Lypes of cases, regardless of Lhelr severlLy, and Lo all sLakeholders,
regardless of Lhelr role.
1he lramework ldenLlfles Lhe key sLrucLural elemenLs of a sysLem lnformed by evldence. lL
deflnes a vlslon of safer communlLles. lL puLs forward Lhe bellef LhaL rlsk and harm reducLlon
are fundamenLal goals of Lhe [usLlce sysLem, and LhaL Lhese can be achleved wlLhouL sacrlflclng
offender accounLablllLy or oLher lmporLanL [usLlce sysLem ouLcomes. lL boLh expllcaLes Lhe
premlses and values LhaL underlle our [usLlce sysLem and puLs forward a proposed seL of
prlnclples Lo gulde evldence-based declslon maklng aL Lhe local level-prlnclples LhaL are,
Lhemselves, evldence-based. 1he lramework also hlghllghLs some of Lhe mosL groundbreaklng
of Lhe research-evldence LhaL clearly demonsLraLes LhaL we can reduce preLrlal mlsconducL
and offender recldlvlsm. lL ldenLlfles Lhe key sLakeholders who musL be acLlvely engaged ln a
collaboraLlve parLnershlp lf an evldence-based sysLem of [usLlce ls Lo be achleved. lL also seLs
ouL Lo begln Lo ouLllne some of Lhe mosL dlfflculL challenges we wlll face as we seek Lo
dellberaLely and sysLemaLlcally lmplemenL such an approach ln local communlLles.
ln sharlng Lhls lramework, we celebraLe all LhaL has come before lL and all Lhose laborlng so
hard on our sLreeLs, ln our courLrooms, and ln our [alls and prlsons. We bulld upon a foundaLlon
of research and noLeworLhy pracLlce from [urlsdlcLlons around Lhe counLry LhaL share a vlslon
of Lhe communlLles of Lomorrow-sLronger and more vlbranL as a resulL of less crlme, fewer
vlcLlms, resLored famllles, and offenders engaged ln healLhy llfesLyles.
3
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
AL Lhe same Llme, we openly acknowledge LhaL Lhere ls much work Lo be done. An earnesL
revlew of Lhe research reveals large bodles ln some areas and slgnlflcanL deflclLs ln oLhers,
parLlcularly ln preLrlal [usLlce and prosecuLlon. We musL work Lo flll Lhese. Larly revlewers of
Lhe lramework have suggesLed lL ls lncompleLe ln oLher ways, lncludlng lnsufflclenL guldance
around lmporLanL lmplemenLaLlon lssues. We agree and seek Lo answer Lhese concerns ln Lhe
nexL phase of our work. 1hese are buL a few of Lhe challenges LhaL lle ahead.
ln Lhe second phase of Lhls lnlLlaLlve, we wlll seek Lo ldenLlfy [urlsdlcLlons LhaL are lnLeresLed ln
plloLlng Lhe lramework. ln so dolng, we wlll work LogeLher Lo bulld lnformaLlon and Lools Lo
supporL lLs lmplemenLaLlon and Lo sLruggle Lhrough Lhe Lhorny lssues Lhls lramework wlll
surface. lL wlll undoubLedly challenge our processes, our pollcles, and even our phllosophles.
Lxperlences from earller crlmlnal [usLlce reform efforLs, such as communlLy pollclng,
demonsLraLe LhaL ma[or shlfLs ln approach are ofLen confronLed by challenges and meL wlLh
reslsLance. ln Llme, however, Lhose LhaL are well concelved, well documenLed, and LhaL produce
measurable ouLcomes Lake rooL and grow. lL ls our lnLenLlon, Lherefore, Lo engage ln a
dellberaLe process of documenLlng and evaluaLlng Lhe efforLs of plloL slLes. 1hls ls, afLer all, Lhe
essence of Lhls lnlLlaLlveXLo use research Lo lnform our approaches and Lo evaluaLe and learn
from Lhelr resulLs. 1hese lessons wlll offer valuable lnformaLlon Lo gulde us Lo a safer fuLure.
-Morrls 1hlgpen, ulrecLor
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
4
TABLE OF CONTENTS
F!rew!rd ................................................................................................................................... 2
Preface: The Evidence-Based Decisi!n Making in L!cal Criminal Justice
Systems Initiative.......................................................................................................................6
Initiative Appr!ach and Pr!ducts ........................................................................................ 6
Initiative Partners........................................................................................................... 6
Intr!ducti!n: A New Paradigm f!r the Justice System.............................................................7
Why a New Paradigm? .................................................................................................... 7
The Justice System Can D! Better ...................................................................................... 8
Other Systems Have Made Pr!gress; S! T!! Can the Justice System............................................ 8
The 100,000 Lives Campaign ............................................................................................ 9
Calling !n the Best in Pe!ple: The 1 Milli!n Fewer Victims Campaign ......................................10
Making the C!mparis!n between Healthcare and Justice System Ref!rm......................................10
An Overview !f Key Research Findings Related t! Risk Reducti!n and Their Implicati!ns
f!r the Justice System.....................................................................................................12
7 Ways t! Reduce Recidivism ..........................................................................................13
Secti!n 1: Underlying Premises............................................................................................... 17
The C!re Values !f the Justice System................................................................................17
Secti!n 2: The Key Decisi!n P!ints, Decisi!n Makers, and Stakeh!lders in the Criminal
Justice System.......................................................................................................................... 19
Key Decisi!n P!ints.......................................................................................................19
Key Decisi!n Makers and Stakeh!lder Gr!ups.......................................................................19
Secti!n 3: Examining Justice System Decisi!n Making Thr!ugh the Lens !f
Harm Reducti!n...................................................................................................................... 20
Crime Harms the Entire C!mmunity..................................................................................20
The Justice System Strives t! Achieve Risk and Crime Reducti!n ...............................................20
The Justice System Can Result in Harm Reducti!n.................................................................21
Achieving, Measuring, and Maintaining Harm Reducti!n and Advancing C!mmunity Wellness ..........22
Exhibit 1: Reducing Crime with Evidence-Based Opti!ns: What W!rks, and Benefits & C!sts ...........23
Secti!n 4: The Principles Underlying the Framew!rk............................................................ 25
Principle One: The pr!fessi!nal judgment !f criminal justice system decisi!n makers is
enhanced when inf!rmed by evidence-based kn!wledge ..........................................................25
Principle Tw!: Every interacti!n within the criminal justice system !ffers an !pp!rtunity
t! c!ntribute t! harm reducti!n........................................................................................26
Principle Three: Systems achieve better !utc!mes when they !perate c!llab!ratively.......................27
Principle F!ur: The criminal justice system will c!ntinually learn and impr!ve when pr!fessi!nals
make decisi!ns based !n the c!llecti!n, analysis, and use !f data and inf!rmati!n............................28
3
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Secti!n 5: Applying Evidence-Based Principles t! Practice ................................................... 30
A L!gic M!del f!r Harm Reducti!n Decisi!n Making at the System Level ....................................30
Evidence-Based Decisi!n Making System-Level Implementati!n Steps .........................................30
Evidence-Based Decisi!n Making L!gic M!del ......................................................................32
Secti!n 6: Key Challenges in Implementing this Framew!rk ................................................. 33
Risk Reducti!n and Evidence-Based Decisi!n Making in the C!ntext !f Sancti!ning Purp!se .............33
Plea Neg!tiati!ns ..........................................................................................................34
The Management !f L!w Level Offenses .............................................................................35
L!!king t! Phase II and Bey!nd.........................................................................................35
Secti!n 7: C!llab!rati!n: A Key Ingredient !f an Evidence-Based System............................. 36
Aligning the Criminal Justice System t! Achieve Harm Reducti!n ..............................................36
Bringing the Stakeh!lders t! the Table t! F!rm P!licy Teams....................................................36
Secti!n 8: Building Evidence-Based Agencies......................................................................... 38
Aligning Criminal Justice Agencies t! Achieve Harm Reducti!n.................................................38
Making What W!rks W!rk ...........................................................................................39
Appendix 1: Advis!ry B!ard Members: Evidence-Based Decisi!n Making in
L!cal Criminal Justice Systems Initiative................................................................................ 40
Appendix 2: Meth!d!l!gy Used t! C!mpute 1 Milli!n Fewer Victims ................................. 41
2.3 Milli!n Justice System Empl!yees.................................................................................41
Appendix 3: Research Findings Matrix................................................................................... 42
H!w t! Read the Matrix .................................................................................................42
Using Evidence t! Inf!rm Decisi!n Making ..........................................................................56
Appendix 4: 2009 Z!gby Internati!nal Public Opini!n Survey.............................................. 57
Appendix 5: Gl!ssary !f Terms................................................................................................ 58
Appendix 6: Bibli!graphy....................................................................................................... 59
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
6
PREFACE: THE EVIDENCE-BASED DECISION
MAKING IN LOCAL CRIMINAL JUSTICE
SYSTEMS INITIATIVE
ln !une 2008, Lhe naLlonal lnsLlLuLe of CorrecLlons awarded Lhe CenLer for LffecLlve ubllc
ollcy, ln parLnershlp wlLh Lhe reLrlal !usLlce lnsLlLuLe, Lhe !usLlce ManagemenL lnsLlLuLe, and
1he Carey Croup, a cooperaLlve agreemenL Lo address "Lvldence-8ased ueclslon Maklng ln
Local Crlmlnal !usLlce SysLems." 1he goal of Lhe lnlLlaLlve ls Lo bulld a sysLemwlde framework
(arresL Lhrough flnal dlsposlLlon and dlscharge) LhaL wlll resulL ln more collaboraLlve, evldence-
based declslon maklng and pracLlces ln local crlmlnal [usLlce sysLems. 1he lnlLlaLlve ls grounded
ln Lhe accumulaLed knowledge of Lwo decades of research on Lhe facLors LhaL conLrlbuLe Lo
crlmlnal reoffendlng and Lhe processes and meLhods Lhe [usLlce sysLem can employ Lo lnLerrupL
Lhe cycle of reoffense. 1he efforL seeks Lo equlp crlmlnal [usLlce pollcymakers ln local
communlLles wlLh Lhe lnformaLlon, processes, and Lools LhaL wlll resulL ln measurable
reducLlons of preLrlal mlsconducL and posL-convlcLlon reoffendlng.
INITIATIVE APPROACH AND PRODUCTS
1he prlnclple producL of Lhe lnlLlal 18-monLh phase of Lhls lnlLlaLlve ls Lhls documenL-
Aoltomewotkofot vlJeoce-8oseJoueclsloooMokloqoloolocoloctlmlooloIostlceo5ystemso(tbeo
ltomewotk)-deslgned Lo advance consLrucLlve change ln local level crlmlnal [usLlce declslon
maklng. 1he lramework descrlbes key crlmlnal [usLlce declslons, evldence-based knowledge
abouL effecLlve [usLlce pracLlces, and pracLlcal local level sLraLegles for applylng rlsk and harm
reducLlon prlnclples and Lechnlques. ln developlng Lhe lramework, Lhe lnlLlaLlve has drawn
upon Lhe experLlse of naLlonal lnsLlLuLe of CorrecLlons sLaff and Lhe lnlLlaLlve parLners, an
acLlve, mulLldlsclpllnary Advlsory CommlLLee, lnpuL from pollcymakers and pracLlLloners
(law enforcemenL offlclals, [all admlnlsLraLors, preLrlal offlclals, defense, prosecuLlon, courL
admlnlsLraLors, [udges, communlLy supervlslon represenLaLlves, vlcLlm advocaLes, and
clLy/counLy commlssloners and managers) Lhrough a serles of focus group dlscusslons and
lndlvldual lnLervlews, a llLeraLure revlew, Lhe experlences of an assembled group of non-
crlmlnal [usLlce, evldence-based managemenL experLs, and a publlc oplnlon survey. 1he
lramework wlll be plloL LesLed ln selecLed [urlsdlcLlons ln hase ll of Lhe lnlLlaLlve.
INITIATIVE PARTNERS
"67869X:;9X*::6<8=>6X+?@A=<X+;A=<B +9689=CAX,?D8=<6X(7D8=8?86
Madellne CarLer, rlnclpal (lnlLlaLlve ulrecLor) 1lm Murray, LxecuLlve ulrecLor
8achelle Clguere, rogram AssoclaLe Cherlse 8urdeen, Chlef CperaLlng Cfflcer
'E6X,?D8=<6X-C7CF6G678X(7D8=8?86 'E6X"C96BX/9;?H
8arry Mahoney, resldenL Mark Carey, resldenL
Llalne nugenL-8orakove, vlce-resldenL lrank uomurad, vlce-resldenL
7
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
What D! We Mean By Evidence?
In the justice system, the term
evidence is used in a variety !f
ways. It can refer t! items c!llected at
a crime scene, eyewitness acc!unts,
!r security camera f!!tage. These
types !f evidence are referred t! as
legal evidence.
F!r the purp!ses !f this Framew!rk,
h!wever, the term evidence is used
t! describe findings fr!m empirically
s!und s!cial science research. The
Framew!rk refers t! the results !f this
research as evidence-based p!licy
and practice.
It is imp!rtant t! n!te that all research
is n!t !f equal strength; this is
discussed further in Appendix 3.
INTRODUCTION: A NEW PARADIGM FOR THE
JUSTICE SYSTEM
Aoltomewotkofot vlJeoce-8oseJoueclsloooMokloqoloolocoloctlmlooloIostlceo5ystems deflnes Lhe
core prlnclples and acLlon sLraLegles LhaL crlmlnal [usLlce pollcymakers may employ Lo reduce
Lhe harm Lo communlLles caused by crlme. lL ls bullL on decades of experlence worklng wlLh
lndlvldual pollcymakers and pracLlLloners
and wlLh sLakeholder Leams ln local
[usLlce sysLems. lL ls based on Lhe
evldence from emplrlcal sLudles ln Lhe
flelds of organlzaLlonal managemenL,
evldence-based pracLlce ln crlmlnal [usLlce
and behavloral healLh, and collaboraLlve
processes. lL ls framed by a renewed
opLlmlsm regardlng Lhe poLenLlal Lhe
[usLlce sysLem has for reduclng harm and
vlcLlmlzaLlon and maklng communlLles
safer LhroughouL Lhe naLlon.
WHY A NEW PARADIGM?
1he [usLlce sysLem-along wlLh oLher
publlc secLor servlce sysLems-faces Lhe
21
sL
cenLury challenges of undersLandlng
emerglng sclence, LranslaLlng emplrlcal
flndlngs lnLo pollcy and pracLlce and, ln so
dolng, reLoollng long-held approaches,
and reLralnlng a workforce Lo adopL more
effecLlve pracLlces and embrace new
skllls. 1hese challenges are daunLlng, buL
crlLlcally lmporLanL.
Accordlng Lo Lhe u.S. ueparLmenL of
!usLlce, 8ureau of !usLlce SLaLlsLlcs,
1
67
of lndlvlduals released from prlson are
rearresLed wlLhln Lhree years afLer dlscharge. An esLlmaLed 30 of probaLloners supervlsed ln
Lhe communlLy are reconvlcLed for a new crlme. 1hese recldlvlsm raLes have remalned
relaLlvely sLable for decades.
2
lurLhermore, on any glven day, flve ouL of slx defendanLs
provlded wlLh a flnanclal release condlLlon are unable Lo make Lhe bond amounL seL by
Lhe courL.
3

1
Hughes & Wils!n, 2003.
2
Ibid.
3
C!hen & Reaves, 2008.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
8
1hese sLaLlsLlcs are parLlcularly soberlng when conslderlng Lhe Lens of Lhousands of new vlcLlms
each year
4
and Lhe lmmense loss of human llfe, dlgnlLy, and sense of safeLy Lhey experlence,
Lhe sLaggerlng cosLs of supporLlng law enforcemenL, Lhe courLs, correcLlons, and Lhe behavloral
and healLh sysLems, and, perhaps mosL lmporLanLly, Lhe rlpple effecL" of crlme on
communlLles ln Lerms of decaylng nelghborhoods, chlldren's exposure Lo vlolence, and
Lhe shlfLlng of resources from parks and schools Lo [alls and prlsons.
THE JUSTICE SYSTEM CAN DO BETTER
8esearch over Lhe pasL Lwo decades has demonsLraLed LhaL beLLer resulLs from our [usLlce
sysLem's efforLs and lnvesLmenLs can be reallzed. lor example, research demonsLraLes LhaL
a 30 reducLlon ln recldlvlsm ls posslble
3
lf Lhe [usLlce sysLem applles currenL knowledge
6
conslsLenLly and wlLh fldellLy. Moreover, Lhe research also shows LhaL appllcaLlon of Lhls
knowledge can produce slgnlflcanL cosL beneflLs Lo clLles, counLles, and sLaLes.
7
OTHER SYSTEMS HAVE MADE
PROGRESS; SO TOO CAN THE
JUSTICE SYSTEM
A 2000 reporL by Lhe lnsLlLuLe of Medlclne
(lCM)
8
revealed LhaL hosplLal medlcal
errors across Lhe naLlon resulLed ln a loss
of nearly 100,000 llves each year. 1he
reporL demonsLraLed LhaL Lhese mlsLakes
dld noL resulL from lndlvldual
lncompeLence, buL lnsLead were prlmarlly
Lhe resulL of sysLem fallures. Pe!ple
w!rking in health care are am!ng the
m!st educated and dedicated
w!rkf!rce in any industry, Lhe auLhors
wroLe. The pr!blem is n!t bad pe!ple;
the pr!blem is that the system needs t!
be made safer.
1he lCM reporL propelled Lhe medlcal
professlon lnLo a sLaLe of alarm. PealLhcare
professlonals had always vlewed
Lhemselves as beloq safe and sovloq llves,
noL cosLlng llves. Whlle Lhe medlcal code of
eLhlcs afflrms a commlLmenL Lo

4
In 2007 al!ne, U.S. residents age 12 !r !lder experienced appr!ximately 23 milli!n crimes. Of these, 17.5 milli!n (76%)
were pr!perty crimes, 5.2 milli!n (23%) were crimes !f vi!lence, and 194,100 (1%) were pers!nal thefts (BJS, 2008).
5
See Andrews et al., 1990; Andrews & B!nta, 1998; Landenberger & Lipsey, 2005; McGuire, 2002 & 2001.
6
Current kn!wledge refers t! inf!rmati!n regarding !ffender risk, dynamic risk fact!rs (i.e., crimin!genic needs), applying
interventi!ns appr!priately, and utilizing specific t!!ls and techniques.
7
A!s, Miller, & Drake, 2006b; see Secti!n 3 f!r additi!nal inf!rmati!n.
8
K!hn, C!rrigan, & D!nalds!n, 2000.
A nati!nal public !pini!n survey
c!mmissi!ned by the Nati!nal Institute !f
C!rrecti!ns and its partners in the Evidence-
Based Decisi!n Making in L!cal Criminal
Justice Systems Pr!ject illuminates the
publics views !n justice system practices and
recidivism reducti!n eff!rts. Key findings
fr!m this survey are included thr!ugh!ut this
d!cument. Further inf!rmati!n ab!ut the
study itself is c!ntained in Appendix 4.
When resp!ndents are t!ld that ab!ut half !f
the pe!ple released fr!m pris!n eventually g!
back t! pris!n and ab!ut a third !f th!se !n
pr!bati!n c!mmit new crimes, just 19%
indicate that these rates are acceptable; 80%
indicate that these rates are unacceptable.
Z!gby Internati!nal, August 2009
9
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
"The pr!blem with m!st
pe!ple is n!t that they aim
t!! high and miss the mark,
but that they aim t!! l!w
and hit it."
Michelangel!
The shared nature !f !ur g!al (and the fact that we did n!t seek t! exp!se any h!spital f!r
p!!r perf!rmance) changed the ten!r !f the campaign; it was a p!sitive initiative that called
!n the best in pe!ple, drawing them back t! the reas!ns they first were interested in this
w!rk. There was s! much untapped energy and s! much unleashed j!y, centered !n the
pr!viders' c!mmitment t! their patients.
Stanf!rd Graduate Sch!!l !f Business, 2008, p. 22.
c!mpetence and a commlLmenL Lo study, apply,
and advance scientific kn!wledge,
9
Lhe lCM reporL
revealed someLhlng qulLe dlfferenL. AcLlons on Lhe parL of
medlcal professlonals-and ln some cases lnacLlon-were
acLually lncreaslng Lhe deaLh raLe.
ln Lhe eyes of one organlzaLlon, Lhe reporL presenLed an
opporLunlLy. 1he lnsLlLuLe for PealLhcare lmprovemenL
(lPl) had been worklng for a decade Lo lnLroduce
sysLemlc change ln hosplLals ln an efforL Lo prevenL loss
of llfe due Lo human error. under Lhe leadershlp of
resldenL and CLC ur. uonald 8erwlck, Lhe lPl's
phllosophy was Lo vlew problems noL as a base meLal"
Lo be hldden and lgnored, buL as a deslrable Lreasure" or resource LhaL, when mlned and
undersLood, could lead Lo lmprovemenL and advancemenL. lor ur. 8erwlck, Lhe lCM reporL was
a verlLable gold mlne.
THE 100,000 LIVES CAMPAIGN
lPl launched a naLlonal campalgn Lo reduce Lhe devasLaLlng-and somewhaL embarrasslng-
loss of 100,000 accldenLal hosplLal and cllnlcal deaLhs Lo a more accepLable levelXzero. CreaLlng
Lhe slogan s!me is n!t a number; s!!n is n!t a time, 8erwlck launched Lhe 100,000 Llves
Campalgn. Pe proposed a meLhod Lo reduce 100,000 needless, error-drlven hosplLal deaLhs
wlLhln Lwo years.
lPl's efforLs were meL wlLh unprecedenLed success. WlLh roughly 3,100 of Lhe naLlon's
hosplLals-represenLlng 73 of Lhe avallable paLlenL bed space-enrolled ln Lhe lnlLlaLlve, an
esLlmaLed 122,342 deaLhs were prevenLed.
10
WhaL was Lhe key Lo Lhe success of Lhe 100,000 Llves Campalgn? Accordlng Lo !oe McCannon,
Lhe Campalgn's managerX
llve key lessons from Lhe lPl experlence-Lhose wlLh Lhe mosL dlrecL appllcaLlon Lo Lhe [usLlce
fleld-are lnLerspersed LhroughouL Lhe remalnder of Lhls documenL.

9
See American Medical Ass!ciati!n, 2002.
10
The C!mm!nwealth Fund, 2006.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
10
CALLING ON THE BEST IN PEOPLE: THE 1 MILLION FEWER
VICTIMS CAMPAIGN
1he lPl lnlLlaLlve soughL Lo save 100,000 llves Lhrough Lhe appllcaLlon of research-based
Lechnlques. 1he [usLlce sysLem could achleve equally dramaLlc resulLs.
lL ls esLlmaLed LhaL Lhe unlLed SLaLes could experlence 1,000,000 fewer vlcLlmlzaLlons.
11
1o
achleve Lhese resulLs, a slmllar approach Lo Lhe lPl lnlLlaLlve-adopLlng key sLraLegles LhaL are
evldence-based-musL be falLhfully adopLed. 1he publlc deserves and expecLs noLhlng less.
12
1hls lramework deflnes Lhe sLraLegy. hase ll of Lhe Lvldence-8ased ueclslon Maklng ln Local
Crlmlnal !usLlce SysLems lnlLlaLlve wlll begln LesLlng lL emplrlcally.
13
IHI LESSON #1: QUANTIFY THE GOAL
Drawing !n the advice !f experienced civic activist Gl!ria Steinem, IHI s!ught t!
m!bilize supp!rters and critics alike by flatly naming the pr!blem they were
attempting t! address (deaths as a result !f medical err!r) and quantifying the
g!al: The 100,000 Lives Campaign. S! p!werful was this message that when
the campaign was publicly launched at IHIs 16
th
Annual Nati!nal F!rum !n
Quality Impr!vement in December 2004, speaker after speaker expressed what
am!unted t! the equivalent !f m!ral !utrage that any !f their c!lleagues might
even c!nsider n!t j!ining the campaign. In the w!rds !f Sister Mary Jean Ryan,
President and CEO !f SSM Health Care, !ne !f the largest Cath!lic healthcare
systems in the c!untry: N! needless deaths is fundamental t! any healthcare
!rganizati!n, s! I think that CEOs sh!uld really w!rry m!re ab!ut n!t declaring
c!mmitment t! this g!al than t! declaring it. The less!n f!r criminal justice?
1 MILLION FEWER VICTIMS IS POSSIBLE;
THE TIME TO START IS NOW.
MAKING THE COMPARISON BETWEEN HEALTHCARE AND JUSTICE
SYSTEM REFORM
lPl's success ln reduclng unnecessary deaLhs ls well documenLed. Lessons learned from lPl are
lnLended Lo serve as helpful ways of Lhlnklng abouL advanclng evldence-based declslon maklng
ln Lhe [usLlce sysLem. WlLhouL quesLlon, Lhere are slgnlflcanL dlfferences ln Lhese sysLems.
PosplLals and cllnlcs are noL managed by lndlvlduals elecLed by Lhe general publlc. 1hey are noL
operaLed by a seL of acLors who, for all lnLenLs and purposes, are lndependenL and have
unllaLeral declslon maklng auLhorlLy. 1hey were noL deslgned wlLh a sysLem of checks and
balances ln mlnd, where one Leam of docLors produces evldence ln an aLLempL Lo prevall over
anoLher medlcal Leam. Cn Lhe oLher hand, whlle employees reporL Lo a slngle admlnlsLraLor and
share a common overarchlng goal, hosplLals are sLaffed by lndlvldual labor unlLs, wlLh dlsLlncL
areas of experLlse and responslblllLles, LhaL compeLe for llmlLed resources and work ln
envlronmenLs fraughL wlLh dlfferlng vlewpolnLs, communlcaLlon barrlers, and performance

11
See Appendix 2 f!r the meth!d!l!gy used t! c!mpute this figure.
12
The NIC-c!mmissi!ned 2009 Z!gby study reflects the publics expectati!n that, am!ng !thers, the current rate !f !ffender
failure is unacceptable; spending sh!uld be increased !n appr!aches pr!ven t! reduce crime; and criminal justice pr!fessi!nals
sh!uld rely !n research in their decisi!n making.
13
In Phase II, interested jurisdicti!ns will be c!mpetitively selected t! pil!t test the Framew!rk as presented in this d!cument.
Alth!ugh many !f the c!ncepts c!ntained in the Framew!rk have been implemented in part, the Phase II pil!t will be the first
kn!wn eff!rt t! fully integrate these strategies in wh!le. In additi!n, Phase II will serve as an !pp!rtunity t! further expand the
ideas and c!ncepts put f!rth in the Framew!rk and m!re specifically address the implementati!n issues it will generate.
11
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
pressures. 1hey coordlnaLe and collaboraLe wlLh conLracLed and governmenLal agencles,
lnsurance companles, and funders, and, as such, Lhey face many of Lhe same consLralnLs
professlonals ln oLher dlsclpllnes face.
romoLlng shlfLs ln aLLlLudes and behavlors LhaL supporL raLher Lhan defy a sysLem's vlslon,
overcomlng Lhe obsLacles presenLed by a large workforce, sLaylng currenL and conversanL wlLh
Lhe laLesL research, creaLlng change ln Lhe face of unprecedenLed work demands and ever
LlghLenlng resources, adapLlng Lo new Lechnologles, overcomlng sklll and knowledge deflclLs-
Lhese are buL a few of Lhe challenges common Lo large sysLems, wheLher [usLlce or healLhcare
or anoLher fleld. Whlle Lhe conLexL and complexlon of crlmlnal [usLlce cerLalnly dlffers from
Lhose of healLhcare, Lhe lessons of lPl bear conslderaLlon by Lhose lnLeresLed ln advanclng
change on a slgnlflcanL level.
The IHI Experience and Its Relevance t! Criminal Justice
There is n! d!ubt that alth!ugh there are similarities, there are als! many
differences between healthcare and justice systems. N!netheless, the IHI
experience is instructive in several ways. S!me !f the key less!ns have
relevance t! p!ssible ref!rms t! justice system practices. But perhaps m!re
imp!rtantly, the br!ader g!al !f impr!ving !utc!mes in the face !f daunting
challenges (e.g., c!mplicated systems and pr!cesses, multiple players,
c!mpeting g!als such as patient wellness versus c!st c!ntainment, etc.) is
perhaps the m!st fundamental similarity. In the w!rds !f !ne !f this initiatives
advis!rs:
IHI pr!ceeded fr!m the f!ll!wing premises, which are definitely applicable t! the
criminal justice system:
1. Things can be impr!ved.
2. Impr!vement will c!me !ver time, thr!ugh a successi!n !f acti!ns, each !f
which will pr!vide the !pp!rtunity f!r learning.
3. Better than the status qu! is, by definiti!n, "better" and we sh!uld n!t wait t!
s!lve everything bef!re beginning t! impr!ve s!me things.
4. We sh!uld be m!dest and realistic ab!ut !ur insights and abilities.
5. We need t! d! s!mething, because in the absence !f inf!rmed acti!n,
n!thing will change. And we can learn as we pr!ceed.
Jeffrey Pfeffer, Stanf!rd Graduate Sch!!l !f Business
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
12
AN OVERVIEW OF KEY RESEARCH FINDINGS RELATED TO RISK
REDUCTION AND THEIR IMPLICATIONS FOR THE JUSTICE SYSTEM
SLudles examlnlng Lhe quesLlon of how besL Lo prevenL fuLure crlme have lmporLanL
lmpllcaLlons for [usLlce sysLem pollcy and pracLlce. Whlle Lhese sLudles (and clLaLlons) are
deLalled more comprehenslvely ln Appendlx 3 of Lhls documenL-and Lhelr pollcy lmpllcaLlons
wlll be explored more Lhoroughly ln hase ll of Lhls lnlLlaLlve-Lhe slgnlflcance of Lhls body of
research ls lllusLraLed ln 7 Ways Lo 8educe 8ecldlvlsm" (pp. 13-13).
54% !f resp!ndents indicate that punishing th!se wh! c!mmit crimes sh!uld be the primary
purp!se !f the criminal justice system; 31% indicate that reducing the likelih!!d that
c!nvicted !ffenders will c!mmit new crimes sh!uld be the primary purp!se.
H!wever, 87% !f resp!ndents indicate they w!uld be m!re likely t! supp!rt alternatives t!
jail if research c!nsistently sh!wed there are ways !ther than jail t! reduce the likelih!!d that
n!n-vi!lent !ffenders will c!mmit new crimes.
When it c!mes t! vi!lent crime, 40% !f resp!ndents were in fav!r !f alternatives t! jail if
they w!uld reduce the likelih!!d !f re!ffense.
Z!gby Internati!nal, August 2009
13
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
7 WAYS TO REDUCE RECIDIVISM
!"X#$%&X'()*+,X-+.X)/+*X%++.++0.1&X&((2+X&(X/3.1&/45X)/+*X&(X).(44.13X%13X6)/0/1(7.1/6X1..3+"
8.+.%)6$X4/13/179 SLrucLured assessmenL Lools predlcL preLrlal mlsconducL and rlsk of reoffense more
effecLlvely Lhan professlonal [udgmenL alone.
14
8rlef screenlng Lools provlde a qulck assessmenL of rlsk,
comprehenslve Lools provlde lnformaLlon on rlsk Lo reoffend and effecLlve LargeLs of lnLervenLlon Lo reduce
fuLure crlme.
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 Law enforcemenL uses assessmenLs Lo lnform clLe versus arresL declslons,
prosecuLors and [udges use assessmenLs Lo lnform plea and senLenclng declslons, [alls use assessmenLs Lo
deLermlne houslng asslgnmenLs and work release placemenLs, and communlLy correcLlons uses assessmenLs
Lo deLermlne lnLenslLy of supervlslon.
:;%0<2.+X(4X<)%6&/6.X/0<2/6%&/(1+9 Law enforcemenL offlcers admlnlsLer brlef assessmenLs prlor Lo maklng
clLe/release declslons, preLrlal servlces and communlLy correcLlons conducL assessmenLs prlor Lo key
declslons.
="X#$%&X'()*+,X>/).6&X<)(7)%00/17X%13X/1&.)?.1&/(1+X&(X0.3/@0X%13X$/7$.)X)/+*X(44.13.)+"
8.+.%)6$X4/13/179 8ecldlvlsm raLes are reduced an average of 30 when medlum and hlgh rlsk offenders
recelve approprlaLe behavlor changlng programmlng.
13
Conversely, offenders assessed as low rlsk Lo
reoffend do noL beneflL from behavlor changlng programmlng
16
and are sllghLly more llkely Lo recldlvaLe
when Lhey are overly supervlsed or programmed.
17
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 lor low rlsk offenders, prosecuLors use dlverslonary programs, prosecuLors
and [udges avold excesslve condlLlons, defense counsel advocaLes for low lnLenslLy lnLervenLlons, communlLy
correcLlons uses mlnlmal supervlslon. !udges, prosecuLors, and defense counsel LargeL medlum and hlgh rlsk
offenders for programmlng deslgned Lo poslLlvely lnfluence behavlor.
:;%0<2.+X(4X<)%6&/6.X/0<2/6%&/(1+9XAgencles performlng assessmenLs color code case flles of hlgh, medlum,
and low rlsk offenders for easy ldenLlflcaLlon by declslon makers, communlLy supervlslon agencles use call-ln
or klosk reporLlng for low rlsk offenders, LreaLmenL programs modlfy admlsslon crlLerla Lo admlL only
medlum and hlgh rlsk offenders.

14
Andrews & B!nta, 1998; Andrews, B!nta, & W!rmith, 2006; Andrews et al., 1990; B!nta, 2007; Gendreau, Little, & G!ggin,
1996; Gr!ve & Meehl, 1996; Gr!ve et al., 2000; Harris, 2006; Smith, Gendreau, & Swartz, 2009.
15
Andrews, 2007; Andrews & B!nta, 2007; Andrews, B!nta, & W!rmith, 2006; Andrews & D!wden, 2007; Andrews,
D!wden, & Gendreau, 1999; B!nta, 2007; D!wden, 1998; Gendreau, G!ggin, & Little, 1996; Lipsey & Cullen, 2007; Smith,
Gendreau, & Swartz, 2009.
16
Ibid.
17
Andrews & B!nta, 2007; B!nta, Wallace-Capretta, & R!!ney, 2000; Cullen & Gendreau, 2000; Gendreau, G!ggin, Cullen, &
Andrews, 2001; L!wenkamp & Latessa, 2004; L!wenkamp, Latessa, & H!lsinger, 2006.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
14
7 WAYS TO REDUCE RECIDIVISM
A"X#$%&X'()*+,XB(6@+X/1&.)?.1&/(1+X4()X0.3/@0X%13X$/7$.)X)/+*X(44.13.)+X(1X&$./)X/13/?/3@%2X6)/0/1(7.1/6X
1..3+"
8.+.%)6$X4/13/179 CognlLlve behavloral programs are generally Lhe mosL effecLlve programmlng lnLervenLlons
for hlgher rlsk offenders.
18
lurLhermore, employlng program lnLervenLlons LhaL lnfluence Lhe LralLs LhaL lead
Lo fuLure crlme (l.e., crlmlnogenlc needs) yleld sLronger reducLlons ln recldlvlsm (up Lo an average of 30
reducLlon).
19
1he neL value (Lhe cosL of Lhe program less Lhe savlngs derlved from prevenLlng crlme) of Lhe
average LargeLed, evldence-based cognlLlve behavloral program, uslng a cosL/beneflL formula, ls $10,299 per
adulL offender.
20
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 !udges ensure LhaL senLenclng condlLlons allgn wlLh speclflc crlmlnogenlc
needs, communlLy correcLlons and LreaLmenL provlders use assessmenL lnsLrumenLs Lo ldenLlfy offenders'
crlmlnogenlc LralLs, LreaLmenL provlders avold one slze flLs all" programs, cognlLlve behavloral servlces are
sysLemaLlcally uLlllzed.
:;%0<2.+X(4X<)%6&/6.X/0<2/6%&/(1+9X1reaLmenL provlders provlde program llsLlngs LhaL ldenLlfy Lhe
crlmlnogenlc needs Lhelr servlces address, communlLy correcLlons refers offenders Lo programs based upon
Lhe maLch beLween offenders' needs and programs' servlces, counLy execuLlves/managers ensure LhaL
servlce conLracLs wlLh LreaLmenL provlders lnclude accounLablllLy measures Lo make cerLaln LhaL Lhe servlces
provlded lnclude cognlLlve behavloral lnLervenLlons.
C"X#$%&X'()*+,X8.+<(13X&(X0/+6(13@6&X'/&$X+'/4&1.++DX6.)&%/1&5DX%13X<)(<()&/(1%2/&5"
8.+.%)6$X4/13/179 CraduaLed sancLlons (l.e., sancLlons LhaL lncrease ln severlLy based on Lhe number and
naLure of acLs of mlsconducL) lncrease compllance wlLh supervlslon and LreaLmenL.
21
SwlfL,
22
cerLaln,
23
and
proporLlonal
24
acLlons LhaL reflecL dlsapproval of behavloral mlsconducL are more effecLlve ln reduclng
recldlvlsm Lhan acLlons LhaL are dlsproporLlonaLe, delayed, or lnconslsLenL.
:;%0<2.X(4X<(2/65X/0<2/6%&/(1+9 CourL admlnlsLraLors develop pollcles Lo move cases swlfLly Lhrough Lhe
courL sysLem, [udges, prosecuLors, and communlLy correcLlons agencles esLabllsh vlolaLlon declslon maklng
guldellnes LhaL Lake lnLo accounL Lhe rlsk of Lhe offender and Lhe severlLy of Lhe vlolaLlon behavlor, all
vlolaLlon behavlor ls responded Lo ln some fashlon, [udges and communlLy correcLlons sLreamllne
procedures LhaL allow for swlfL acLlon followlng offender mlsbehavlor.
:;%0<2.X(4X<)%6&/6. /0<2/6%&/(1+9 CourL admlnlsLraLors manage dockeLs LhaL sLreamllne case processlng,
communlLy correcLlons uses a declslon maklng Lool Lo ald supervlslon offlcers ln sLrucLurlng Lhelr responses
Lo vlolaLlon behavlor, communlLy correcLlons provldes admlnlsLraLlve sancLlonlng processes Lo address
mlsbehavlor qulckly.

18
Andrews, 2007; A!s, Miller, & Drake, 2006a; Landenberger & Lipsey, 2005; Lipsey & Landenberger, 2006; Lipsey,
Landenberger, & Wils!n, 2007; Smith, Gendreau, & Swartz, 2009.
19
Andrews, 2007; Andrews et al., 1990; Smith, Gendreau, & Swartz, 2009.
20
A!s, Miller, & Drake, 2006b.
21
Andrews & Janes, 2006; Burke, 2004; Harrell et al., 2003; Hay, 2001; Taxman, S!ule, & Gelb, 1999; Tayl!r & Martin, 2006.
22
Rhine, 1993.
23
Grasmack & Bryjak, 1980; Nich!ls & R!ss, 1990; Patern!ster, 1989.
24
T!nry, 1996; V!n Hirsh, 1993.
13
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
7 WAYS TO REDUCE RECIDIVISM
E"X#$%&X'()*+,X-+.X0().X6%))(&+X&$%1X+&/6*+"
8.+.%)6$X4/13/179 1he use of lncenLlves and poslLlve relnforcemenL are effecLlve ln promoLlng behavloral
change.
23
oslLlve relnforcemenL should be provlded aL a raLe of four relnforcers for every expresslon of
dlsapproval (or sancLlon).
26
8esearch demonsLraLes LhaL Lhls formula enhances offenders' moLlvaLlon Lo
conLlnue exhlblLlng prosoclal behavlors and aLLlLudes.
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 !udges and communlLy correcLlons develop pollcles around Lhe sLrucLured
and speclflc use of rewards Lo relnforce poslLlve behavlor.
:;%0<2.+X(4X<)%6&/6.X/0<2/6%&/(1+9 uefense counsel requesLs revlew hearlngs when cllenLs reach slgnlflcanL
mllesLones, communlLy correcLlons acknowledges progress Lhrough Lhe posLlng of awards, wrlLlng leLLers of
afflrmaLlon, provldlng compllmenLary bus passes, pralslng offenders' behavlor Lo Lhelr famllles, or reduclng
reporLlng requlremenLs, law enforcemenL acknowledges law abldlng behavlor of known offenders.
F"X#$%&X'()*+,X>.2/?.)X+.)?/6.+X/1X1%&@)%2X.1?/)(10.1&+X'$.).X<(++/G2."
8.+.%)6$X4/13/179 AlLhough LreaLmenL servlces provlded ln sLrucLured (e.g., resldenLlal, lnsLlLuLlonal) seLLlngs
are demonsLraLed Lo be effecLlve, servlces dellvered ln naLural envlronmenLs (l.e., seLLlngs ln offenders'
lmmedlaLe surroundlngs LhaL mosL closely resemble prosoclal, supporLlve envlronmenLs) lmprove offenders'
bondlng Lo Lhe prosoclal communlLy and ald ln reduclng recldlvlsm.
27
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 Law enforcemenL refers Lo communlLy-based crlsls servlces for offenders
wlLh menLal healLh condlLlons, [udges and prosecuLors use communlLy-based raLher Lhan resldenLlal or
lnsLlLuLlonally based programs when Lhe safeLy of Lhe communlLy ls noL ln [eopardy, counLy
execuLlves/managers provlde supporL for fundlng and zonlng communlLy-based programmlng opLlons.
:;%0<2.+X(4X<)%6&/6.X/0<2/6%&/(1+9 !udges, prosecuLors, defense counsel, communlLy correcLlons, and oLhers
Lake lnvenLory of avallable servlces Lo ensure a conLlnuum of servlce opLlons, communlLy correcLlons uLlllzes
prosoclal famlly members, employers, and menLors Lo supporL Lhe offender, resource dlrecLorles are
developed and shared among sLakeholders.
H"X#$%&X'()*+,XI%/)X+%16&/(1+X'/&$X/1&.)?.1&/(1+X&$%&X%33).++X6)/0/1(7.1/6X1..3+"
8.+.%)6$X4/13/179 8esearch demonsLraLes LhaL sancLlons wlLhouL programmlng (e.g., booL camps wlLhouL a
LreaLmenL componenL,
28
elecLronlc monlLorlng,
29
lnLenslve supervlslon,
30
lncarceraLlon
31
) do noL conLrlbuLe
Lo reducLlons ln reoffense raLes. ModesL lncreases ln Llme served may even lncrease recldlvlsm.
32
:;%0<2.+X(4X<(2/65X/0<2/6%&/(1+9 rosecuLors and [udges employ a comblnaLlon of sancLlons and behavlor
changlng programmlng for purposes of rlsk reducLlon, counLy execuLlves/managers fund a balance of
behavlor changlng programmlng and accounLablllLy measures, communlLy correcLlons agencles address
offender mlsbehavlor wlLh behavlor changlng, raLher Lhan solely punlLlve, responses.

25
Andrews & B!nta, 2006; Cullen & Gendreau, 2000; Drake & Barn!ski, 2008; Latessa, Cullen, & Gendreau, 2002; Nati!nal
Research C!uncil, 2007; Petersilia, 2007; Petersilia, 2004; Taxman, S!ule, & Gelb, 1999.
26
Andrews & B!nta, 2006; Gendreau & G!ggin, 1996; Gendreau, Little, & G!ggin, 1996.
27
Andrews, 2007; B!nta et al., 2002; Clear & Sumter, 2002; Elgelk! et al., 1998; Emrick et al., 1993; Galanter, 1993; Higgins
& Silverman, 1999; Meyers et al., 2002; Meyers & Smith, 1997; OC!nn!r & Perryclear, 2003; Shapir! & Schwartz, 2001.
28
MacKenzie et al., 1995; MacKenzie, Wils!n, & Kider, 2001.
29
MacKenzie, 1997.
30
A!s, Miller, & Drake, 2006a; A!s, Phipps, Barn!ski, & Lieb, 2001; Petersilia, 1999; Petersilia & Turner, 1993; T!nry, 1997.
31
Andrews, 2007; Gendreau, G!ggin, & Cullen, 1999; Gendreau, G!ggin, Cullen, & Andrews, 2001.
32
Hughes, Wils!n, & Beck, 2001; Langan & Levin, 2002; Smith, G!ggin, & Gendreau, 2002.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
16
IHI LESSON #2: MAKE IT PROFOUNDLY SIMPLE
IHI realized that establishing a l!fty g!al and leaving it t! h!spital staff acr!ss the
c!untry t! find their !wn ways t! reach it was a recipe f!r failure. Ad!pting
evidence-based practice places an additi!nal burden !n decisi!n makers and
staff. On t!p !f meeting their r!utine resp!nsibilities, they have t! c!llect and
analyze research, determine the !ptimal meth!d t! integrate it int! the existing
culture, and define the practical steps t! implementing it !n a day-t!-day basis.
These additi!nal tasks layered !ver existing duties can easily create resistance
even !n the part !f the best-intenti!ned pr!fessi!nals. IHI s!ught t! ameli!rate
this danger by defining, !n behalf !f the pr!fessi!n, six evidence-based steps
(such as using pr!ven pr!cesses t! prevent ventilat!r-related pneum!nia,
elevating the head !f the patients bed t! between 30 and 45 degrees at all
times, and reducing surgical !n-site infecti!ns thr!ugh the use !f simple
pr!cedures such as frequent and careful hand washing). The less!n f!r criminal
justice?
TRANSLATE EVIDENCE-BASED RESEARCH INTO
PROFOUNDLY SIMPLE STRATEGIES.
17
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
SECTION 1: UNDERLYING PREMISES
ln developlng Lhe lramework, Lhe followlng premlses were acknowledgedX
1. Clven Lhe currenL sLaLe of knowledge ln Lhe [usLlce and Lhe behavloral healLh flelds, beLLer
ouLcomes Lhan have been reallzed ln Lhe pasL can be expecLed.
2. 8eLLer ouLcomes wlll be derlved lf
exlsLlng resources (lncludlng non-
lncarceraLlve and lncarceraLlve) are used
more effecLlvely.
3. lf, Lhrough Lhe supporL of emplrlcal
evldence, a loglc model for crlmlnal
[usLlce processes and declslon maklng
33
ls deflned and lmplemenLed wlLh
fldellLy, Lhese lmproved ouLcomes wlll
resulL.
4. 1he careful collecLlon and analysls of
daLa and lnformaLlon regardlng Lhe
lmplemenLaLlon of Lhe loglc model wlll
produce clear and convlnclng evldence
Lo gulde furLher advancemenLs ln pollcy
and pracLlce. ln Lhls way, [usLlce sysLem
ouLcomes can conLlnue Lo lmprove over
Llme.
34
3. 1he u.S. [usLlce sysLem has developed around a seL of core values. 1hese are Lo be honored
and proLecLed. 1hey provlde a foundaLlon upon whlch Lhls lramework ls consLrucLed.
THE CORE VALUES OF THE JUSTICE SYSTEM
1he u.S. [usLlce sysLem" ls ln acLuallLy many [usLlce sysLems-each governed by a dlfferenL
comblnaLlon of sLaLe and federal law and each made up of many dlfferenL organlzaLlonal
componenLs. ln Lhelr mlsslons and ln Lhelr lnvolvemenL ln lndlvldual cases, Lhese componenLs
ofLen have speclflc goals LhaL vary conslderably and are someLlmes ln confllcL. Powever, Lhelr
work ls grounded ln values LhaL have a long hlsLory ln Lhe u.S. and LhaL are wldely embraced
across Lhe many componenLs of any [usLlce sysLem. 1hese core values gulde Lhe developmenL
and lmplemenLaLlon of Lhe lramework. 1hey lnclude Lhe followlngX
public safety (assurlng Lhe proLecLlon of Lhe communlLy and of lndlvlduals),
fairness (assurlng LhaL processes ln Lhe courLs and oLher [usLlce sysLem agencles are falr
and free from blas),

33
A l!gic m!del is a graphic representati!n !f the the!ry behind a c!nceptual framew!rk; see Secti!n 5 f!r m!re inf!rmati!n.
34
Phase II !f the Evidence-Based Decisi!n Making in L!cal Criminal Justice Systems Pr!ject will include an independent
evaluati!n !f the pil!t sites t! determine their fidelity t! the l!gic m!del and the Framew!rk, and the results that are derived.
M!dificati!ns t! the l!gic m!del will be made as needed.
Outc!mes under a risk reducti!n
m!del are defined as decreases in the
rate !r severity !f re!ffense by
!ffenders, decreases in the harm
caused t! c!mmunities as a result !f
crime, increases in the level !f
satisfacti!n with the justice system by
victims, and increases in the level !f
public c!nfidence in the justice system.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
18
individual liberty (recognlzlng LhaL a prlmary funcLlon of Lhe [usLlce sysLem ls Lo proLecL
Lhe rlghLs and freedoms of lndlvlduals and Lo guard agalnsL an arblLrary exerclse of
governmenLal auLhorlLy),
respect f!r the rights, needs, and c!ncerns !f victims !f crime;
respect f!r the rights !f pers!ns accused !f crime;
respect f!r the rule !f law;
discreti!n (recognlzlng LhaL Lhe sound and lnformed exerclse of dlscreLlon, wlLhln Lhe
parameLers esLabllshed by law, ls an essenLlal parL of [usLlce sysLem declslon maklng), and
appreciati!n f!r differences in perspectives and practices acr!ss jurisdicti!ns
(recognlzlng LhaL local dlfferences ln pollcy and pracLlce exlsL and can fosLer lnnovaLlon and
conLrlbuLe Lo lmprovemenLs ln pracLlce and ouLcomes).
19
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
SECTION 2: THE KEY DECISION POINTS, DECISION
MAKERS, AND STAKEHOLDERS IN THE CRIMINAL
JUSTICE SYSTEM
1hls lramework was developed wlLh key declslon polnLs, declslon makers, and sLakeholders
ln mlnd.
33
KEY DECISION POINTS
ArresL declslons (clLe, deLaln, dlverL, LreaL, release)
reLrlal sLaLus declslons (release on recognlzance, release on flnanclal bond, release wlLh
supervlslon condlLlons, deLaln, vlolaLlon response, supervlslon condlLlons reassessmenL)
Charglng declslons (charge, dlverL, defer, dlsmlss)
lea declslons (plea Lerms)
SenLenclng declslons (senLence Lype, lengLh, Lerms and condlLlons)
Local lnsLlLuLlonal lnLervenLlon declslons (securlLy level, LreaLmenL lnLervenLlons)
Local lnsLlLuLlonal release declslons (Llmlng of release, condlLlons of release)
CommunlLy lnLervenLlon declslons (supervlslon level, supervlslon condlLlons, LreaLmenL
lnLervenLlons)
vlolaLlon response declslons (response level, sancLlons, LreaLmenL lnLervenLlons)
ulscharge from crlmlnal [usLlce sysLem declslons (Llmlng of dlscharge)
KEY DECISION MAKERS AND STAKEHOLDER GROUPS
Law enforcemenL offlcers
reLrlal offlclals
vlcLlm advocaLes
rosecuLors
uefense aLLorneys
!all admlnlsLraLors
CourL admlnlsLraLors
!udges
robaLlon/arole offlclals
ClLy/CounLy managers/Commlssloners
SLaLe leglslaLors
CommunlLy represenLaLlves (e.g., clvlc leaders, members of falLh-based organlzaLlons,
servlce provlders)

35
While this list is n!t exhaustive, f!r purp!ses !f this Framew!rk these are c!nsidered the primary decisi!n p!ints, decisi!n
makers, and stakeh!lders. Omissi!n !f !ther stakeh!lders, including defendants/!ffenders and their family members,
researchers, and !thers, is n!t intended t! diminish the imp!rtant c!ntributi!n they play t! advancing evidence-based decisi!n
making.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
20
SECTION 3: EXAMINING JUSTICE SYSTEM DECISION
MAKING THROUGH THE LENS OF HARM
REDUCTION
CRIME HARMS THE ENTIRE
COMMUNITY
Whlle crlme ofLen resulLs ln Lhe speclflc
paln and sufferlng of lndlvlduals, all
crlme dlsrupLs Lhe fabrlc of our
communlLles, [eopardlzes our lndlvldual
and collecLlve sense of safeLy, and
exLracLs a flnanclal penalLy by dlverLlng
publlc monles Lo Lhe [usLlce sysLem LhaL
mlghL oLherwlse supporL bulldlng Lhe
healLh of our communlLles (e.g., schools
for our chlldren, parks for our famllles).
Lveryone ls a vlcLlm of crlme. And whlle
some suffer more Lhan oLhers, everyone
beneflLs-dlrecLly and lndlrecLly-from
crlme prevenLlon and reducLlon efforLs.
THE JUSTICE SYSTEM STRIVES
TO ACHIEVE RISK AND CRIME
REDUCTION
8lsk reducLlon resulLs from Lhe
successful appllcaLlon of prlnclples and
Lechnlques LhaL have been
demonsLraLed Lo reduce Lhe llkellhood,
frequency, or severlLy of reoffense by
known defendanLs/offenders.
36
A
growlng body of sclence provldes [usLlce
sysLem professlonals wlLh Lhe lnformaLlon and Lools Lo esLlmaLe Lhe level of rlsk an lndlvldual
poses and provldes prlnclples for lnLervenLlon Lo reduce Lhe llkellhood, severlLy, and/or
frequency of fuLure rlsk. 1hls approach does noL devalue offender accounLablllLy. ln facL, lL
assures LhaL Lhe sLeps Laken by [usLlce sysLem declslon makers Lo hold offenders accounLable
produce Langlble and meanlngful ouLcomes-reduced rlsk Lo reoffend. A rlsk reducLlon
phllosophy Lherefore poslLs tbeooffeoJet as Lhe focus.

36
This d!cument is intended t! address the entire criminal justice system and as such there is equal interest in pretrial and p!st-
sentence system activities and !ffender c!nduct. F!r sake !f ease t! the reader, the use !f defendant/!ffender is n!t repeated.
The term !ffender is used t! refer t! b!th defendants and !ffenders.
Harm reducti!n, as used in the
Framew!rk, refers t! decreases in the
ill effects !f crime experienced br!adly
by c!mmunities (e.g., res!urces
all!cated t! the justice system that
c!uld !therwise be directed t!
alternative public pri!rities, unsafe
streets, aband!ned businesses, etc.),
by victims (e.g., fear !f reprisal !r
revictimizati!n, financial l!sses, etc.),
by citizens (e.g., lack !f c!nfidence in
c!mmunity pr!tecti!n eff!rts,
generalized fears !f victimizati!n,
etc.), by families !f !ffenders (e.g.,
l!ss !f wages by a family member wh!
is justice-system inv!lved, inability !f
incarcerated fathers/m!thers t! fulfill
their parenting r!les, etc.), and by
!ffenders themselves (e.g.,
h!melessness, unempl!yment, etc.).
21
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
THE JUSTICE SYSTEM CAN RESULT IN HARM REDUCTION
AlLhough Lhe lmpacL of crlme ls generally LhoughL of ln Lerms of Lhe offender and Lhe vlcLlm,
crlme affecLs Lhe healLh and welfare of Lhe communlLy ln a much broader way. A harm
reducLlon phllosophy poslLs tbeocommoolty as Lhe focus and acknowledges Lhese broad
lmpacLs. Some of Lhese very slgnlflcanL collaLeral consequences are
hlgh cosLs of lncarceraLlon, leadlng Lo lncreased Laxes for resldenLs and buslnesses,
eroslon of properLy values and decreased properLy Lax revenue, leadlng Lo decreaslng Lax
bases as resldenLs move ouL of crlme-plagued nelghborhoods,
loss of buslness revenue ln hlgh crlme nelghborhoods, leadlng Lo fewer [ob opporLunlLles
for Lhe communlLy,
unravellng of resldenLs' sense of
commlLmenL Lo local communlLles, whlch
ls crlLlcal Lo ensurlng safe, healLhy, and
prosperous nelghborhoods,
growLh of crlme culLures, where crlmlnal
acLlvlLy ls so commonplace lL becomes
vlewed as a normal parL of llfe,
negaLlve lnfluence of crlmlnal behavlor
from one generaLlon Lo Lhe nexL,
dlsrupLlon of normal everyday acLlvlLles
LhaL promoLe soclal lnLeracLlon and
vlbranL communlLles,
overall dlsLrusL of Lhe [usLlce sysLem Lo
be responslve Lo communlLy, vlcLlm, or
offender needs,
unsafe condlLlons for chlldren-
parLlcularly ln vlolenL nelghborhoods,
places where drugs are manufacLured
(e.g., meLh labs), and schools plagued by
gangs,
removal of slgnlflcanL segmenLs of some demographlc subgroups (e.g., males ln age groups
prone Lo hlgh crlme) from Lhe communlLy, and
repercusslons (e.g., flnanclal, emoLlonal) experlenced by famllles and chlldren of
lncarceraLed persons.
Actuarial instruments are !ne
example !f the research-supp!rted
t!!ls available t! criminal justice
pr!fessi!nals. These instruments
enable pr!fessi!nals t! assess the
level !f risk an individual !ffender is
likely t! p!se. While these instruments
cann!t determine any !ne individuals
risk level with abs!lute certainty, they
canlike the actuarial t!!ls used t!
determine that a 17-year-!ld b!y is
m!re likely t! get int! a traffic
accident than a 40-year-!ld w!man
statistically predict the likelih!!d !f an
!utc!me am!ng a large gr!up !f
individuals with similar characteristics.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
22
ACHIEVING, MEASURING, AND MAINTAINING HARM REDUCTION AND
ADVANCING COMMUNITY WELLNESS
!usLlce sysLems focused on harm reducLlon and communlLy wellness can creaLe real and
meanlngful change aL Lhe communlLy level. undersLandlng whaL Lhese changes are and how Lo
measure Lhem requlres esLabllshlng a seL of Langlble performance measures. 8roadly, Lhese
performance measures can be grouped lnLo four caLegorlesX1) lncreases ln publlc safeLy, 2)
lmprovemenLs ln Lhe wellness of Lhe communlLy, 3) lncreases ln saLlsfacLlon wlLh Lhe [usLlce
sysLem, and 4) lmprovemenLs ln Lhe soclal and flscal cosLs of [usLlce sysLem lnLervenLlons.
37,38
Lxamples of posslble performance measures lnclude Lhe followlngX
Increases in public safety, as measured by
reduced physlcal, psychologlcal, and economlc
harm Lo prlmary vlcLlms,
fewer released offenders arresLed for new
offenses,
longer elapsed Llme from release Lo reoffense,
fewer released offenders arresLed for a more
serlous offense Lhan Lhelr orlglnal offense,
decreased average number of new offenses for released offenders,
fasLer case processlng Llmes (l.e., shorLer elapsed Llme from arresL Lo flnal ad[udlcaLlon) LhaL
decrease Lhe llkellhood of preLrlal mlsbehavlor and lncrease swlfLness of punlshmenL,
fewer people vlcLlmlzed by released offenders,
fewer vlcLlms revlcLlmlzed" by orlglnal perpeLraLors,
decreased number of proLecLlon order/sLay-away orders vlolaLed,
fewer reporLs of crlme from hoL spoLs" lnvolvlng elLher known offenders or new offenders,
and
lncreases ln Lhe proporLlon of [all and prlson beds occupled by hlgh rlsk offenders compared
Lo low rlsk offenders.
Impr!ved c!mmunity wellness, as measured by
decreased number of drug/alcohol-relaLed Lrafflc accldenLs,
fewer drug/alcohol-relaLed Lrafflc faLallLles,
decreases ln emergency-room admlsslons for crlme-relaLed and drug-relaLed ln[urles,
lncreased number of drug-free bables born,
fewer chlld welfare lnLervenLlons ln famllles of offenders,
lncreases ln Lhe number of people successfully compleLlng LreaLmenL programs, and
fewer [all and prlson admlsslons for people wlLh menLal healLh lssues.

37
In 2007, taxpayers spent $1,223 per h!useh!ld !n the criminal justice system; see A!s, 2009.
38
Exhibit 1 pr!vides an analysis !f the c!sts and benefits !f 571 studies !n specific crime reducti!n strategies. Analyses !f this
kind equip p!licymakers t! make inf!rmed ch!ices regarding the investment !f res!urces and the benefits that can be derived
fr!m these investments.
93% !f resp!ndents indicate the
criminal justice system sh!uld make
neighb!rh!!ds safer.
Z!gby Internati!nal, August 2009
23
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
E"HIBIT 1: REDUCING CRIME WITH EVIDENCE-BASED OPTIONS:
WHAT WORKS, AND BENEFITS & COSTS
SourceXAos, Mlller, & urake, 2006b
Washingt!n State Institute f!r
Public P!licy
Estimates as !f Oct!ber, 2006
N!tes:
n/e means n!t estimated at this time
Preventi!n pr!gram c!sts are partial c!sts, pr!-
rated t! match crime !utc!mes
Effect !n Crime
Outc!mes
Percent change in
crime !utc!mes &
the number !f
evidence-based
studies !n which the
estimate is based (in
parentheses)
Benefits and C!sts
(Per Participant, Net Present Value, 2006 D!llars)
Benefits t!
Crime
Victims
(!f the
reducti!n)
Benefits t!
Taxpayer
(!f the
reducti!n)
C!sts
(marginal
pr!gram c!st,
c!mpared t!
the c!st !f
alternative)
Benefits
(t!tal)
Minus
C!sts
(per
participant)
Pr!grams f!r Pe!ple in the Adult Offender System
vocaLlonal educaLlon ln prlson -9.0 (4) $8,114 $6,806 $1,182 $13,738
lnLenslve supervlslonXLreaLmenL-
orlenLed programs
-16.7 (11) $9,318 $9,369 $7,124 $11,363
Ceneral educaLlon ln prlson (baslc
educaLlon or posL-secondary)
-7.0 (17) $6,323 $3,306 $962 $10,669
CognlLlve behavloral Lherapy ln prlson or
communlLy
-6.3 (23) $3,638 $4,746 $103 $10,299
urug LreaLmenL ln communlLy -9.3 (6) $3,133 $3,493 $374 $10,034
CorrecLlonal lndusLrles ln prlson -3.9 (4) $3,360 $4,496 $417 $9,439
urug LreaLmenL ln prlson (LherapeuLlc
communlLles or ouLpaLlenL)
-3.7 (20) $3,133 $4,306 $1,604 $7,833
AdulL drug courLs -8.0 (37) $4,393 $4,703 $4,333 $4,767
LmploymenL and [ob Lralnlng ln
communlLy
-4.3 (16) $2,373 $2,386 $400 $4,339
LlecLronlc monlLorlng Lo offseL [all Llme 0 (9) $0 $0 -$870 $870
Sex offender LreaLmenL ln prlson wlLh
afLercare
-7.0 (6) $6,442 $2,883 $12,383 -$3,238
lnLenslve supervlslonXsurvelllance-
orlenLed programs
0 (23) $0 $0 $3,747 -$3,747
WashlngLon's uangerously MenLally lll
Cffender program
-20.0 (1) $18,020 $13,116 n/e n/e
urug LreaLmenL ln [all -4.3 (9) $2,481 $2,636 n/e n/e
AdulL booL camps 0 (22) $0 $0 n/e n/e
uomesLlc vlolence educaLlon/cognlLlve
behavloral LreaLmenL
0 (9) $0 $0 n/e n/e
!all dlverslon for menLally lll offenders 0 (11) $0 $0 n/e n/e
Llfe skllls educaLlon programs for adulLs 0 (4) $0 $0 n/e n/e
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
24
Increased satisfacti!n with the criminal justice system, as measured by
lncreased number of vlcLlms saLlsfled wlLh Lhe [usLlce sysLem's responses,
lncreased number of offenders maklng resLlLuLlon paymenLs,
lncreased vlcLlm cooperaLlon wlLh Lhe [usLlce sysLem,
lncreased cooperaLlon of Lhe publlc wlLh Lhe
crlmlnal [usLlce sysLem,
fewer people who belleve Lhe [usLlce sysLem ls a
revolvlng" door, and
lncreases ln Lhe number of poslLlve medla reporLs
abouL Lhe [usLlce sysLem.
Impr!vements in the s!cial and fiscal c!sts !f
justice system interventi!ns, as measured by
fewer famlly members of known offenders who
become lnvolved wlLh Lhe [usLlce sysLem,
decreases ln Lhe cosLs for lncarceraLlon,
greaLer flnanclal reLurn on lnvesLmenL ln LreaLmenL,
rehablllLaLlon, and alLernaLlves Lo lncarceraLlon,
decreased crlme raLe,
lncreased Lax base,
lncreases ln Llmely chlld supporL paymenLs, and
lncreases ln courL-lmposed fees collecLed.
A harm reducLlon phllosophy focuses more broadly on
Lhe overall and long-Lerm healLh and welfare of Lhe
communlLy, parLlcularly ln Lerms of creaLlng a
collecLlve sense of publlc safeLy.
74% !f resp!ndents agree with the
statement, We sh!uld increase
spending !n appr!aches pr!ven t!
reduce the chances that !ffenders
will c!mmit new crime.
Z!gby Internati!nal, August 2009
90% !f resp!ndents indicate that
the criminal justice system
sh!uld w!rk t! increase the
publics c!nfidence.
Z!gby Internati!nal, August 2009
23
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
SECTION 4: THE PRINCIPLES UNDERLYING THE
FRAMEWORK
lour prlnclples, each based upon emplrlcal research, underlle Aoltomewotkofot vlJeoce-
8oseJoueclsloooMokloqoloolocoloctlmlooloIostlceo5ystems. 1hey deflne, ln broad Lerms, Lhe
way crlmlnal [usLlce professlonals wlll work LogeLher, make declslons, and operaLe Lhelr
agencles under Lhls approach.
PRINCIPLE ONE: THE PROFESSIONAL JUDGMENT OF CRIMINAL
JUSTICE SYSTEM DECISION MAKERS IS ENHANCED WHEN
INFORMED BY EVIDENCE-BASED KNOWLEDGE
39
uecades of research ln Lhe [usLlce and behavloral healLh flelds have resulLed ln emplrlcal
flndlngs LhaL supporL pracLlces and lnLervenLlons LhaL resulL ln crlme reducLlon. Lnhanced
awareness and Lhe conslsLenL appllcaLlon of LhaL knowledge LhroughouL Lhe crlmlnal [usLlce
sysLem offer Lhe promlse of decreased preLrlal mlsconducL and posL-senLence crlme and
communlLy harm. 1he crlmlnal [usLlce sysLem's dlscreLlon polnLs provlde for Lhe use of
professlonal [udgmenL Lo ensure LhaL lndlvldual facLors and Lhe LoLallLy of clrcumsLances
are Laken lnLo conslderaLlon when declslons are made.
Implicati!ns !f Principle One
lor professlonal [udgmenL Lo be lnformed by evldence-based knowledge
evldence-based knowledge musL be
documenLed and readlly avallable,
Lhe pollcy lmpllcaLlons of knowledge-and Lhelr
poLenLlal ouLcomes-musL be ldenLlfled,
Lhe meLhods for applylng knowledge Lo pracLlce
musL be dellneaLed,
professlonal [udgmenL should Lake lnLo accounL
boLh evldence-based knowledge and lndlvldual
clrcumsLances, and
where declslons are made LhaL counLer
emplrlcal evldence, Lhe raLlonale for Lhose
excepLlons should be explalned.

39
See the f!ll!wing research citati!ns that supp!rt this principle: Andrews & B!nta, 1998; A!s, Miller, & Drake, 2006a; Cullen
& Gendreau, 2000; Gendreau, G!ggin, Cullen, & Andrews, 2001; Gendreau, Little, & G!ggin, 1996; Gr!ve & Meehl, 1996;
Gr!ve, Zald, Leb!w, Snitz, & Nels!n, 2000; L!wenkamp, Latessa, & H!lsinger, 2006; L!wenkamp, Latessa, & Smith, 2006;
L!wenkamp, Pealer, Smith, & Latessa, 2007.
61% !f resp!ndents indicate that
when criminal justice pr!fessi!nals
make decisi!ns, research !n what
w!rks in preventing crime sh!uld be
the m!st imp!rtant thing they rely
!n. 24% say pr!fessi!nal experience
and 9% say pers!nal beliefs sh!uld be
the maj!r determinant.
Z!gby Internati!nal, August 2009
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
26
PRINCIPLE TWO: EVERY INTERACTION WITHIN THE CRIMINAL
JUSTICE SYSTEM OFFERS AN OPPORTUNITY TO CONTRIBUTE TO
HARM REDUCTION
40
Cffenders lnLeracL wlLh an array of professlonals (e.g., law enforcemenL offlcers, preLrlal
offlclals, [allers, [udges, eLc.) as Lhelr cases are processed Lhrough Lhe crlmlnal [usLlce sysLem.
Llkewlse, an array of professlonals-and Lhe agencles Lhey represenL-lnLeracL wlLh ooeo
oootbet (e.g., law enforcemenL wlLh prosecuLors, prosecuLors wlLh defenders, [udges wlLh
preLrlal offlclals, eLc.). 1hree separaLe buL equally lmporLanL bodles of research are relevanL
Lo Lhese crlmlnal [usLlce sysLem condlLlons. llrsL, research demonsLraLes LhaL professlonals'
lnLeracLlons wlLh offenders can have a slgnlflcanL poslLlve lmpacL on offenders' behavlor.
Second, parallel research demonsLraLes LhaL professlonals' poslLlve lnLeracLlons wlLh vlcLlms
can promoLe a sense of saLlsfacLlon and falrness. 1hlrd, research demonsLraLes LhaL sysLems
are mosL effecLlve ln achlevlng Lhelr ulLlmaLe ouLcomes when Lhey operaLe as value chalns."
under a value chaln sysLem, each componenL of a sysLem provldes addlLlve raLher Lhan
dupllcaLlve or deLracLlng value. lor Lhls Lo be Lrue, Lhe componenLs' lnLeracLlve operaLlons
musL be fully coordlnaLed wlLh one anoLher.
Implicati!ns !f Principle Tw!
lor Lhe crlmlnal [usLlce sysLem Lo Lake advanLage of lLs lnLeracLlon poLenLlal
all professlonals ln Lhe crlmlnal [usLlce sysLem musL undersLand Lhelr lndlvldual
poLenLlal Lo poslLlvely lnfluence offender behavlor,
all professlonals ln Lhe crlmlnal [usLlce sysLem musL undersLand Lhelr lndlvldual
poLenLlal Lo poslLlvely lnfluence vlcLlms' experlences wlLh Lhe [usLlce sysLem,
crlmlnal [usLlce professlonals musL have Lhe knowledge and skllls LhaL wlll enable Lhem
Lo maxlmlze Lhese opporLunlLles,
agency
41
pollcles LhroughouL Lhe crlmlnal [usLlce sysLem musL enable professlonals Lo
exerclse Lhls knowledge and apply Lhese skllls,
crlmlnal [usLlce sysLem processes musL be evaluaLed Lo ensure LhaL lnLerchanglng
sysLems are coordlnaLed and allgned wlLh one anoLher (l.e., lnformaLlon ls shared,
pollcles are compaLlble, lnLeresLs and ouLcomes are ln agreemenL), and
where lnLerchanglng sysLems lack coordlnaLlon, processes musL be reallgned.

40
See the f!ll!wing research citati!ns that supp!rt this principle: Bazem!re & Schiff, 2004; B!nta, Rugge, Sc!tt, B!urg!n, &
Yessine, 2008; D!wden & Andrews, 2004; Henggeler, Sch!enwald, B!rduin, R!wland, & Cunningham, 1998; Lind & Tyler,
1988; MacDuffie & Helper, 2006; P!rter, 1985; Tyler, 2007; Tyler, 2000; Tyler & Hu!, 2002; Umbreit, 1998; WSIPP, 2004.
41
Thr!ugh!ut this d!cument we use the term agency t! indicate a discrete entity !rganized t! serve a particular functi!n, such
as a p!lice agency, pr!secut!rs !ffice, c!urt, etc.
27
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
PRINCIPLE THREE: SYSTEMS ACHIEVE BETTER OUTCOMES
WHEN THEY OPERATE COLLABORATIVELY
42
8esearch demonsLraLes LhaL speclflc acLlvlLles, processes, and approaches-when lnsLlLuLed
and adhered Lo across componenLs-wlll more llkely resulL ln Lhe achlevemenL of arLlculaLed
ouLcomes. As dlsLlngulshed from value chaln research, whlch addresses Lhe lmporLance of
Lhe lnLeracLlons of sobsystemso(componenLs of a larger whole), Lhe research on collaboraLlon
speaks Lo Lhe manner ln whlch Lhe loJlvlJools who represenL dlfferenL lnLeresLs and
organlzaLlons (e.g., courL admlnlsLraLlon, [all operaLlons, eLc.) work LogeLher Lowards a
shared ouLcome (decreased crlme and harm, lncreased communlLy safeLy).
Implicati!ns !f Principle Three
lor crlmlnal [usLlce leadershlp Lo
achleve effecLlve collaboraLlon
key declslon makers and
sLakeholders musL be ldenLlfled,
a formal, ongolng process of
collaboraLlve pollcymaklng musL
be esLabllshed,
parLners musL ensure LhaL
collaboraLlon occurs aL Lhe sysLem
and case level only lnasmuch as lL
does noL lnfrlnge upon Lhe
lndlvldual rlghLs of Lhe accused or
Lhe responslblllLles and auLhorlLy
of Lhe sysLem acLors, and
pollcy Leams musL esLabllsh and
adhere Lo emplrlcally derlved
collaboraLlon meLhods LhaL have
been demonsLraLed Lo be
successful ln faclllLaLlng goal
aLLalnmenL.
43

42
See the f!ll!wing research citati!ns that supp!rt this principle: Adler, Kw!n, & Heckscher, 2008; C!llins & P!rras, 1997;
Heckscher & Adler, 2006; Henggeler, Sch!enwald, B!rduin, R!wland, & Cunningham, 1998; Lars!n & LaFast!, 1989.
43
A b!dy !f literature !n successful c!llab!rative pr!cesses exists and sh!uld guide this w!rk. As addressed in Secti!n 6,
supp!rting d!cuments will describe these research findings and translate findings int! specific steps c!llab!rative teams can
f!ll!w.
While ethical questi!ns regarding the
participati!n !f judges !n c!llab!rative
teams have arisen in a number !f
circumstances, judges acr!ss the
c!untry have led !r participated !n
teams that have addressed jail
cr!wding, established specialty c!urts,
revised p!licy and practice related t! the
management !f a particular !ffender
p!pulati!n, !r !therwise led t!
impr!vements in c!urt and justice
system !perati!ns. The ABA M!del C!de
!f Judicial C!nduct and the maj!rity !f
state judicial rules !f ethics supp!rt the
participati!n !f judges in c!mmissi!ns
!r p!licy-level gr!ups that are dev!ted
t! the impr!vement !f the law, the legal
system, !r the administrati!n !f justice.
F!r a m!re in-depth discussi!n !f the
ethical c!nduct !f judges !n
c!llab!rative teams, see Str!ker, 2006
and Gray, 2002.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
28
PRINCIPLE FOUR: THE CRIMINAL JUSTICE SYSTEM WILL
CONTINUALLY LEARN AND IMPROVE WHEN PROFESSIONALS
MAKE DECISIONS BASED ON THE COLLECTION, ANALYSIS, AND
USE OF DATA AND INFORMATION
44
Learnlng sysLems are Lhose LhaL adapL Lo a dynamlc envlronmenL Lhrough a process of
conLlnuous lnformaLlon collecLlon and analysls. 1hrough Lhls process of lndlvldual and collecLlve
learnlng, enLlLles-wheLher a slngle professlonal worklng wlLh an lndlvldual case, an agency
monlLorlng lLs overall operaLlons, or Lhe crlmlnal [usLlce sysLem as a whole monlLorlng sysLem
efflclency and effecLlveness-lmprove Lhelr processes and acLlvlLles ln a consLanL efforL Lo
achleve beLLer resulLs aL all levels. ln addlLlon Lo faclllLaLlng conLlnuous lmprovemenLs ln harm
reducLlon wlLhln an agency or sysLem, ongolng daLa collecLlon adds Lo Lhe overall body of
knowledge ln Lhe fleld abouL whaL works and whaL does noL.
Implicati!ns !f Principle F!ur
lor Lhe crlmlnal [usLlce sysLem Lo become a learnlng enLlLy, Lhe followlng ls necessaryX
Lhe esLabllshmenL of clear, speclflc, and LransparenL performance measuremenLs LhaL
ldenLlfy and measure approaches and acLlvlLles demonsLraLed or belleved
43
Lo conLrlbuLe
Lo deslred ouLcomes aL Lhe case, agency, and sysLem levels,
Lhe esLabllshmenL of basellne measures aL Lhe case, agency, and sysLem levels,
ongolng and ob[ecLlve collecLlon of daLa aL Lhe case, agency, and sysLem levels,
crlLlcal and ob[ecLlve analysls of Lhese daLa Lo compare agency and sysLem performance
wlLh esLabllshed LargeLs,
commlLmenL Lo quallLy assurance ln Lhe performance of acLlvlLles and ln Lhe collecLlon of
meanlngful daLa,
conLlnual feedback loops Lo ensure LhaL lnformaLlon ls shared, muLually undersLood, and
collaboraLlvely dellberaLed,
commlLmenL Lo vlew less-Lhan-deslrable resulLs
as opporLunlLles Lo lmprove, and
modlflcaLlon of pollcy and pracLlce as
performance measures and quallLy conLrol
monlLorlng lndlcaLe.

44
See the f!ll!wing research citati!ns which supp!rt this principle: Peters & Austin, 1986; Peters & Waterman, 2004; Senge,
2006.
45
Where the evidence falls sh!rt !r is inc!mplete, data c!llecti!n and critical analysis are particularly imp!rtant.
89% !f resp!ndents indicate that
criminal justice !fficials sh!uld tell
the public h!w well they are d!ing at
reducing crime.
Z!gby Internati!nal, August 2009
29
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Less!ns in Using Evidence...Fr!m M!neyball
In the b!!k M!neyball: The Art !f Winning an Unfair Game (2003), Michael Lewis
examined the questi!n !f h!w the Oakland Athletics, the sec!nd p!!rest team in maj!r
league baseball, repeatedly excelled against better-financed teams. Unable t! match the
financial strength !f perennial fav!rites such as the New Y!rk Yankees, the Oakland
Athletics used an!ther strategy t! achieve c!nsistently high perf!rmance: they
used evidence.
Oakland Athletics General Manager Billy Beane challenged baseballs c!nventi!nal
wisd!m ar!und c!mm!n decisi!ns such as the advantage !f drafting p!wer hitters and
when t! bunt. By using statistics and !ther evidence, Beane determined, f!r example, that
a walk is n!t an inferi!r way t! get !n base; it is in fact as g!!d as a single. With this
c!nclusi!n, Beane set !ut t! recruit n!t the p!wer hitters, but th!se with the best walk-t!-
at-bat statistics. In this way, players were recruited based !n their !verall value-
add t! the team.
Applying this type !f analysis t! every aspect !f baseball, Beane established a meth!d !f
decisi!n making that relied !n data and inf!rmati!n t! supp!rt the c!st-benefit decisi!ns
that w!uld lead t! a higher perf!rming team, dem!nstrating that it matters less h!w much
m!ney is spent and m!re h!w it is spent.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
30
SECTION 5: APPLYING EVIDENCE-BASED
PRINCIPLES TO PRACTICE
A LOGIC MODEL FOR HARM REDUCTION
DECISION MAKING AT THE SYSTEM LEVEL
A loglc model ls a graphlc represenLaLlon of Lhe Lheory behlnd a concepLual framework. 1he
loglc model supporLlng Aoltomewotkofot vlJeoce-8oseJoueclsloooMokloqoloolocoloctlmloolo
Iostlceo5ystems addresses Lhe lmplemenLaLlon of Lhe lramework aL Lhe system level. lL ls
bullL upon Lhe four prlnclples underlylng Lhe lramework (as descrlbed ln SecLlon 4). lL
ouLllnes Lhe loglcal flow of boLh Lhe processes and acLlvlLles lnvolved ln lmplemenLaLlon,
and lL demonsLraLes Lhe expecLed harm reducLlon lmpacLs LhaL wlll resulL from Lhese
processes and acLlvlLles.
Loglc models are bullL uslng several key elemenLsX
lnpuLs, whlch represenL exlsLlng res!urces (boLh flnanclal and human), pollcles, pracLlces,
faclllLles, and capablllLles LhaL [urlsdlcLlons brlng Lo Lhe Lable ln lmplemenLlng Lhe
lramework,
acLlvlLles, whlch represenL Lhe speclflc strategies Lo be puL ln place Lo lmplemenL Lhe
lramework and apply evldence-based declslon maklng Lo achleve harm reducLlon,
ouLpuLs, whlch speclfy Lhe immediate results LhaL occur as acLlvlLles and sLraLegles are
lmplemenLed (e.g., change ln pollcy/pracLlce, adopLlon of new Lools/proLocols, number of
people Lralned, eLc.),
ouLcomes, whlch serve as indicat!rs LhaL change ls occurrlng aL key declslon polnLs ln Lhe
[usLlce sysLem as a resulL of Lhese acLlvlLles and whlch demonsLraLe LhaL evldence-based
declslon maklng has been lmplemenLed, and
lmpacLs, whlch deflne Lhe Lypes of l!ng-term results LhaL can be anLlclpaLed and
measured as a resulL of Lhe lramework's lmplemenLaLlon.
underlylng Lhe enLlre loglc model are assumpLlons and conLexLual condlLlons. 1he assumpLlons
are based on Lhe prlnclples ln Lhe lramework and serve as Lhe raLlonale for how [urlsdlcLlons
can achleve harm reducLlon by lmplemenLlng Lhls lramework. 8ecause Lhe loglc model ls
deslgned Lo serve as a roadmap, each [urlsdlcLlon wlll Lallor speclflc aspecLs of Lhe acLlvlLles and
Lypes of ouLcomes/lmpacLs expecLed based on local clrcumsLances. 1hese local clrcumsLances,
referred Lo as conLexLual condlLlons, underlle Lhe enLlre loglc model.
EVIDENCE-BASED DECISION MAKING SYSTEM-LEVEL
IMPLEMENTATION STEPS
lmplemenLaLlon of evldence-based declslon maklng requlres a deslre and commlLmenL Lo
change how Lhe crlmlnal [usLlce sysLem responds Lo crlmlnal offendlng ln a way LhaL enhances
publlc safeLy and reduces harm Lo communlLles, vlcLlms, and offenders. Such change
necessarlly lnvolves a complex seL of lmplemenLaLlon sLeps LhaL need Lo occur aL mulLlple levels
wlLhln Lhe sysLem-aL Lhe overall sysLem level (l.e., lnvolvlng all sLakeholders wlLhln Lhe [usLlce
sysLem), wlLhln each agency/enLlLy LhaL engages ln Lhe crlmlnal [usLlce process (e.g., pollce,
prosecuLors, defense, preLrlal servlces, courLs, communlLy correcLlons, and correcLlons), and aL
31
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Lhe case level (e.g., ln Lerms of how declslons are made ln lndlvldual cases). 1he lramework
provldes an overall vlslon for how evldence-based declslon maklng can work ln local crlmlnal
[usLlce sysLems and Lhe Lypes of
ouLcomes and lmpacLs LhaL mlghL be
expecLed lf evldence-based declslon
maklng ls lmplemenLed.
ln general, Lhe lmplemenLaLlon process
lncludes four sLagesX1) developlng a
sysLemwlde vlslon and process for
evldence-based declslon maklng,
2) developlng a plan Lo lmplemenL Lhe
pollcy and procedural changes necessary
Lo supporL Lhe lmplemenLaLlon of
evldence-based declslon maklng, 3)
lmplemenLlng evldence-based declslon
maklng, and 4) lnsLlLuLlonallzlng and
reflnlng evldence-based declslon maklng
Lhrough an ongolng process of revlew
and reflnemenL.
A sysLem-level loglc model LhaL deplcLs
Lhe relaLlonshlps beLween acLlvlLles,
ouLcomes, and lmpacLs ls provlded Lo
show generally how lmplemenLaLlon of
evldence-based declslon maklng can
change Lhe sysLem's response Lo crlmlnal offendlng, enhance publlc safeLy, and reduce harm.
46
IHI LESSON #3: A MARATHON IS RUN ONE STEP AT A TIME
IHI ann!unced fr!m the start that n!t every Campaign participant had t!
implement all six interventi!ns at !nce. Rec!gnizing that small wins w!uld
unleash an appetite f!r larger vict!ries, their m!tt! became !ne step at a time.
This appr!ach res!lved the pr!blem !f implementing change acr!ss a very large
and diverse nati!n: what was p!ssible in an urban research facility in
Massachusetts, f!r example, might n!t be practical f!r a small, rural h!spital in
Minnes!ta. Yet each had the !pp!rtunity t! succeed, !ne step at a time. The
less!n f!r criminal justice?
PROVIDE THE TOOLS TO WIN THE RACE,
LET THE RUNNERS SET THEIR PACE.

46
The specifics !f the implementati!n steps will be the subject !f an interactive dial!gue with each !f the pil!t sites and a
jurisdicti!n-specific l!gic m!del will be devel!ped. In additi!n, during Phase II, agency-specific and case-level l!gic m!dels will
als! be devel!ped.
Results-Based Management
What gets measured gets d!ne.
If results are n!t measured, successes
cann!t be distinguished fr!m failures.
If successes cann!t be distinguished,
they cann!t be replicated.
If failures cann!t be identified, they
cann!t be c!rrected.
If results cann!t be dem!nstrated,
supp!rt cann!t be secured.
Adapted fr!m Osb!rne & Gaebler, 1992.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
32
EVIDENCE-BASED DECISION MAKING LOGIC MODEL
33
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
SECTION 6: KEY CHALLENGES IN IMPLEMENTING
THIS FRAMEWORK
WlLhouL a doubL, lmplemenLaLlon of Lhls lramework wlll ralse a number of challenges and
Lhorny" lssues for crlmlnal [usLlce sysLem declslon makers. Some of Lhese are pragmaLlc, some
operaLlonal, oLhers phllosophlcal. Whlle we do noL aLLempL Lo ldenLlfy all of Lhese, a few key
lssues are noLed as among Lhe mosL complex. Pow Lhey are addressed and puL lnLo pracLlce wlll
be among Lhe lmporLanL lmplemenLaLlon lessons of hase ll.
RISK REDUCTION AND EVIDENCE-BASED DECISION MAKING IN THE
CONTE"T OF SANCTIONING PURPOSE
Different Cases, Different Purp!ses
Much has been wrlLLen abouL Lhe purposes of senLenclng. Lach ([usL desserLs/reLrlbuLlon,
deLerrence, rehablllLaLlon, and lncapaclLaLlon) offers a raLlonale for sancLlonlng offenders. 1he
mosL noLable of Lhe dlfferences among Lhem ls Lhe dlsLlncLlon beLween uLlllLarlan goals-Lhose
LhaL alm Lo produce some good as a resulL of Lhe sancLlon (such as dlscouraglng crlmlnal
behavlor, helplng offenders learn Lo avold fuLure crlmlnal engagemenL, or resLralnlng Lhose
LhoughL llkely Lo pose a LhreaL ln Lhe fuLure)-and Lhe non-uLlllLarlan [usL deserLs" approach
whlch asserLs LhaL offenders deserve Lo be punlshed for Lhelr crlmes, regardless of wheLher or
noL LhaL punlshmenL wlll lnfluence fuLure behavlor.
47
Cnly some of Lhese aLLend Lo Lhe lssue of
rlsk reducLlon.
1he unlque facLors and conslderaLlons of a glven case may resulL ln one sancLlonlng purpose
Laklng precedence over anoLher. ln Lhose lnsLances where rlsk reducLlon ls noL ldenLlfled as Lhe
prlmary purpose of sancLlonlng, lLs slgnlflcance and lmporLanL role should noneLheless be
fully consldered.
The Weight !f the Evidence
As descrlbed prevlously, Lhere ls a wlde body of research Lo supporL Lhe clalm LhaL rlsk
reducLlon ls posslble.
48
1he evldence regardlng oLher sancLlonlng purposes ls, Lhus far, less
compelllng.
49
Lvldence-based declslon maklng requlres LhaL declslon makers undersLand Lhe
relaLlve lmpacL of varlous sancLlonlng opLlons and Lake Lhls lnLo accounL when deLermlnlng Lhe
ouLcome of a parLlcular case.
nlneLy-flve percenL of offenders wlll ulLlmaLely be released Lo Lhe communlLy.
30
1he welghL of
Lhe evldence demonsLraLlng Lhe efflcacy of rlsk reducLlon approaches provldes [usLlce sysLem

47
See particularly the writings !f M. Kay Harris !n the t!pic !f sancti!ning phil!s!phies (e.g., Harris, 1986).
48
See What W!rks in Reducing Pretrial Misbehavi!r and Offender Recidivism in Appendix 3 and 7 Ways t! Reduce
Recidivism in the Preamble.
49
F!r instance, research finds that incarcerati!n and !ther punitive sancti!ns, in is!lati!n !f !ther interventi!ns, d! n!t reduce
future !ffending; see Gendreau & G!ggin, 1996; Gendreau, G!ggin, & Cullen, 1999; Lipsey & Cullen, 2007; Smith, G!ggin, &
Gendreau, 2002. Research als! suggests that deterrent effects are inc!nsistent and depend !n individual characteristics,
em!ti!ns, experiences, etc. Theref!re, !utc!mes derived s!lely fr!m deterrence are difficult t! predict; see B!uffard et al.,
1999; Exum, 2002; Matsueda, Kreager, & Huizinga, 2006; Nagin, 1998; Piquer! & P!garsky, 2002; P!garsky, 2007; P!garsky,
2002; Staff!rd & Warr, 1993.
50
See Hughes & Wils!n, 2003.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
34
erhaps no oLher [usLlce sysLem
process has as profound an effecL
on harm reducLlon as plea
negoLlaLlons. 1o be successful ln
reachlng Lhe goal of publlc safeLy,
plea negoLlaLlon pracLlces should
be gulded by research.
acLors wlLh confldence LhaL Lhe goal of rlsk reducLlon can be achleved
31
elLher slngularly or ln
con[uncLlon wlLh oLher sancLlonlng purposes. ln Lhls way, rlsk reducLlon should be noL
sldellned" when oLher sancLlonlng goals are consldered Lo be of equal or hlgher value. lL ls noL
an elLher/or" proposlLlon, alLhough bow rlsk conslderaLlons are facLored lnLo a case may vary.
Several case scenarlos may besL lllusLraLe Lhls polnLX
A low rlsk offender who has commlLLed a serlous crlme mlghL be senLenced Lo serve hls
Llme ln [all raLher Lhan prlson lf lL ls deLermlned LhaL [all would be less llkely Lo expose Lhe
offender Lo Lhe anLlsoclal lnfluences LhaL lead Lo lncreases ln crlme among lower rlsk
offenders.
A moderaLe rlsk offender senLenced Lo prlson mlghL be placed ln an lnsLlLuLlon closer Lo home,
where supporLlve famlly members have a greaLer opporLunlLy Lo offer poslLlve lnfluence. Pe
may also be provlded rlsk reduclng programmlng durlng and followlng lncarceraLlon.
A hlgh rlsk offender convlcLed of a low level offense mlghL be placed on lnLenslve
supervlslon and be requlred Lo compleLe a hlgh lnLenslLy LreaLmenL program.
ln each of Lhese scenarlos, rlsk reducLlon ls a conslderaLlon ln Lhe crafLlng of an approprlaLe
dlsposlLlon, ln some cases olooqslJe oLher sancLlonlng purposes.
ln hase ll of Lhe Lvldence-8ased ueclslon Maklng ln Local Crlmlnal !usLlce SysLems lnlLlaLlve, nlC
and lLs lnlLlaLlve parLners wlll work wlLh plloL slLes Lo sorL Lhrough and develop a model(s) Lo
address Lhe quesLlon of Lhe role of rlsk reducLlon ln Lhe conLexL of mulLlple sancLlonlng purposes.
PLEA NEGOTIATIONS
ln mosL [urlsdlcLlons, well over 90 of felony crlmlnal cases are handled Lhrough pleas, wlLh Lhe
ma[orlLy of Lhe courLs accepLlng Lhose pleas as negoLlaLed.
32
ln many [urlsdlcLlons, plea
negoLlaLlons are ofLen crafLed ln hlghly prescrlpLlve ways, dlcLaLlng, for example, noL only Lhe
lengLh of lncarceraLlon and probaLlon supervlslon, buL also Lhe speclflc condlLlons of
supervlslon. ?eL, few [urlsdlcLlons have avallable Lo Lhem lnformaLlon abouL an offender's rlsk
Lo reoffend or crlmlnogenlc needs aL Lhe polnL of plea negoLlaLlon, meanlng LhaL key declslon
makers-prosecuLors and defenders-negoLlaLe Lhese agreemenLs absenL lnformaLlon abouL
how besL Lo lnfluence fuLure crlmlnal behavlor based on Lhe unlque characLerlsLlcs of Lhe
offender belng senLenced. As a resulL, ln mosL [urlsdlcLlons, cases are passed along Lo
correcLlons and/or probaLlon, whlch Lhen assess
rlsk/needs and, ln many cases, work Lo reLroflL
research-based lnLervenLlons Lo courL-lmposed
senLenclng parameLers.
Arguably, Lhe lnLroducLlon of rlsk/need lnformaLlon
aL Lhe plea sLage-and perhaps earller-could have
a profound effecL on [udlclal declslons, and yeL Lhls
ls noL wlLhouL lLs due process and resource
challenges. 1hls ls anoLher of Lhe lmporLanL lssues
Lhe lnlLlaLlve wlll address ln hase ll.

51
F!r a review !f s!me !f the research, see Appendix 3.
52
See BJS, 2009; BJS, 2007. While misdemean!r cases !utweigh fel!nies 4 t! 1 (LaF!untain et al., 2008), n! nati!nal data is
available t! indicate the percentage !f these cases that are settled thr!ugh plea agreement.
33
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
THE MANAGEMENT OF LOW LEVEL OFFENSES
Many [usLlce sysLems across Lhe counLry are lnundaLed wlLh mlnor crlmlnal maLLers. 1hese
peLLy or nulsance" crlmes, as Lhey are ofLen called, consume enormous sysLem resources,
lncludlng pollce offlcer Llme, preLrlal assessmenLs and perhaps preLrlal supervlslon, hearlngs
before ball commlssloners and maglsLraLes, [all beds, courL dockeLs, eLc. CfLen, Lhe defendanLs
charged wlLh Lhese crlmes are lndlgenL, menLally lll, and/or homeless, many are revolvlng
door" cases, lndlvlduals who are apprehended and processed numerous Llmes over Lhe course
of a year. LfforLs Lo process and manage Lhelr cases consume a slgnlflcanL porLlon of Lhe [usLlce
sysLem budgeL. lnsufflclenL fundlng or servlces and/or Lhe press of overwhelmlngly hlgh
caseloads can resulL ln qulck-flx responses LhaL may address Lhe lmmedlaLe, presslng problem
of movlng Lhe case forward wlLhln esLabllshed Llmeframes, buL Loo ofLen fall shorL of resolvlng
Lhe sysLemlc lnfluences LhaL lle aL Lhe hearL of Lhe crlmlnal behavlor.
Crlmlnal [usLlce enLlLles and agencles across Lhe counLry process hundreds or even Lhousands of
Lhese cases ln a glven day or week,
33
ofLenLlmes wlLhouL Lhe opporLunlLy Lo dlagnose Lhe
facLors leadlng Lo Lhe crlmlnal behavlor or consLrucL a soluLlon wlLh long-Lerm poLenLlal.
AssessmenLs are rarely conducLed ln Lhese cases, resulLlng ln a slLuaLlon ln whlch llLLle
lnformaLlon oLher Lhan a crlmlnal hlsLory and arresL reporL are avallable Lo gulde
declslon maklng.
1he lramework seeks Lo apply evldence-based knowledge Lo all crlmlnal [usLlce declslons and ln
all Lypes of cases-peLLy, serlous, and all Lhose ln beLween. 1here are aL leasL Lwo challenges ln
dolng soXflrsL, Lhere ls a dearLh of knowledge Lo gulde pollcy and pracLlce ln some areas.
Second, Lhe volume of cases, shorLage of labor, press of Llme, dlfflculLles assoclaLed wlLh
unlque challenges such as Lhe serlously menLally lll, and, ln some cases, lnsufflclenL physlcal
space Lo conducL lnLervlews, provlde servlces, eLc., comblne Lo creaLe seemlngly lmposslble
barrlers Lo evldence-based pracLlces wlLh all cases. 1hls ls yeL a Lhlrd key lmplemenLaLlon
challenge LhaL wlll be squarely addressed ln hase ll of Lhe lnlLlaLlve.
LOOKING TO PHASE II AND BEYOND
lmplemenLaLlon of Lhe lramework wlll undoubLedly surface a varleLy of Lhorny lssues" such as
Lhose addressed hereXrlsk reducLlon as opposed Lo or alongslde oLher sancLlonlng purposes,
wheLher and how rlsk and crlmlnogenlc need daLa should be consldered aL early declslon polnLs
(LhaL ls, aL Lhe arresL, preLrlal, and plea negoLlaLlon sLages), and how besL Lo effecLlvely and
efflclenLly use research Lo end Lhe revolvlng door of low level crlmlnal cases. CLhers are
anLlclpaLed. And whlle lL ls expecLed LhaL Lhe dlscusslons and debaLes abouL how Lo address
Lhem wlll be dlfflculL-and wlll ralse quesLlons LhaL compel pollcymakers Lo confronL dlrecLly
Lhelr phllosophles, values, commlLmenL Lo pasL pracLlce, and ablllLles Lo creaLlvely deslgn new
[usLlce sysLem approaches-Lhere ls no doubL LhaL Lhese dellberaLlons wlll move Lhe fleld
forward ln Lhe advancemenL of evldence-based declslon maklng and lmproved [usLlce sysLem
ouLcomes. 1o be sure, one of Lhe key sLraLegles Lo maklng Lhls posslble ls collaboraLlve
pollcymaklng.

53
The actual numbers vary widely by jurisdicti!n and in s!me jurisdicti!ns will be very l!w. See LaF!untain et al., 2008;
Nati!nal Ass!ciati!n !f Criminal Defense Lawyers, 2009.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
36
SECTION 7: COLLABORATION: A KEY INGREDIENT
OF AN EVIDENCE-BASED SYSTEM
ALIGNING THE CRIMINAL JUSTICE SYSTEM TO ACHIEVE
HARM REDUCTION
ComponenLs of Lhe crlmlnal [usLlce sysLem-and Lhe agencles and acLors LhaL represenL Lhem-
frequenLly operaLe wlLhouL clarlLy of, or consensus on, Lhe ouLcomes Lhe sysLem seeks Lo
achleve and/or Lhe opLlmal meLhods Lo reach Lhem.
new ways of Lhlnklng abouL how Lhls sysLem" coolJ work, evldence-based knowledge abouL
how besL Lo produce lnLended ouLcomes aL Lhe sysLem, agency, and case levels, and emplrlcal
evldence abouL meLhods Lo achleve effecLlve collaboraLlve processes offer guldance Lo
[urlsdlcLlons lnLeresLed ln worklng collaboraLlvely Lo achleve harm reducLlon ln Lhelr
communlLles.
BRINGING THE STAKEHOLDERS TO THE TABLE TO FORM POLICY TEAMS
54
CollaboraLlon ln Lhe crlmlnal [usLlce sysLem seeks Lo overcome Lhe llmlLaLlons of LradlLlonal
and non-sysLemlc approaches Lo crlmlnal [usLlce problem solvlng and soluLlon developmenL by
brlnglng LogeLher sLakeholders Lo share lnformaLlon, work Loward Lhe developmenL of common
goals, and [olnLly creaLe pollcles Lo supporL Lhose goals. 5tokebolJets are deflned as Lhose who
lnfluence and have an lnvesLmenL ln Lhe crlmlnal [usLlce sysLem's ouLcomes. 1hese sysLemwlde
sLakeholder groups are referred Lo as pollcyoteoms.
ldeally, pollcy Leams are comprlsed of Lhe
crlmlnal [usLlce agencles and communlLy
organlzaLlons LhaL lmpacL, or are lmpacLed
by, declslons LhaL wlll be made by Lhe
collaboraLlve Leam. 1he speclflc
composlLlon of Lhe collaboraLlve Leam
varles from [urlsdlcLlon Lo [urlsdlcLlon.
1hose wlLh Lhe poslLlonal or personal
power Lo creaLe change wlLhln Lhelr own
agencles and organlzaLlons are approprlaLe
members of Lhe collaboraLlve Leam. 1he
chlef [udge, courL admlnlsLraLor, elecLed prosecuLor, chlef publlc defender and represenLaLlve
of Lhe prlvaLe defense bar, admlnlsLraLor of Lhe communlLy correcLlons agency, pollce chlef and
elecLed sherlff, preLrlal admlnlsLraLor, vlcLlm advocaLes, local elecLed offlclals (clLy manager,
counLy commlssloner), servlce provlders, and communlLy represenLaLlves all play a parL ln Lhe
admlnlsLraLlon of [usLlce and brlng valuable lnformaLlon, resources, and perspecLlves Lo Lhls
collaboraLlve endeavor.

54
Carl Lars!n and Frank LaFast! (1989) studied an array !f public and private sect!r w!rking gr!ups in an eff!rt t! identify the
characteristics !f highly effective teams. Their findings pr!vide a r!admap f!r jurisdicti!ns that seek t! w!rk t!gether in a truly
c!llab!rative manner. A c!mpani!n guide t! this Framew!rk will pr!vide inf!rmati!n and guidance t! jurisdicti!ns as they seek
t! establish p!licy teams t! undertake a c!llab!rative, evidence-based decisi!n making pr!cess.
C!llab!rati!n is the pr!cess !f w!rking
t!gether t! achieve a c!mm!n g!al that
is imp!ssible t! reach with!ut the eff!rts
!f !thers.
37
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
IHI LESSON #4: INDIVIDUALS ARE NOT FLAWED, SYSTEMS ARE
In its campaign t! save 100,000 lives, IHI refused t! view individual failure as the
way t! acc!unt f!r the needless l!ss !f 100,000 lives. Instead, they f!cused !n
c!rrecting the system !f medical care. In the w!rds !f Berwick, Every system is
perfectly designed t! achieve exactly the result it gets. IHI ad!pted the p!siti!n
that individual healthcare pr!fessi!nals did n!t need t! w!rk harder, smarter, !r
faster; instead, they needed t! change a flawed system that led smart and
dedicated pe!ple t! make mistakes. The less!n f!r criminal justice?
BUILD A SYSTEM THAT WILL NATURALLY RESULT
IN THE OUTCOMES WE SEEK.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
38
SECTION 8: BUILDING EVIDENCE-BASED AGENCIES
ALIGNING CRIMINAL JUSTICE AGENCIES TO ACHIEVE HARM
REDUCTION
lor evldence-based declslon maklng Lo be effecLlve, lL musL occur wlLh conslsLency LhroughouL
Lhe [usLlce sysLem. 1haL ls, Lhe rellance on evldence Lo lnform declslon maklng should occur aL
Lhe sysLem level, aL Lhe agency level, and aL Lhe case level.
1he precedlng secLlon on collaboraLlon suggesLs LhaL sysLem-level allgnmenL can besL be
achleved Lhrough a collaboraLlve pollcy Leam process. Agency- and case-level allgnmenL requlre
a dlfferenL approach, Lhey requlre a speclflc focus on organlzaLlonal developmenL wlLhln each
of Lhe [usLlce sysLem agencles.
AdopLlng a pracLlce of relylng on evldence Lo lnform declslon maklng-raLher Lhan relylng on
LradlLlon, personal bellefs, or oLher facLors-wlll undoubLedly requlre some (buL more llkely all)
agencles ln Lhe crlmlnal [usLlce sysLem Lo reevaluaLe Lhelr pollcles and pracLlces.
33
uolng so
lnvolves
reevaluaLlng agency mlsslon, goals, and values Lo supporL a vlslon LhaL ls shared by all Lhe
[usLlce sysLem sLakeholders as well as Lhe workforce wlLhln Lhe agency,
reconslderlng agency pollcy and pracLlce ln llghL of evldence-based knowledge,
ln some lnsLances, reLoollng organlzaLlonal sLrucLure,
addresslng, where necessary, organlzaLlonal culLure Lo allgn wlLh a new vlslon, mlsslon, and
goals, and
provldlng new knowledge and skllls for sLaff.
lor Lhese change efforLs Lo Lake hold, Lhey
musL prove Lhemselves Lo be rellable and Lo
beLLer supporL sLaffs' ablllLy Lo effecLlvely
carry ouL Lhelr duLles. lor example, lf aL Lhe
senLenclng sLage, ob[ecLlve daLa ls provlded
Lo defense counsel, prosecuLors, and [udges
LhaL effecLlvely lnforms and shapes Lhe
senLenclng declslon, declslon makers wlll
come Lo noL only expecL buL also Lo rely on
Lhls lnformaLlon ln Lhe fuLure. lf, on Lhe oLher
hand, Lhe lnformaLlon provlded ls nelLher useful nor rellable, Lhe new approach of conslderlng
ob[ecLlve daLa wlll be abandoned and pasL pracLlce wlll prevall.
CrganlzaLlonal change ls noL easy, nor ls lL always successful. Accordlng Lo experLs
36
up Lo 83 of organlzaLlonal change lnlLlaLlves fall, and
up Lo 70 of Lhese fallures are due Lo flawed execuLlon.

55
Appendix 3 is a c!mpilati!n !f evidence-based kn!wledge that has p!licy implicati!ns f!r justice system pr!fessi!nals.
56
R!gers, Wellins, & C!nn!r, 2002.
Organizati!nal devel!pment is the
practice !f changing internal systems,
and pe!ple, f!r the purp!ses !f visi!n
and missi!n advancement.
39
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
IHI LESSON #5: MAKE THE NEW EASIER THAN THE OLD
IHI underst!!d that if the practices they were pr!m!ting did n!t appeal t! th!se
wh! w!uld implement themif they were seen as n!thing m!re than additi!nal
w!rk burdenschange w!uld n!t !ccur. On the !ther hand, if the new practices
c!uld save staff time and eff!rt and enhance patient safety, staff w!uld be quick
t! embrace and integrate the new practices. M!dels f!r replacing f!rmer
practices with newer, streamlined appr!aches were ad!pted by inv!lving staff in
the pr!cess. Seni!r physician Steven Tremain, C!ntra C!sta Regi!nal Medical
Center, summarized the results: We basically exp!sed pe!ple wh! were hungry
t! learn h!w [t! achieve better results with!ut additi!nal burden]and they t!!k
it and ran with it. What [we]created is the belief that it can be d!ne. The less!n
f!r criminal justice?
REPLACE CURRENT PRACTICES WITH THOSE THAT
ARE MORE EFFECTIVE AND EASIER TO IMPLEMENT.
MAKING WHAT WORKS WORK
An enormous lnvesLmenL of publlc funds ls made each year ln Lhe name of publlc safeLy. 1he
sLraLeglc use of Lhose funds can produce a profoundly poslLlve lmpacL, as measured by fewer
new vlcLlms and fewer new crlmes commlLLed by offenders under crlmlnal [usLlce conLrol.
Powever, changlng pollcy and pracLlce aL Lhe sysLem, agency, or case level ls no slmple Lask,
parLlcularly when Lhese changes challenge currenL phllosophles, undersLandlngs of Lhe
research, and Lhe day-Lo-day pracLlce rouLlnes of agencles and sLaff. 1o reach Lhelr full
poLenLlal, evldence-based pracLlces cannoL slmply be placed alongslde pasL pracLlce or Lhrough
Lhe plecemeal exchange of one pasL pracLlce for a new one. lnsLead, an evidence-based
decisi!n making pr!cess-a sysLemlc approach LhaL uses research Lo lnform declslons aL all
levels-offers Lhe greaLesL promlse for recldlvlsm reducLlon and Lhe poLenLlal for a Lremendous
reLurnXone mllllon fewer vlcLlms.
4;G6X5;DXC7IX5;7J8DX;:X&6<=I=>=DGX&6I?<8=;7
uo. use rlsk assessmenL LoolsXLhey are Lhe gaLeway Lo rlsk reduclng sLraLegles.
uo. rovlde evldence-based programmlng LhaL LargeLs crlmlnogenlc needs for medlum and
hlgher rlsk offenders.
uo. Address crlmlnal Lhlnklng and problem solvlng skllls.
uo. 8espond Lo mlsconducL wlLh swlfLness and cerLalnLy.
uo. use more carroLs Lhan sLlcks.
uo. uellver servlces ln naLural (communlLy) envlronmenLs.
uoot.oLxpecL sancLlons alone Lo change behavlor.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
40
APPENDIX 1: ADVISORY BOARD MEMBERS:
EVIDENCE-BASED DECISION MAKING IN LOCAL
CRIMINAL JUSTICE SYSTEMS INITIATIVE
Shlrley Abrahamson, Chlef !usLlce, Supreme CourL of Wlsconsln, Madlson, Wlsconsln
Suzanne 8rown-Mc8rlde, LxecuLlve ulrecLor, Callfornla CoallLlon AgalnsL Sexual AssaulL,
SacramenLo, Callfornla
Ldwln 8urneLLe, Chlef ubllc uefender (former), Cook CounLy ubllc uefender's Cfflce,
Chlcago, llllnols
Cary ChrlsLensen, rlnclpal, CorrecLlons arLners, lnc., CllnLon Corners, new ?ork
Cary uarllng, Crlmlnal !usLlce lannlng Manager, Larlmer CounLy, lorL Colllns, Colorado
Adrlan Carcla, Parrls CounLy Sherlff, PousLon, 1exas
8oberL !ohnson, Anoka CounLy ALLorney, Anoka, MlnnesoLa
uale ooch, Senlor !udge, MulLnomah CounLy CourLhouse, MulLnomah CounLy ClrculL CourL,
orLland, Cregon
Sally oreamer, ulrecLor, llfLh !udlclal ulsLrlcL, ueparLmenL of CorrecLlonal Servlces, ues
Molnes, lowa
Mlchael Marcus, !udge, MulLnomah CounLy ClrculL CourL, orLland, Cregon
Carlos MarLlnez, ubllc uefender, Law Cfflces of Lhe ubllc uefender, Mlaml, llorlda
eggy McCarry, ulrecLor, CenLer on SenLenclng and CorrecLlons, vera lnsLlLuLe, new ?ork,
new ?ork
Ceraldlne nagy, ulrecLor, 1ravls CounLy CommunlLy Supervlslon and CorrecLlons
ueparLmenL, AusLln, 1exas
Wendy nlehaus, ulrecLor, ueparLmenL of reLrlal Servlces, PamllLon CounLy, ClnclnnaLl,
Chlo
Mlchael laneL, LxecuLlve Cfflcer, venLura CounLy Superlor CourL, venLura, Callfornla
8onald 8elnsLeln, ulrecLor, CenLer for Lvldence 8ased SenLenclng, Arlzona Supreme CourL,
hoenlx, Arlzona
Susan Shaffer, ulrecLor, ulsLrlcL of Columbla, reLrlal Servlces Agency, WashlngLon, u.C.
. uavld Soares, ulsLrlcL ALLorney, Albany !udlclal CenLer, Cfflce of Lhe ulsLrlcL ALLorney,
Albany, new ?ork
Mark 1hompson, !udlclal ulsLrlcL AdmlnlsLraLor, Pennepln CounLy ulsLrlcL CourL,
Mlnneapolls, MlnnesoLa
8oger Warren, resldenL LmerlLus, naLlonal CenLer for SLaLe CourLs, Wllllamsburg, vlrglnla
1homas WhlLe, ulrecLor of CperaLlons (former), CourL SupporL Servlces ulvlslon,
ConnecLlcuL !udlclal 8ranch, WeLhersfleld, ConnecLlcuL
41
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
APPENDIX 2: METHODOLOGY USED TO
COMPUTE 1 MILLION FEWER VICTIMS
1hls lramework was developed Lo asslsL crlmlnal [usLlce sysLem sLakeholders ln applylng
evldence Lo declslon maklng. Applylng evldence Lo declslon maklng can conLrlbuLe Lo
reducLlons ln Lhe raLe of recldlvlsm among offenders and ln collaLeral harm Lo communlLles. A
speclflc goal-fewer vlcLlms-has been ldenLlfled as a means Lo gauge success and galvanlze
sLakeholders around Lhls naLlonal lnlLlaLlve. 1he lnlLlaLlve has esLabllshed Lhe goal of one mllllon
fewer vlcLlms.
Accordlng Lo Lhe 8ureau of !usLlce SLaLlsLlcs Sourcebook of Crlmlnal !usLlce SLaLlsLlcs,
37
ln 2003
Lhere were 2,361,193 full-Llme employees worklng ln federal, sLaLe, and local crlmlnal [usLlce
sysLems. 1he llsLlng lncludes Lhose lnvolved ln correcLlons (748,230), [udlclal and legal poslLlons
(494,007), and pollce proLecLlon (1,118,936). lL does noL lnclude parL-Llme employees or Lhose
engaged ln worklng dlrecLly wlLh offenders ln programmlng (such as non-governmenLal,
conLracLual servlce provlders ln communlLy seLLlngs).
2.3 MILLION JUSTICE SYSTEM EMPLOYEES
2.3 mllllon [usLlce sysLem employees means LhaL every day, Lhere are 2.3 mllllon opporLunlLles
Lo reduce harm and Lhe llkellhood LhaL an offender wlll commlL anoLher crlme. lf [usL bolf of
Lhese lndlvlduals were Lo effecLlvely apply evldence-based pracLlces on [usL one case resulLlng
ln one less offender wlLh one less vlcLlm, Lhe neL effecL would be one mllllon fewer vlcLlms.
1hls lramework and lnlLlaLlve form Lhe basls of Lhe Cne Less ______" campalgn because every
lndlvldual who works ln Lhe [usLlce sysLem coo make a dlfference. lL ls noLhlng less Lhan a call
Lo acLlon.
One less !ffender.
One less crime.
One less victim.

57
BJS, 2003.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
42
APPENDIX 3: RESEARCH FINDINGS MATRIX
1he research ln Lhls maLrlx ls a snapshoL, raLher Lhan a Lhorough revlew, of all currenL research
on reduclng preLrlal mlsbehavlor and offender recldlvlsm. 1he summarles provlded here are
lnLended Lo brlefly descrlbe Lhe ma[or concluslons of Lhe research sLudles. Lach of Lhe sLudles
clLed has been revlewed by an experL researcher ln Lhe crlmlnal [usLlce sysLem for
meLhodologlcal soundness and lnLerpreLaLlon of Lhe flndlngs.
38
Many of Lhe sLudles focus on
general populaLlons and may noL be generallzable Lo speclal populaLlons, such as women
offenders, sex offenders, and so on. 8eaders are encouraged Lo refer Lo Lhe source documenLs
for more ln-depLh deLall abouL Lhe sLudy meLhodology, how concepLs were measured, Lhe
sLudy populaLlon, and oLher conLexLual lnformaLlon LhaL help puL Lhe flndlngs lnLo perspecLlve.
ln addlLlon, cerLaln areas of Lhe crlmlnal [usLlce sysLem have been sLudled more rlgorously Lhan
oLhers and as a resulL Lhere are gaps ln Lhe research LhaL wlll be evldenL Lo Lhe reader. lor
example, Lhere ls very llLLle research on pollce declslons Lo arresL or lssue clLaLlons. Also, some
of Lhe sLudles presenLed here are very recenL, oLhers are noL because Lhere are no currenL
research sLudles LhaL have produced beLLer or dlfferenL resulLs. llnally, new research ls
publlshed rouLlnely, and readers should be mlndful LhaL new sLudles may have relevanL
flndlngs LhaL are noL lncluded ln Lhls maLrlx.
HOW TO READ THE MATRI"
1he research sLudles have been caLegorlzed lnLo one of four caLegorlesXWhaL uoesn'L Work,
WhaL Works, WhaL's romlslng, and WhaL's noL Clear.
1he WhaL uoesn'L Work" caLegory lncludes flndlngs based on rlgorous and
meLhodologlcally sound research LhaL repeaLedly shows (elLher Lhrough numerous
slngle sLudles or meLa-analysls sLudles) LhaL Lhe lnLervenLlon does noL have Lhe lnLended
or deslred resulLs.
1he WhaL Works" caLegory ls based on rlgorous and meLhodologlcally sound research
LhaL demonsLraLes slgnlflcanL poslLlve flndlngs (elLher Lhrough numerous slngle sLudles
or meLa-analysls sLudles).
1he WhaL's romlslng" caLegory lncludes flndlngs LhaL show promlse buL requlre more
rlgorous emplrlcal sLudy.
1he flnal caLegory, WhaL's noL Clear," lncludes sLudles LhaL have confllcLlng flndlngs
(l.e., one sLudy shows someLhlng works whlle anoLher sLudy shows LhaL lL doesn'L).
1hese flndlngs requlre addlLlonal emplrlcal sLudy.
1he flrsL column conLalns a brlef summary of Lhe meLhodology and ma[or flndlngs LhaL are
relevanL for evldence-based declslon maklng ln Lhe crlmlnal [usLlce sysLem. 1he second column
noLes meLhodologlcal conslderaLlons LhaL may lmpacL Lhe generallzablllLy of Lhe flndlngs. 1he
Lhlrd column hlghllghLs Lhe varlous declslon polnLs wlLhln Lhe crlmlnal [usLlce sysLem for whlch
Lhe flndlngs are relevanL and a summary of posslble pollcy and pracLlce lmpllcaLlons.

58
The auth!rs wish t! ackn!wledge the significant c!ntributi!ns !f the f!ll!wing researchers, wh!se reviews appear in wh!le !r
in part in this matrix: Melissa Alexander, Tim!thy Bynum, Ed Latessa, Chris L!wenkamp, R!ger Pryzybylski, and Ralph Serin.
43
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
What D!esnt W!rk In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-
analyses that dem!nstrate null !r negative !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A revlew of seven meLa-analyses lnvesLlgaLlng
Lhe rlsk prlnclple (deflned as Lhe probablllLy of
reoffendlng) found LhaL provldlng lnLense
correcLlonal lnLervenLlons Lo low rlsk offenders
does noL decrease recldlvlsm and may even
lncrease recldlvlsm raLes. 1he reasons clLed for
fallure lncluded exposure of low rlsk offenders
Lo hlgh rlsk offenders (l.e., anLlsoclal peers)
and dlsrupLlon of Lhe facLors LhaL make Lhem
low rlsk (l.e., sLrong famlly Lles, [ob, eLc.).
ltlmotyocltotloo.oLowenkamp & LaLessa (2004)
none noLed. lmpllcotloos.
1he ma[orlLy of servlces
and more lnLenslve
supervlslon should be
dlrecLed Lo hlgher rlsk
offenders.
ulverslon declslons
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
supervlslon sLraLegy
A meLa-analysls of 29 sLudles found LhaL Lhere
ls no overall effecL of booL camps on recldlvlsm
(l.e., Lhere was nearly equal odds of
recldlvaLlng beLween Lhe booL camp and
comparlson groups). !uvenlle booL camps were
less effecLlve overall Lhan adulL booL camps.
ltlmotyocltotloo.oMacoenzle, Wllson, & older
(2001)
5oppottloqocltotloo. Wllson, Macoenzle, &
MlLchell (2003)
1he sLudy lncluded
29 experlmenLal and quasl-
experlmenLal sLudles and used
offlclal daLa and mulLlple lndlces
of recldlvlsm.
1here was conslderable varlaLlon
among Lhe sLudles. ln nlne
sLudles, booL camp parLlclpanLs
had lower recldlvlsm raLes Lhan
dld comparlson groups, ln elghL
sLudles, comparlson groups had
lower recldlvlsm raLes, and ln Lhe
remalnlng sLudles, no slgnlflcanL
dlfferences were found.
Cf Lhe 29 ellglble sLudles, only 9
were publlshed ln peer-revlewed
[ournals and Lhe year of publlc-
caLlon was noL consldered. Also,
Lhere was lnsufflclenL
lnformaLlon on sample
demographlcs (gender, eLhnlclLy)
for comparlsons, some adulL booL
camps lncluded [uvenlles, and
programmlng lnformaLlon was
lncompleLe.
lmpllcotloos.
8ooL camps (especlally
[uvenlle booL camps) are
of doubLful efflcacy.
CommunlLy
lnLervenLlon sLraLegy
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
44
What D!esnt W!rk In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-
analyses that dem!nstrate null !r negative !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of 117 sLudles lnvolvlng
442,471 offenders showed LhaL none of Lhe
Lhree LreaLmenL" condlLlons-lengLh of Llme
lncarceraLed, servlng an lnsLlLuLlonal senLence
versus recelvlng a communlLy-based sancLlon,
and recelvlng an lnLermedlaLe sancLlon-were
assoclaLed wlLh a reducLlon ln recldlvlsm. ln
facL, longer Llme perlods ln prlson were
assoclaLed wlLh %1X/16).%+.X/1X).6/3/?/+0,
compared Lo shorLer Llme perlods ln prlson.
1hese effecLs held across gender, adulLs/
[uvenlles, race, and rlsk level of Lhe offender.
1here was some evldence LhaL more sLrlngenL
sancLlons may affecL females more adversely
Lhan males.
ltlmotyocltotloo.oSmlLh, Coggln, & Cendreau
(2002)
5oppottloqocltotloos.oCendreau, Coggln, &
Cullen (1999), Llpsey & Cullen (2007)
1o be lncluded ln Lhe meLa-
analysls, Lhe sLudy musL have
used a follow-up perlod of aL
leasL slx monLhs and musL have
provlded sufflclenL lnformaLlon Lo
calculaLe an effecL slze beLween
Lhe sancLlon and recldlvlsm.
SLudles of LreaLmenL servlces LhaL
also employed a sancLlon were
ellglble for lncluslon ln Lhe
analysls.
Many of Lhe prlson-based sLudles
lncluded ln Lhe analysls lacked
essenLlal descrlpLlve lnformaLlon
regardlng sLudy meLhodology
(e.g., condlLlons of conflnemenL).
lmpllcotloos.
SancLlons on Lhelr own
do noL change offender
behavlor or reduce
recldlvlsm. More severe
sancLlons (l.e., longer
prlson senLences) may
lncrease recldlvlsm.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
A sLudy of 14 lnLenslve Supervlslon
uemonsLraLlon rograms found LhaL a hlgher
percenLage of lndlvlduals on lS were
lncarceraLed durlng Lhe one year follow-up
perlod Lhan Lhe conLrol group. 1here were no
dlfferences ln arresLs for new crlmes beLween
Lhe LreaLmenL and conLrol groups. Powever,
lS was assoclaLed wlLh more Lechnlcal
vlolaLlonsX81 of Lhe lS offenders had
Lechnlcal vlolaLlons compared wlLh 33 of
Lhose ln Lhe conLrol group. ln addlLlon, flve
Llmes as many lS offenders were reLurned Lo
prlson for Lechnlcal vlolaLlons as compared Lo
Lhe conLrol group (21 compared Lo 4). 1he
auLhors also concluded LhaL lS dld noL resulL
ln cosL savlngs durlng Lhe one year follow-up
perlod and LhaL lS ulLlmaLely cosL 30 more
Lhan LradlLlonal probaLlon or parole
supervlslon.
ltlmotyocltotloo. eLersllla & 1urner (1993)
uaLa were collecLed ln each slLe
on offender demographlcs, prlor
crlmlnal hlsLory, currenL offense,
and dependence and LreaLmenL
hlsLory. uaLa on servlces
recelved, parLlclpaLlon ln
LreaLmenL and work programs,
and recldlvlsm (Lechnlcal
vlolaLlons, arresLs, and
lncarceraLlon) were collecLed aL
Lhe slx- and Lwelve-monLh polnLs
of supervlslon.
lmpllcotloos.
SLrlngenL supervlslon
condlLlons Lend Lo
produce more Lechnlcal
vlolaLlons and more
lncarceraLlon and do noL
reduce recldlvlsm by
Lhemselves.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
supervlslon sLraLegy
43
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
What D!esnt W!rk In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-
analyses that dem!nstrate null !r negative !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of more Lhan 400 research
sLudles LhaL examlned Lhe effecLs of
punlshmenL on recldlvlsm found LhaL
punlshmenL produced almosL ldenLlcal effecLs
on recldlvlsm as dld no punlshmenL or reduced
punlshmenL. 1hls lncluded drug LesLlng,
elecLronlc monlLorlng, flnes, lnLermlLLenL
lncarceraLlon, resLlLuLlon, Scared SLralghL
programs, and lncarceraLlon.
ltlmotyocltotloo.oCendreau & Coggln (1996)
Whlle all sLudles lncluded had a
comparlson group, Lhe crlLerla for
sLudy lncluslon were noL
provlded and no conLrols were
added (e.g., quallLy of research
deslgn, dosage, eLc.).
lmpllcotloos.
SancLlons on Lhelr own
do noL change offender
behavlor or reduce
recldlvlsm.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
lnLervenLlon sLraLegy
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
46
What W!rks In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-analyses
dem!nstrating significant p!sitive !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
MeLa-analyses of more Lhan 100 correcLlonal programs and
LreaLmenL research sLudles show LhaL Lhe rlsk of recldlvlsm
ls greaLly reduced (10-30 on average) when aLLenLlon ls
pald Lo deallng wlLh crlmlnogenlc needs (dynamlc rlsk
facLorsKXe.g., anLlsoclal aLLlLudes and values, anLlsoclal peers,
cerLaln personallLy and LemperamenL LralLs, famlly and
relaLlonal facLors, subsLance abuse, employmenL, school and
occupaLlonal Lralnlng, and Lhe use of personal and lelsure Llme).
1hese sLudles also found LhaLXLhe mosL powerful
approaches Lo changlng offender behavlor lnclude cognlLlve
behavloral and soclal learnlng sLraLegles (e.g., modellng,
relnforcemenL, and sklll acqulslLlon) ln Lhe conLexL of a
quallLy lnLerpersonal relaLlonshlp, more lnLenslve levels of
LreaLmenL are mosL effecLlve wlLh hlgher rlsk offenders (Lhe
rlsk prlnclple), lnLervenLlon efforLs should LargeL mulLlple
crlmlnogenlc needs (Lhe need prlnclple), and effecLlve
lnLervenLlons are Lhose LhaL are responslve Lo Lhe
moLlvaLlon, cognlLlve ablllLy, and oLher characLerlsLlcs
of Lhe offender (Lhe responslvlLy prlnclple).
lurLher flndlngs lncludeXrecldlvlsm reducLlon effecLs are
sllghLly greaLer when communlLy-based servlces and
lnLervenLlons are dellvered ln Lhe communlLy as compared
Lo servlces dellvered ln resldenLlal/lnsLlLuLlonal seLLlngs,
afLercare and follow-up servlces LhaL provlde a conLlnuum of
care are also necessary Lo manage and prevenL relapse,
recldlvlsm sllghLly lncreased when lnapproprlaLe
correcLlonal servlces were provlded (l.e., LreaLmenL servlces
LhaL do noL adhere Lo Lhe rlsk, need, and responslvlLy
prlnclples).
1hese flndlngs hold across communlLy correcLlons,
resldenLlal correcLlons, dlverslonary programs, males and
females, [uvenlle and adulL correcLlons, resLoraLlve and non-
resLoraLlve [usLlce programs, dlfferenL Lypes of LreaLmenL,
and dlfferenL Lypes of needs LargeLed.
ltlmotyocltotloo.oAndrews (2007)
5oppottloqocltotloos. Andrews & uowden (2007), Andrews
eL al. (1990), Andrews & 8onLa (2006), 8onLa (2007)
1he auLhors
acknowledge LhaL
furLher meLa-analyLlc
revlew on responslvlLy
ls needed, and LhaL
undersLandlng of Lhe
rlsk prlnclple ls sLlll
llmlLed by Lhe relaLlvely
few sLudles LhaL reporL
separaLe effecLs for
lower and hlgher rlsk
cases.
lmpllcotloos.
8ecldlvlsm ls more llkely
reduced when Lhe
[usLlce sysLem focuses
on crlmlnogenlc needs,
uses a cognlLlve
behavloral approach,
reserves more lnLenslve
servlces for Lhe hlgher
rlsk offender, and uses
afLercare servlces.
Charglng declslon
lea negoLlaLlons
SenLenclng declslon
CommunlLy
lnLervenLlon sLraLegy
47
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
What W!rks In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-analyses
dem!nstrating significant p!sitive !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of more Lhan 800 rlgorous program
evaluaLlons found LhaL a number of approaches
demonsLraLed a reducLlon ln recldlvlsm raLes, lncludlng
LreaLmenL-orlenLed lnLenslve supervlslon (22 reducLlon)
compared Lo no reducLlon for survelllance-orlenLed
lnLenslve supervlslon, cognlLlve behavloral LreaLmenL for sex
offenders ln prlson (13), vocaLlonal educaLlon ln prlson
(13), drug LreaLmenL ln Lhe communlLy (12), adulL drug
courLs (11), and cognlLlve behavloral programs ln general
(8). CognlLlve behavloral LreaLmenL for low rlsk sex
offenders on probaLlon achleved a 31 reducLlon ln
recldlvlsm. Cverall, cognlLlve behavloral approaches were
conslsLenLly found Lo be more effecLlve ln reduclng Lhe
recldlvlsm raLe across a varleLy of correcLlonal conLexLs and
offender populaLlons
CosL savlngs were also subsLanLlal. ApproxlmaLe per person
cosL savlngs examples lnclude $11,000 for LreaLmenL-
orlenLed lnLenslve supervlslon, $13,700 for vocaLlonal
educaLlon ln prlson, $10,000 for communlLy drug LreaLmenL,
and $10,000 for cognlLlve behavloral approaches. Whlle
Lhe absoluLe dlfferences ln Lhe recldlvlsm raLes ln some
slLuaLlons may have been modesL, even small reducLlons ln
Lhe raLe can have conslderable economlc and soclal beneflLs.
ltlmotyocltotloos. Aos, Mlller, & urake (2006a), Aos, Mlller,
& urake (2006b)
none noLed. lmpllcaLlonsX
Lmphasls should be
placed on LreaLmenL
LargeLs (l.e.,
crlmlnogenlc needs)
uslng a varleLy of
lnLervenLlons, especlally
cognlLlve behavloral
programmlng. ueclslons
regardlng correcLlonal
lnvesLmenLs should
conslder Lhe
cosL/beneflL of Lhe
lnLervenLlon.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
supervlslon sLraLegy
robaLlon/parole
vlolaLlon response
A meLa-analysls of several hundred sLudles of crlmlnal
[usLlce lnLervenLlons found LhaL when core correcLlonal
pracLlces (e.g., Lhe effecLlve use of auLhorlLy, modellng and
relnforclng prosoclal aLLlLudes, Leachlng concreLe problem-
solvlng skllls, advocaLlng for communlLy resources, and
bulldlng a relaLlonshlp LhaL allows for open communlcaLlon
and respecL) were used, parLlcularly ln comblnaLlon wlLh
adherence Lo Lhe rlsk, need, and responslvlLy prlnclples,
programs had beLLer LreaLmenL ouLcomes Lhan programs
LhaL dld noL use core correcLlonal pracLlces. 1he flndlngs
were parLlcularly Lrue for hlgher rlsk cases, programs LhaL
LargeLed crlmlnogenlc needs, and cllnlcally approprlaLe
LreaLmenL. 1he flndlngs of Lhe analysls held for varlous
offender and program characLerlsLlcs. 1he only core
correcLlonal pracLlce LhaL was noL assoclaLed wlLh slgnlflcanL
reducLlons ln raLes of reoffendlng was Lhe effecLlve use of
auLhorlLy.
ltlmotyocltotloo. uowden & Andrews (2004)
5oppottloqocltotloos. 8onLa eL al. (2008), 1roLLer (1996)
none noLed. lmpllcotloos.
ALLenLlon Lo sLaff
characLerlsLlcs and skllls
ls necessary Lo enhance
ouLcomes wlLh
offenders.
CommunlLy
lnLervenLlon sLraLegy
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
48
What W!rks In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Rig!r!us and meth!d!l!gically s!und research and meta-analyses
dem!nstrating significant p!sitive !utc!mes
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of randomlzed or quasl-experlmenLal
sLudles found LhaL cognlLlve behavloral Lherapy (C81) ls
effecLlve ln reduclng recldlvlsm by as much as 23 Lo 30
under cerLaln condlLlons. 1he effecLs lncreased when Lhe
LreaLmenL dosage was lncreased, when hlgher rlsk offenders
were LargeLed, and when Lhe quallLy of lmplemenLaLlon was
monlLored. 1he effecLs held for all brands of currlculum,
adulL and [uvenlle offenders, male and female offenders,
and mlnorlLy/non-mlnorlLy offenders.
ltlmotyocltotloo. Llpsey, Landenberger, & Wllson (2007)
5oppottloqocltotloos. Landenberger & Llpsey (2003), Wllson,
8ouffard, & Macoenzle (2003)
1he analysls lncluded a
llmlLed number of
sLudles by caLegory.
lmpllcotloos.
rogrammlng dosage
should maLch offenders'
rlsk levels.
lea negoLlaLlons
ulverslon declslons
SenLenclng declslons
CommunlLy
lnLervenLlon sLraLegy
robaLlon vlolaLlon
response
A synLhesls of 18 meLa-analyses of correcLlonal lnLervenLlons
found slmllar resulLs wlLh regard Lo reduclng recldlvlsm.
lnLervenLlons LhaL uLlllzed lnLenslve crlmlnal sancLlonlng" or
were excluslvely deLerrence-based Lended Lo be lneffecLlve
or even lncreased recldlvlsm. Cn Lhe oLher hand, Lhere were
some lnLervenLlons LhaL were found Lo reduce recldlvlsm
by an average of 23 Lo 30. 1hls group of more effecLlve
lnLervenLlons predomlnanLly employed behavloral and/or
cognlLlve skllls Lralnlng meLhods." 1he overall concluslon
was LhaL Lhe programs LhaL work besL
are founded on an expllclL emplrlcally based model of
crlme causaLlon,
have a sound meLhod of assesslng rlsk of reoffendlng, and
offenders are asslgned dlfferenL levels of servlce and
supervlslon accordlngly,
conLaln a sound meLhod of assesslng crlmlnogenlc needs
and dynamlc rlsk facLors LhaL are llnked Lo offendlng,
requlre skllled and sLrucLured engagemenL by sLaff,
uLlllze cognlLlve behavloral approaches, and
are dellvered by personnel who have adequaLe Lralnlng
and resources.
ltlmotyocltotloo.oMcCulre (2001)
none noLed. lmpllcaLlonsX
rograms deslgned Lo
reduce recldlvlsm
should be monlLored
Lhrough conLlnuous
quallLy lmprovemenL
Lechnlques Lo ensure
LhaL Lhe program
condlLlons for
behavloral change
are meL.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
supervlslon sLraLegy
49
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A sLudy on a sancLlons grld used by parole fleld sLaff ln Chlo
Lo deLermlne Lhe approprlaLe response Lo vlolaLlons of
condlLlons of posL-release supervlslon lndlcaLed LhaL
moderaLe and hlgh rlsk offenders ln all supervlslon caLegorles
had a lower llkellhood of recldlvlsm afLer compleLlng a
halfway house program. Powever, low and low/moderaLe
rlsk offenders recldlvaLed more frequenLly when Lhey were
placed ln Lhese hlgher securlLy seLLlngs Lhan lnLo a sLralghL
communlLy placemenL. ln addlLlon, offenders ln Lhe parole
vlolaLor caLegory were Lhe only group LhaL experlenced a
slgnlflcanLly lower level of recldlvlsm across all rlsk levels
when placed ln halfway houses.
ltlmotyocltotloo.oAndrews & !anes (2006)
Cffenders ln a halfway
house program were
Lracked for Lwo years
posL release Lo
deLermlne Lhe
basellne recldlvlsm
raLe and Lhe
characLerlsLlcs of
Lhose mosL llkely Lo
succeed. 8ased on Lhls
research, a supervlslon
grld was creaLed Lo
classlfy offenders lnLo
four rlsk levels and
Lhree supervlslon
caLegorles.
1he arLlcle does noL
provlde deLalls on
Lhe research
meLhodology. 1he
research was
conducLed wlLh
offenders ln one sLaLe.
lmpllcotloos.
Palfway house
lnLervenLlons wlLh
supervlslon geared Lo
level of rlsk/need can
be effecLlve wlLh
hlgher rlsk offenders.
Low rlsk offenders
may do worse when
placed ln hlgh
securlLy/lnLenslve
supervlslon halfway
house programs.
!all or prlson release
declslons
A randomlzed experlmenL explorlng drug courL monlLorlng
found LhaL offenders asslgned Lo adapLlve lnLervenLlon (l.e., a
LreaLmenL-orlenLed response as opposed Lo a [udge-orlenLed
response) were more llkely Lo graduaLe, had fewer warranLs
lssued, and had more negaLlve drug screens (l.e., clean). 1he
effecLs were presenL for boLh low and hlgh rlsk offenders,
alLhough low rlsk offenders performed beLLer.
ltlmotyocltotloo. Marlowe eL al. (2008)
1he sample slze was
small-31 offenders.
ln addlLlon, Lhe
experlmenL was
conducLed ln a slngle
drug courL, whlch
makes generallzaLlon
problemaLlc.
lmpllcotloos.
urug courLs should be
admlnlsLered wlLh a
LreaLmenL orlenLaLlon.
lea negoLlaLlons
SenLenclng declslons
CommunlLy
lnLervenLlon
sLraLegy
robaLlon vlolaLlon
response
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
30
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A quasl-experlmenLal sLudy compared ouLcomes beLween
8reaklng Lhe Cycle counLles and non-8reaklng Lhe Cycle
counLles wlLh a LoLal sample slze of 3,600 adulL offenders.
(8reaklng Lhe Cycle ls a communlLy-based drug
LreaLmenL/lnLervenLlon program deslgned Lo address drug-
relaLed crlme.) 1he 8reaklng Lhe Cycle group had a sllghL buL
sLaLlsLlcally slgnlflcanL lower llkellhood of arresL for any
offense and slgnlflcanLly fewer drug arresLs overall. ln Lhe
8reaklng Lhe Cycle counLles LhaL admlnlsLered more drug
LesLs and sancLlons, offenders wlLh drug condlLlons had a
sLaLlsLlcally slgnlflcanL lower llkellhood of arresL for any
offense and slgnlflcanLly fewer drug arresLs.
An analysls of Lhe cosLs and beneflLs of Lhe 8reaklng Lhe Cycle
program found LhaL lL reLurned $2.30 Lo $3.70 for every dollar
lnvesLed. 1he concluslon was LhaL Lhe 8reaklng Lhe Cycle
program ls an effecLlve sLraLegy for reduclng drug arresLs for
offenders wlLh drug condlLlons.
ltlmotyocltotloo.oParrell eL al. (2003)
1he ma[or llmlLaLlon
ls Lhe rellance on
secondary daLa, whlch
llmlLed Lhe analyses
(for example, Lhere
were no daLa on
LreaLmenL uLlllzaLlon).
ln addlLlon, alLhough
some of Lhe flndlngs
were sLaLlsLlcally
slgnlflcanL, mosL
observed dlfferences
were modesL.
lmpllcotloos.
rograms deslgned Lo
achleve speclflc
ouLcomes should be
evaluaLed Lo
deLermlne Lhelr
effecLlveness and
overall cosL/beneflL.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
A sLudy of 130 low rlsk and 37 hlgh rlsk offenders found
sLrong supporL for Lhe rlsk prlnclple ln drug courLs. Plgh rlsk
offenders (who were scheduled Lo blweekly sLaLus hearlngs)
performed beLLer ln drug courL Lhan Lhose who were asslgned
Lo sLaLus hearlngs as usual (Lhey had more negaLlve drug
screens and beLLer aLLendance aL counsellng sesslons).
ltlmoty cltotloo. Marlowe eL al. (2006)
5oppottloqocltotloos.oLowenkamp, Polslnger, & LaLessa (2003)
1he sample slze for
Lhe hlgh rlsk group
was small (37 hlgh rlsk
offenders compared
Lo 130 low rlsk
offenders), and Lhere
was llmlLed follow-up
on lllegal behavlor,
whlch llmlLs Lhe ablllLy
Lo generallze abouL
Lhe sLaylng power of
Lhe effecLs.
lmpllcotloos.
urug courL
parLlclpanLs should be
selecLed based on rlsk
level (l.e., Lhe rlsk
prlnclple holds ln drug
courL seLLlngs).
ulverslon declslons
lea negoLlaLlons
SenLenclng declslons
CommunlLy
lnLervenLlon
sLraLegles
A sLudy found LhaL Lhe more Llme a probaLlon offlcer spenL
addresslng crlmlnogenlc needs and uslng behavloral
Lechnlques wlLh probaLloners, Lhe lower Lhe raLe of
recldlvlsm. Powever, only one Lhlrd of Lhe probaLlon offlcers
spenL a slgnlflcanL amounL of Llme ln Lhelr sesslons dlscusslng
Lhese needs. lurLher, Lhe more Llme spenL dlscusslng Lhe
condlLlons of probaLlon, Lhe hlgher Lhe recldlvlsm raLe. ln
slLuaLlons where less Lhan 13 mlnuLes were spenL dlscusslng
probaLlon condlLlons, Lhe recldlvlsm raLe was 19 compared
Lo 42 when more Llme was devoLed Lo dlscusslng probaLlon
condlLlons.
ltlmotyocltotloo.o8onLa eL al. (2008)
1hls was a slngle slLe
sLudy and Lhere were
problems ln
lmplemenLlng Lhe
lnLervenLlon model as
deslgned.
lmpllcotloos.
Supervlslon offlcers
should spend Lhe
ma[orlLy of Lhelr Llme
worklng wlLh
offenders on
crlmlnogenlc needs
(raLher Lhan focuslng
on condlLlons of
supervlslon LhaL are
non-crlmlnogenlc), use
behavloral Lechnlques,
and devoLe aL leasL 13
mlnuLes per sesslon Lo
lssues relaLed Lo
crlmlnogenlc needs.
CommunlLy
supervlslon sLraLegy
31
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A sLudy found LhaL [udges who used ball guldellnes were
more conslsLenL ln Lhelr declslon maklng regardlng release on
recognlzance Lhan [udges who dld noL use ball guldellnes. 1he
[udges who used guldellnes were more llkely Lo granL 8C8 Lo
non-serlously charged defendanLs and Lo be more sLrlngenL
wlLh defendanLs faclng more serlous charges Lhan Lhe conLrol
group, who lacked Lhls level of conslsLency ln Lhelr declslons.
ln addlLlon, wlLh regard Lo defendanLs classlfled wlLhln Lhe
cash ball declslon group ln Lhe guldellnes, 63 of Lhe [udges
who used guldellnes seL ball ln Lhls range, whlle only 38 of
Lhe [udges ln Lhe conLrol group seL ball slmllarly.
1he equlLy of ball declslons lnvolves declslon maklng ln whlch
one would expecL slmllarly slLuaLed" defendanLs Lo be
LreaLed ln a slmllar manner, whlch was conflrmed by Lhls
sLudy. 1he varlaLlon ln ball amounLs was subsLanLlally reduced
among Lhe [udges uslng guldellnes.
ltlmotyocltotloo.oColdkamp & CoLLfredson (1983)
1hls was an
experlmenLal sLudy of
ball guldellnes looklng
aL 960 cases and
conducLed over a 14-
monLh perlod. !udges
were randomly
asslgned Lo an
experlmenLal group,
whlch would use ball
guldellnes, or a
comparlson group,
whlch would seL ball
declslons as Lhey had
ln Lhe pasL.
1hls was a slngle slLe
sLudy.
lmpllcotloos.
rovldlng [udlclal
offlcers wlLh ob[ecLlve
lnformaLlon abouL
offenders'
backgrounds and
communlLy Lles (as
well as abouL Lhe
charges agalnsL Lhe
defendanL) coupled
wlLh Lhe use of a
valldaLed lnsLrumenL
helps produce more
equlLable and effecLlve
preLrlal declslons.
reLrlal release
declslons
A revlew of 30 sLudles (of 33 drug courLs) found LhaL Lhe
recldlvlsm raLe (for boLh drug and non-drug offenses) was
lower on average for drug courL parLlclpanLs Lhan for Lhose ln
Lhe comparlson group (38 compared Lo 30). 1hree sLudles
LhaL used random asslgnmenL and dld noL have a hlgh
parLlclpanL aLLrlLlon raLe demonsLraLed a reducLlon from 30
Lo 43. ln addlLlon, oLher sLudles LhaL used a group of ellglble
buL non-referred offenders as Lhe comparlson group also
observed a moderaLe reducLlon ln reoffendlng.
rograms LhaL used elLher a pre-plea or posL-plea model were
more effecLlve Lhan Lhose LhaL employed a mlxed model.
Moreover, programs LhaL offered a clear lncenLlve for
compleLlon (e.g., dlsmlssal of charges) had greaLer success
Lhan Lhose LhaL dld noL. llnally, drug courLs LhaL used a slngle
dedlcaLed provlder were more successful because Lhey were
more llkely Lo use a cognlLlve behavloral model.
ltlmotyocltotloo.oWllson, MlLchell, & Macoenzle (2006)
none noLed. lmpllcotloos.
urug courLs should
conslder adopLlng a
pre-plea or posL-plea
model, provldlng
offenders wlLh
lncenLlves for
compleLlon, and uslng
cognlLlve behavloral
Lechnlques.
ulverslon declslons
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
32
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of 140 sLudles of communlLy (lnLermedlaLe)
sancLlons and 323 sLudles of lncarceraLlon found LhaL, for
lnLermedlaLe sancLlons, Lhere appeared Lo be a neL
wldenlng" effecL Lhrough Lhe LargeLlng of lndlvlduals who
would noL have prevlously recelved as severe a sancLlon. ln
addlLlon, Lhere was no lndlcaLlon LhaL Lhese more severe
sancLlons were more effecLlve Lhan LradlLlonal communlLy
supervlslon. ln Lhe 47 sLudles of lnLenslve supervlslon
lncluded ln Lhls revlew, Lhere was no dlfference beLween Lhe
groups, wlLh each havlng a recldlvlsm raLe of 29. Powever,
Lhere was an lndlcaLlon LhaL Lhe lncluslon of a LreaLmenL
componenL wlLh Lhe lnLenslve supervlslon program resulLed
ln a 10 reducLlon ln recldlvlsm.
1he analysls of wheLher longer perlods of lncarceraLlon
produced lower recldlvlsm raLes lncluded Lwo componenLsX
one comparlng slmllar offenders who spenL more Llme
(averaglng over 30 monLhs) ln prlson compared wlLh less
(averaglng less Lhan 17 monLhs) and Lhe second comparlng
offenders who were senL Lo prlson for a brlef Llme wlLh a
slmllar group noL recelvlng a prlson senLence. nelLher of
Lhese analyses exhlblLed dlfferenL effecLs on recldlvlsm.
ltlmotyocltotloo.oCendreau, Coggln, Cullen, & Andrews (2001)
MeLhodologlcal rlgor
was noL lncluded as a
crlLerlon for lncluslon
ln Lhe meLa-analysls.
lmpllcotloos.
lnLermedlaLe
sancLlons should
be uLlllzed wlLh
recognlLlon of boLh
Lhelr ablllLy Lo achleve
cerLaln ouLcomes and
Lhelr llmlLaLlons, such
as accounLablllLy as
opposed Lo rlsk
reducLlon. Careful
conLrols should be puL
ln place when
lmplemenLlng
lnLermedlaLe sancLlons
Lo avold unlnLended
neL wldenlng.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
A meLa-analysls of 131 sLudles for almosL 730,000 adulL
offenders found LhaL Lhe sLrongesL predlcLors of recldlvlsm
proved Lo be crlmlnogenlc need, crlmlnal hlsLory/hlsLory of
anLlsoclal behavlor, soclal achlevemenL, age/gender/race, and
famlly facLors. 8oLh sLaLlc and dynamlc predlcLors proved
lmporLanL. Cverall, valldaLed rlsk assessmenL lnsLrumenLs
proved Lo be superlor Lo sLaLlc measures and lndlces of
anLlsoclallLy. Larly famlly facLors and pre-adulL anLlsoclal
behavlor are correlaLed wlLh recldlvlsm buL are rarely
lncluded ln adulL offender rlsk assessmenLs. locus on
personal dlsLress, soclal class, and, Lo a lesser exLenL,
lnLelllgence ls conLralndlcaLed based on Lhe emplrlcal
evldence.
ltlmotyocltotloo. Cendreau, Coggln, & LlLLle (1996)
5oppottloqocltotloo. Andrews eL al. (1990), lrench &
Cendreau (2003)
1he sLudles lncluded
ln Lhe meLa-analysls
had an over-
represenLaLlon of
males ln Lhelr samples.
lmpllcotloos.
valldaLed rlsk
assessmenLs should be
used and lnclude boLh
sLaLlc and dynamlc rlsk
facLors.
Charglng declslons
ulverslon declslons
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
lnLervenLlon
sLraLegy
33
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A meLa-analysls of 70 prlson-based LreaLmenL sLudles found
hlgher effecL slzes resulLlng from behavloral programs and
programs wlLh greaLer lnLegrlLy ln Lerms of lmplemenLaLlon.
ln parLlcular, programs LhaL LargeLed crlmlnogenlc needs had
lncreased effecLs on recldlvlsm, whlch lncreased wlLh Lhe
number of crlmlnogenlc needs LargeLed. Cverall, Lhe sLudy
found LhaL mlsconducL was reduced by abouL 26 Lhrough
programmlng.
ltlmotyocltotloo. lrench & Cendreau (2003)
1he meLa-analysls had
few sLudles of women
offenders, and lL dld
noL conLrol for facLors
LhaL have been
demonsLraLed Lo
lnfluence mlsconducL
(l.e., prlson
overcrowdlng,
populaLlon lnsLablllLy
Lhrough Lransfers,
securlLy level, eLc.).
1he auLhors noLe LhaL
lmporLanL offender
characLerlsLlcs (rlsk,
need, mlsconducL
hlsLory) may moderaLe
Lhe flndlngs.
lmpllcotloos.
Lnhanced prlson
managemenL wlll
resulL Lhrough a
sLraLegy ln whlch
programmlng has a
cenLral role.
SenLenclng declslons
CorrecLlonal
programmlng
declslons
A summary of 30 meLa-analyses found LhaL overall LreaLmenL
reduces recldlvlsm abouL 9-10, and sllghLly hlgher for
approprlaLe" servlces, when Lhe program ls maLched Lo Lhe
offender's unlque LralLs, communlLy programs have greaLer
effecL slzes, Lhere ls some lnfluence of age of offenders on
recldlvlsm ouLcome, and larger effecL slzes are derlved from
programs wlLh hlgher rlsk offenders.
ltlmotyocltotloo. McCulre (2002)
5oppottloqocltotloo.olrench & Cendreau (2003)
1hls ls a summary of
evaluaLlon sLudles and
does noL have any
conLrols. ln addlLlon,
evaluaLlons of [uvenlle
programs are over-
represenLed ln Lhe
summary, as are
males.
lmpllcotloos.
1reaLmenL
programmlng should
be LargeLed Lo hlgher
rlsk offenders and
Lhelr crlmlnogenlc
needs, and preferably
(Lhough noL
excluslvely) be
communlLy based.
lea negoLlaLlons
SenLenclng
recommendaLlons
SenLenclng declslons
CommunlLy
lnLervenLlon
sLraLegy
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
34
Whats Pr!mising In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Studies that sh!w pr!mising !utc!mes but require m!re rig!r!us research
METHODOLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
1he effecLlveness of graduaLed sancLlons ln deLerrlng non-
compllanL acLs ls conLlngenL on Lhe cerLalnLy, swlfLness, and
falrness (conslsLency and proporLlonallLy) of Lhe response. ln
addlLlon, Lhe supervlslon process musL be proacLlve and have
Lhe followlng crlLlcal elemenLsX(a) lL musL lnform Lhe offender
abouL Lhe behavlor LhaL consLlLuLes an lnfracLlon and abouL
Lhe poLenLlal consequence for LhaL behavlor, (b) lL musL
ensure LhaL Lhe [udlclary, supervlslon agenLs, and oLher
LreaLmenL agencles adhere Lo Lhe sancLlonlng model, and
(c) lL musL uphold Lhe offender's dlgnlLy LhroughouL Lhe
process of change. 1hus, a sound graduaLed sancLlons model
should clearly deflne lnfracLlons, uLlllze a swlfL process for
respondlng Lo lnfracLlons, respond Lo sancLlons uslng a
sLrucLured sancLlon menu wlLh consequences, and employ
behavloral conLracLs for offenders wlLh wrlLLen offender
acknowledgemenL of vlolaLlon behavlor.
ltlmotyocltotloo. 1axman, Soule, & Celb (1999)
1hls ls noL a research
pro[ecL LhaL makes
sLaLlsLlcal lnferences
Lo a larger populaLlon,
however, Lhe
dlscusslon ls
supporLed by clLaLlon
of numerous
lndlvldual sLudles.
lmpllcotloos.
lmmedlacy, falrness,
conslsLency, and
proporLlonallLy ln
respondlng Lo
mlsbehavlor are
lmporLanL.
CommunlLy
lnLervenLlon
sLraLegy
robaLlon/parole
vlolaLlon response
A sLudy predlcLlng rlsk uslng an assessmenL lnsLrumenL
for preLrlal populaLlons examlned Lhe followlng facLorsX
charge Lype, pendlng charges, ouLsLandlng warranLs, prlor
convlcLlons, prlor fallures Lo appear, prlor vlolenL convlcLlons,
lengLh of Llme aL currenL resldence, employmenL sLaLus, and
hlsLory of drug abuse. SLaLlsLlcal analysls showed LhaL Lhe
lnsLrumenL seemed Lo predlcL equally across gender, race,
and geographlc locaLlon.
1he sLudy found LhaL noL only dld Lhe lnsLrumenL predlcL for
fallure Lo appear (l.e., hlgh rlsk defendanLs were less llkely Lo
appear), buL lL also predlcLed for danger Lo Lhe communlLy
(l.e., hlgher rlsk defendanLs were more llkely Lo be arresLed
preLrlal) and for fallure due Lo Lechnlcal vlolaLlons (l.e., hlgher
rlsk defendanLs were more llkely Lo have Lechnlcal vlolaLlons).
A slmllar LesL ln lederal CourL found LhaL offenders wlLh
dlfferenL rlsk levels may respond Lo preLrlal condlLlons
dlfferenLly. ln addlLlon, mosL condlLlons dld noL have an
lmpacL on recldlvlsm rlsk for low rlsk offenders. 1hls flndlng ls
supporLed by anoLher sLudy of lederal ulsLrlcL CourL ln Lhe
ulsLrlcL of Columbla.
ltlmotyocltotloos. vannosLrand (2003), vannosLrand &
oeebler (2009)
5oppottloqocltotloo.oWlnLerfleld, Coggeshall, & Parrell (2003)
1here ls no measure of
assoclaLlon beLween
rlsk score and
ouLcome (e.g., fallure
Lo appear or rearresL).
ln Lhe lederal sLudy,
Lhere were no daLa
on fulflllmenL of
condlLlons or Lhe
quallLy of servlces.
lmpllcotloos.
8y assesslng rlsk,
declslon makers are
able Lo base Lhe use of
preLrlal deLenLlon and
release condlLlons on
level of rlsk.
reLrlal release
declslons
33
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Whats N!t Clear In Reducing Pretrial Misbehavi!r And Offender Recidivism
MAJOR RESEARCH FINDINGS
Findings that c!ntradict !r c!nflict with !ther studies and require additi!nal
rig!r!us research
METHODLOGICAL
CONSIDERATIONS
IMPLICATIONS &
RELEVANT DECISION
POINTS
A sLudy of 2,014 adulL and [uvenlle offenders ln flve slLes found
LhaL offenders placed ln Lhe 1reaLmenL AlLernaLlves Lo SLreeL
Crlme (1ASC) program had lower drug use ln Lhree of Lhe flve
slLes sLudled. 1wo of Lhe slLes reporLed fewer drug crlmes
based on self-reporL daLa, and Lhere was no dlfference ln
reoffendlng ln Lhree slLes. ln addlLlon, 1ASC offenders
performed worse ln Lerms of new arresLs and Lechnlcal
vlolaLlons ln Lwo slLes.
ltlmotyocltotloo.oAnglln, Longshore, & 1urner (1999)
1he follow-up perlod
was only slx monLhs.
Also, Lhe
comparlsons of 1ASC
were made Lo oLher
lnLervenLlons or
probaLlon raLher
Lhan a LreaLmenL/no
LreaLmenL
comparlson.
lmpllcotloos.
noL appllcable
lea negoLlaLlons
SenLenclng
recommendaLlon
SenLenclng declslons
CommunlLy
supervlslon sLraLegy
A randomlzed experlmenL on Lhe effecLs of drug LesLlng durlng
preLrlal release on offender mlsconducL found Lhere was no
sLaLlsLlcally slgnlflcanL dlfference beLween Lhe LreaLmenL and
conLrol groups wlLh regard Lo fallure Lo appear or rearresL. 1he
overall concluslon ls LhaL Lhe use of drug LesLlng durlng Lhe
preLrlal perlod dld noL slgnlflcanLly reduce preLrlal mlsconducL.
ltlmotyocltotloo.o8rlLL, CoLLfredson, & Coldkamp (1992)
1here was slgnlflcanL
aLLrlLlon ln boLh
sLudy slLes. ln
addlLlon, ln one of
Lhe slLes, 20 of Lhe
LreaLmenL group dld
noL recelve a drug
LesL and, among
oLher lndlvlduals, Lhe
amounL of LesLlng
was varled. As such,
Lhere are concerns
abouL Lhe lnLegrlLy of
Lhe lnLervenLlon.
lmpllcotloos.
noL appllcable
reLrlal release
declslons
A sLudy of 1,378 defendanLs from 12 urban and rural counLles
ln norLh Carollna found LhaL Lhe serlousness of charges and Lhe
presence of codefendanLs lnfluenced Lhe flnal dlsposlLlon. 1he
serlousness of charges affecLed Lhe severlLy of Lhe senLence for
defendanLs who were found gullLy. 1he presence of
codefendanLs lncreased Lhe odds of dlsmlssal for Class 1 felony
defendanLs. uefendanLs' prlor crlmlnal hlsLory dld noL affecL
odds of dlsmlssal buL dld lncrease severlLy of senLenclng. 8lack
defendanLs charged wlLh Class 2 felonles were more llkely Lo
have longer sLays ln preLrlal deLenLlon. Longer Llme ln preLrlal
deLenLlon lnfluenced courL dlsposlLlon. WheLher Lhe defendanL
had a prlvaLe versus publlc defender dld noL affecL Lhe
llkellhood of charges belng dlsmlssed. lea bargalnlng was
relaLed Lo Lhe lengLh of senLence for moderaLe Lo hlgh rlsk
groups (where rlsk ls relaLed Lo deLenLlon).
ltlmotyocltotloo.oClarke & ourLz (1983)
8lsk was deflned as
Lhe probablllLy of
deLenLlon, noL Lhe
probablllLy of fuLure
reoffendlng.
lmpllcotloos.
noL appllcable
Charglng declslons
lea negoLlaLlons
reLrlal release
declslons
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
36
USING EVIDENCE TO INFORM DECISION MAKING
uesplLe Lhelr commlLmenL Lo applylng research Lo declslon maklng, some crlmlnal [usLlce
professlonals express confuslon over how Lo apply evldence when lL confllcLs wlLh personal
experlence. Pow much emphasls should be placed on research versus experlence? 1hls Lenslon
ls undersLandable, parLlcularly when research ls ln opposlLlon Lo lnLulLlon or experlence (such
as Lhe emplrlcally supporLed flndlngs LhaL provldlng programmlng Lo lower rlsk offenders can
lncrease recldlvlsm or LhaL lncreaslng Lhe degree of punlshmenL can lncrease recldlvlsm). Lven
when research ls noL ln opposlLlon Lo bellefs or experlence, ouLcomes are never a 100
guaranLee (l.e., some false poslLlves and false negaLlves are Lo be expecLed, regardless of Lhe
sLrengLh of Lhe evldence), alLhough, when followlng Lhe evldence, favorable ouLcomes are
more llkely Lo occur Lhan unfavorable ouLcomes.
1he model below ls presenLed as a way Lo reconclle seemlngly conLradlcLory undersLandlngs.
lL suggesLs Lhe followlngX
1. ueclslon makers begln Lhe declslon maklng process wlLh an undersLandlng of Lhe exlsLlng
research. ln some cases, Lhe relevanL research flndlngs wlll be falrly robusL, ln oLhers, lL wlll
be sparse or absenL.
2. When Lhe research ls lnsufflclenL, declslon makers defer Lo promlslng pracLlce flndlngs.
1hese flndlngs are weaker Lhan research evldence because Lhey elLher have noL been
sub[ecL Lo rlgorous LesLlng or been repllcaLed, noneLheless, Lhey can provlde more exLernal
explanaLory power Lhan bellef or personal experlence alone.
3. When personal experlence confllcLs wlLh research evldence/promlslng pracLlce, declslon
makers welgh Lhe preponderance of evldence wlLh Lhe sLrengLh of experlence.
4. lf Lhe concluslon lnferred from Lhe evldence ls noL followed, declslon makers are
encouraged Lo monlLor ouLcomes Lo deLermlne lf Lhe deslred resulLs are achleved. WlLhouL
Lhls, percepLlons wlll nelLher be afflrmed nor challenged and new learnlng wlll noL resulL.
ersonal
Lxperlence
8esearch
Lvldence
-1-
WhaL do we know
emplrlcally abouL
whaL works and
whaL doesn'L work?
-2-
WhaL has been
Lrled LhaL ls
promlslng?
-3-
WhaL does my
experlence Lell
me?
-4-
ucl5lON
(mooltotoooJo
evolooteotesolts)
'67D=;7X%68L667 *MH69=67<6XC7IX&6D6C9<E
37
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
APPENDIX 4: 2009 ZOGBY INTERNATIONAL
PUBLIC OPINION SURVEY
Zogby lnLernaLlonal was commlssloned by Lhe naLlonal lnsLlLuLe of CorrecLlons and lLs
Lvldence-8ased ueclslon Maklng ln Local Crlmlnal !usLlce SysLems parLners Lo conducL a
Lelephone survey of llkely voLers from !uly 31, 2009 Lo AugusL 4, 2009. 1he LargeL sample was
1,003 lnLervlews, wlLh approxlmaLely 39 quesLlons asked. Samples were randomly drawn from
Lelephone compacL dlscs of a naLlonal llsLed sample.
Zogby lnLernaLlonal surveys employ sampllng sLraLegles ln whlch selecLlon probablllLles are
proporLlonal Lo populaLlon slze wlLhln area codes and exchanges. up Lo slx calls are made Lo
reach a sampled phone number. CooperaLlon raLes are calculaLed uslng one of Lhe Amerlcan
AssoclaLlon of ubllc Cplnlon 8esearch's approved meLhodologles
39
and are comparable Lo
oLher professlonal publlc-oplnlon surveys conducLed uslng slmllar sampllng sLraLegles.
60
WelghLlng by reglon, pollLlcal parLy, age, race, rellglon, and gender ls used Lo ad[usL for
non-response. 1he margln of error ls +/- 3.2 percenLage polnLs.
A reporL from nlC ls anLlclpaLed ln 2010 Lo more fully descrlbe Lhls sLudy and lLs resulLs.

59
The American Ass!ciati!n !f Public Opini!n Research, 2009.
60
Sheppard & Haas, 2003.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
38
APPENDIX 5: GLOSSARY OF TERMS
1he Lerms used ln Lhls documenL have speclflc meanlngs wlLhln Lhe conLexL of a harm reducLlon
phllosophy and an evldence-based declslon maklng model.
J)/0/1(7.1/69 ALLrlbuLes or characLerlsLlcs of Lhe lndlvldual or hls/her envlronmenL LhaL
produce or Lend Lo produce crlmlnal behavlor and recldlvlsm.
>%&%9 A collecLlon of observaLlons or sLaLlsLlcs used Lo measure and analyze lnLervenLlons.
>%&%K3)/?.19X1he ongolng collecLlon and analysls of daLa Lo Lrack performance and lnform
pollcy and pracLlce.
>.4.13%1&9 A person who has been formally charged wlLh a crlme.
:?/3.16.KG%+.39 Concluslons drawn from rlgorous research sLudles LhaL have been
repllcaLed numerous Llmes wlLh deflned, measurable ouLcomes abouL Lhe effecLlveness of
an lnLervenLlon or process.
L(%29 1he deslred end resulL of an efforL.
MGN.6&/?.9XMeasurable, shorL-Lerm lndlcaLors or benchmarks LhaL lndlcaLe progress ls belng
made Loward Lhe goal.
M44.13.)9 A person convlcLed of a crlmlnal charge.
M@&6(0.9XChange LhaL occurs as a resulL of an acLlon or lnLervenLlon.
I.)4()0%16.X0.%+@).9 A quanLlflable measure LhaL ls used Lo supporL Lhe declslon maklng
process by documenLlng how well speclflc funcLlons or processes are carrled ouL.
8.+.%)6$9 1he sysLemaLlc analysls of daLa, uslng sclenLlflc meLhods, Lo sLudy Lhe effecL of an
lnLervenLlon.
39
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
APPENDIX 6: BIBLIOGRAPHY
Adler, . S., owon, S., & Peckscher, C. (2008). erspecLlve-rofesslonal workX1he emergence
of collaboraLlve communlLy. Otqoolzotlooo5cleoce, 19(2), 339-376.
1he Amerlcan AssoclaLlon of ubllc Cplnlon 8esearch. (2009). 5tooJotJoJefloltloos.o
llooloJlsposltloosoofocoseocoJesoooJoootcomeototesoofosotveys, p. 38. 8eLrleved from
hLLpX//www.aapor.org/AM/1emplaLe.cfm?SecLlon=SLandard_ueflnlLlons&1emplaLe=/
CM/ConLenLulsplay.cfm&ConLenLlu=1819
Amerlcan Medlcal AssoclaLlon. (2002). AMAoptloclplesoofomeJlcoloetblcso(2001). Chlcago, lLX
AuLhor. 8eLrleved from hLLpX//www.clrp.org/llbrary/sLaLemenLs/ama/
Andrews, u. A. (2007). rlnclples of effecLlve correcLlonal programs. ln L. L. MoLluk
and 8. C. Serln (Lds.), compeoJlomo2000ooooeffectlveocottectloooloptoqtommloq.
CLLawa, CnXCorrecLlonal Servlce Canada. 8eLrleved from
hLLpX//www.csc-scc.gc.ca/LexL/rsrch/compendlum/2000/lndex-eng.shLml
Andrews, u. A., & 8onLa, !. (1998). 1beopsycboloqyoofoctlmloolocooJocto(2nd ed.). ClnclnnaLl, CPX
Anderson ubllshlng Company.
Andrews, u. A., & 8onLa, !. (2006). 1beopsycboloqyoofoctlmloolocooJocto(4Lh ed.). newark, n!X
Anderson ubllshlng Company.
Andrews, u. A., & 8onLa, !. (2007). klsk-oeeJ-tespooslvltyomoJelofotooffeoJetoossessmeotoooJo
tebobllltotloo (2007-06). CLLawaXubllc SafeLy Canada.
Andrews, u. A., 8onLa, !., & WormlLh, !. S. (2006). 1he recenL pasL and near fuLure of rlsk and/or
need assessmenL. ctlmeoooJouellopoeocy,o52(1), 7-27.
Andrews, u. A., & uowden, C. (2007). 1he rlsk-need-responslvlLy model of assessmenL ln human
servlce and prevenLlon and correcLlons crlme prevenLlon [urlsprudence. coooJloooIootooloofo
ctlmlooloqyoooJoctlmlooloIostlce,oX9(4), 439-464.
Andrews, u. A., uowden, C., & Cendreau, . (1999). cllolcollyotelevootoooJopsycboloqlcollyo
lofotmeJoopptoocbesotooteJoceJoteoffeoJloq.oAometo-ooolytlcostoJyoofobomooosetvlce,otlsk,o
oeeJ,otespooslvlty,oooJootbetocoocetosoloojostlceocootexts. unpubllshed manuscrlpL. CLLawa, CnX
CarleLon unlverslLy.
Andrews, S., & !anes, L. (2006). Chlo's evldence-based approach Lo communlLy sancLlons and
supervlslon. 1oplcsoloocommooltyocottectloos.offectlvelyoMoooqloqovlolotloosoooJokevocotloos,o
Aoooololssoe,o2006. WashlngLon uCXu.S. ueparLmenL of !usLlce, naLlonal lnsLlLuLe of
CorrecLlons.
Andrews, u. A., Zlnger, l., Poge, 8. u., 8onLa, !., Cendreau, ., & Cullen, l. 1. (1990). uoes
correcLlonal LreaLmenL work? A cllnlcally relevanL and psychologlcally lnformed meLa-analysls.
ctlmlooloqy,o28, 369-404.
Anglln, M. u., Longshore, u., & 1urner, S. (1999). 1reaLmenL AlLernaLlves Lo SLreeL Crlme (1ASC)X
An evaluaLlon of flve programs. ctlmlooloIostlceoooJo8ebovlot, 26(2), 168-193.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
60
Aos, S. (2009). cbooqloqotbeoseoteocloqoqtlJ?octeotloqoooctlmeolmpoctostotemeototooosslsto
Jeclslooomokloqo[owerolnL slldes from !une 2009 meeLlng]. 8eLrleved from
www.sgc.wa.gov/MlnuLes/hLm
Aos, S., Mlller, M., & urake, L. (2006a). vlJeoce-boseJooJoltocottectloosoptoqtoms.owbotowotkso
ooJowbotoJoesooot. Clympla, WAXWashlngLon SLaLe lnsLlLuLe for ubllc ollcy.
Aos, S., Mlller, M., & urake, L. (2006b). vlJeoce-boseJopobllcopollcyooptloosotooteJoceofototeo
ptlsooocoosttoctloo,octlmloolojostlceocosts,oooJoctlmeototes. Clympla, WAXWashlngLon SLaLe
lnsLlLuLe for ubllc ollcy.
Aos, S., hlpps, ., 8arnoskl, 8., & Lleb, 8. (2001). 1beocompototlveocostsoooJobeoefltsoofo
ptoqtomsotooteJoceoctlme. Clympla, WAXWashlngLon SLaLe lnsLlLuLe for ubllc ollcy.
8azemore, C., & Schlff, M. (2004). IoveolleojostlceotefotmoooJotestototlveojostlce.o8ollJloqo
tbeotyoooJopollcyoftomoptoctlce. orLland, C8XWlllan ubllshlng.
8onLa, !. (2007). Cffender assessmenLXCeneral lssues and conslderaLlons.
ln L. L. MoLluk and 8. C. Serln (Lds.), compeoJlomo2000ooooeffectlveocottectloool
ptoqtommloq.oCLLawa, CnXCorrecLlonal Servlce Canada. 8eLrleved from
hLLpX//www.csc-scc.gc.ca/LexL/rsrch/compendlum/2000/lndex-eng.shLml
8onLa, !., 8ugge, 1., ScoLL, 1., 8ourgon, C., & ?esslne, A. (2008). Lxplorlng Lhe black box of
communlLy supervlslon. IootooloofoOffeoJetokebobllltotloo,oX7(3), 248270.
8onLa, !., Wallace-CapreLLa, S., & 8ooney, 8. (2000). A quasl-experlmenLal evaluaLlon of an
lnLenslve rehablllLaLlon supervlslon program. ctlmlooloIostlceoooJo8ebovlot, 27(3), 312-329.
8onLa, !., Wallace-CapreLLa, S., 8ooney, !., & McAnoy, o. (2002). An ouLcome evaluaLlon of a
resLoraLlve [usLlce alLernaLlve Lo lncarceraLlon. Iostlceokevlew, 5(4), 319-338.
8ouffard, !. A., Lxum, M. L., & aLernosLer, 8. (2000). WhlLher Lhe beasL? 1he role of emoLlons
ln a raLlonal cholce Lheory of crlme. ln S. S. Slmpson (Ld.), OfoctlmeoooJoctlmloollty.ousloqotbeotyo
looevetyJoyollfe. 1housand Caks, CAXlne lorge ress.
8rlLL, C. L. lll., CoLLfredson, M. 8., & Coldkamp, !. S. (1992). urug LesLlng and preLrlal
mlsconducLXAn experlmenL on Lhe speclflc deLerrenL effecLs of drug monlLorlng defendanLs on
preLrlal release. IootooloofokeseotcbolooctlmeoooJouellopoeocy, 29(1), 62-78.
8ureau of !usLlce SLaLlsLlcs. (2003). !usLlce sysLem employmenL and payroll, Lable 1.17.2003
[ul documenL]. 5ootcebookoofoctlmlooloIostlceo5totlstlcsoOolloe. 8eLrleved from
hLLpX//www.albany.edu/sourcebook/pdf/L1172003.pdf
8ureau of !usLlce SLaLlsLlcs. (2007). lelooyoseoteocesoloostoteocootts,o200X. WashlngLon, uCX
u.S. ueparLmenL of !usLlce, Cfflce of !usLlce rograms.
8ureau of !usLlce SLaLlsLlcs. (2008). ctlmloolovlctlmlzotloo,o2007. WashlngLon, uCXu.S.
ueparLmenL of !usLlce, Cfflce of !usLlce rograms.
8ureau of !usLlce SLaLlsLlcs. (2009). lelooyoseoteocesoloostoteocootts,o2006-5totlstlcolotobles.
WashlngLon, uCXu.S. ueparLmenL of !usLlce, Cfflce of !usLlce rograms.
61
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
8urke, . 8. (2004). lotoleovlolotloosotevlslteJ.oAobooJbookoooostteoqtbeoloqopotoleoptoctlcesofoto
pobllcosofetyoooJosoccessfolottoosltloootootbeocommoolty. WashlngLon, uCXu.S. ueparLmenL of
!usLlce, naLlonal lnsLlLuLe of CorrecLlons. 8eLrleved from hLLpX//www.nlclc.org/llbrary/019833
Clark, S. P., & ourLz, S. 1. (1983). 1he lmporLance of lnLerlm declslons Lo felony Lrlal courL
dlsposlLlons. IootooloofoctlmloololowoooJoctlmlooloqy, 7X, 476-318.
Clear, 1. 8., & SumLer, M. 1. (2002). rlsoners, prlson, and rellglonX8ellglon and ad[usLmenL Lo
prlson. IootooloofoOffeoJetokebobllltotloo, J5, 127-139.
Cohen, 1. P., & 8eaves, 8. A. (2006). lelooyoJefeoJootsoloolotqeootbooocoootles,o2002.
WashlngLon, uCXu.S. ueparLmenL of !usLlce, 8ureau of !usLlce SLaLlsLlcs. 8eLrleved from
hLLpX//b[s.o[p.usdo[.gov/lndex.cfm?Ly=pbdeLall&lld=896
Cohen, 1. P., & 8eaves, 8. A. (2008). ltettloloteleoseoofofelooyoJefeoJootsoloostoteocootts.
WashlngLon, uCXu.S. ueparLmenL of !usLlce, 8ureau of !usLlce SLaLlsLlcs.
Colllns, !. C., & orras, !. l. (1997). 8olltotoolost.o5occessfolobobltsoofovlsloootyocompooles. new
?ork, n?XParperColllns.
CommlLLee on CuallLy PealLhcare ln Amerlcan, lnsLlLuLe of Medlclne. (2000). 1ooettolsobomoo.o
8ollJloqooosofetobeoltbosystem L. 1. oohn, !. M. Corrlgan, & M. S. uonaldson, (Lds.). WashlngLon,
uCX1he naLlonal Academles ress.
1he CommonwealLh lund. (2006). 5ovloqollves,otolsloqobopes. 8eLrleved from
hLLpX//www.commonwealLhfund.org/ConLenL/lrom-Lhe-resldenL/2006/Savlng-Llves--8alslng-
Popes.aspx
Cullen, l. 1., & Cendreau, . (2000). Assesslng correcLlonal rehablllLaLlonXollcy, pracLlce, and
prospecLs. ln !. Porney (Ld.), ctlmloolojostlceo2000.olollcles,optocesses,oooJoJeclsloosoofotbeo
ctlmloolojostlceosystem. WashlngLon, uCXu.S. ueparLmenL of !usLlce, naLlonal lnsLlLuLe of
!usLlce.
uemuLh, S. (2003). 8aclal and eLhnlc dlfferences ln preLrlal release declslons and ouLcomesX
A comparlson of Plspanlc, 8lack, and WhlLe felony arresLees. ctlmlooloqy, X1, 873-907.
uowden, C. (1998). Aometo-ooolytlcoexomlootloooofotbeotlsk,ooeeJoooJotespooslvltyoptloclpleso
ooJotbeltolmpottooceowltblootbeotebobllltotloooJebote. unpubllshed masLer's Lhesls. CLLawa,
CnXCarleLon unlverslLy, ueparLmenL of sychology.
uowden, C., & Andrews, u. A. (2004). 1he lmporLance of sLaff pracLlce ln dellverlng effecLlve
correcLlonal LreaLmenLXA meLa-analyLlc revlew of core correcLlonal pracLlce. lotetootlooolo
IootooloofoOffeoJeto1betopyoooJocompototlveoctlmlooloqy, X8(2), 203-214.
urake, L. o., & 8arnoskl, 8. (2008). locteosloqoeotoeJoteleoseoftomoptlsoo.olmpoctsoofo200Jolowo
oooteclJlvlsmoooJoctlmloolojostlceocosts. Clympla, WAXWashlngLon SLaLe lnsLlLuLe for ubllc
ollcy.
Lgelko, S., CalanLer, M., uermaLls, P., & ueMalo, C. (1998). LvaluaLlon of a mulLlsysLems model
for LreaLlng perlnaLal cocalne addlcLlon. Iootooloofo5obstooceoAboseo1teotmeot, 15(3), 231-239.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
62
Lmrlck, C. u., 1onlgang, !. S., MonLgomery, P., & LlLLle, L. (1993). AlcobollcsoAoooymoos.o
OppottooltlesoooJooltetootlves. new 8runswlck, n!XAlcohol 8esearch uocumenLaLlon, lnc.,
8uLgers CenLer of Alcohol SLudles.
Lxum, M. L. (2002). 1he appllcaLlon and robusLness of Lhe raLlonal cholce perspecLlve ln Lhe
sLudy of lnLoxlcaLed/angry lnLenLlons Lo aggress. ctlmlooloqy, X0, 933-966.
lederal 8ureau of lnvesLlgaLlon. (2002). Crlme ln Lhe unlLed SLaLes, 2002. uolfotmoctlmeo
kepotts. WashlngLon, uCXAuLhor. 8eLrleved from hLLpX//www.fbl.gov/ucr/02clus.hLm
lrench, S., & Cendreau, . (2003). 5ofeoooJobomooeocottectloosotbtooqboeffectlveo
tteotmeoto[ul documenL]. CLLawa, CnXCorrecLlonal Servlce of Canada. 8eLrleved
from hLLpX//www.csc-scc.gc.ca/LexL/rsrch/reporLs/r139/r139_e.pdf
CalanLer, M. (1993). NetwotkotbetopyofotoolcoboloooJoJtoqoobose.oAooewoopptoocbolooptoctlce.
new ?ork, n?X8aslc 8ooks.
Cendreau, ., & Coggln, C. (1996). rlnclples of effecLlve programmlng wlLh offenders.
lotomoooocottectloosokeseotcb, 8(3), 38-40.
Cendreau, ., Coggln, C., & Cullen, l. (1999). 1beoeffectsoofoptlsoooseoteocesooooteclJlvlsm.
CLLawa, CnXSollclLor Ceneral of Canada.
Cendreau, ., Coggln, C., Cullen, l. 1., & Andrews, u. (2001). 1he effecLs of communlLy
sancLlons and lncarceraLlon on recldlvlsm. ln L. L. MoLluk and 8. C. Serln (Lds.), compeoJlomo
2000ooooeffectlveocottectloooloptoqtommloq.ovolomeo1. CLLawa, CnXCorrecLlonal Servlce of
Canada.
Cendreau, ., Coggln, C., & LlLLle, 1. (1996). lteJlctloqooJoltooffeoJetoteclJlvlsm.owbotowotks!o
(1996-07). CLLawa, CnXSollclLor Ceneral of Canada.
Cendreau, ., LlLLle, 1., & Coggln, C. (1996). A meLa-analysls of adulL offender recldlvlsmX
WhaL works? ctlmlooloqy,oJX, 373-607.
Coldkamp, !. S. (1979). 1wooclossesoofooccoseJ.oAostoJyoofobolloooJoJeteotlooolooAmetlco.
Cambrldge, MAX8alllnger.
Coldkamp, !. S., & CoLLfredson, M. (1983). ColJelloesofotoboll.o1beolblloJelpblooexpetlmeot.
hlladelphla, AX1emple unlverslLy ress.
Coldkamp, !. S., & WhlLe, M. u. (2006). 8esLorlng accounLablllLy ln preLrlal releaseX1he
hlladelphla preLrlal release supervlslon experlmenLs. Iootooloofoxpetlmeotoloctlmlooloqy,o
2(2), 143-181.
Crasmack, P. C., & 8ry[ak, C. !. (1980). 1he deLerrenL effecL of percelved severlLy of
punlshmenL. 5oclololotces, 39, 471-491.
Cray, C. (2002). tblcsoooJojoJqesoevolvloqotolesooffotbeobeocb.o5etvloqooooqovetomeotolo
commlssloos. ues Molnes, lAXAmerlcan !udlcaLure SocleLy.
63
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Crove, W. M., & Meehl, . L. (1996). ComparaLlve efflclency of lnformal (sub[ecLlve,
lmpresslonlsLlc) and formal (mechanlcal, algorlLhmlc) predlcLlon proceduresX1he cllnlcal-
sLaLlsLlcal conLroversy. lsycboloqy,olobllcolollcyoooJolow, 2, 293-323.
Crove, W. M., Zald, u. P., Lebow, 8. S., SnlLz, 8. L., & nelson, C. (2000). Cllnlcal versus
mechanlcal predlcLlonXA meLa-analysls. lsycboloqlcoloAssessmeot, 12, 19-30.
Parrell, A., 8oman, !., 8haLl, A., & arLhasaraLhy, 8. (2003). 1beolmpoctoevolootloooofotbeo
MotylooJo8teokotbeocycleololtlotlve. WashlngLon, uCX1he urban lnsLlLuLe.
Parrls, M.o. (1986). 1beoqoolsoofocommooltyosooctlooso[ul documenL]. WashlngLon, uCX
u.S. ueparLmenL of !usLlce, naLlonal lnsLlLuLe of CorrecLlons. 8eLrleved from
hLLpX//www.nlclc.org/pubs/pre/004369.pdf
Parrls, . M. (2006). WhaL communlLy supervlslon offlcers need Lo know abouL acLuarlal
rlsk assessmenL and cllnlcal [udgmenL. leJetololtobotloo,o70(2). 8eLrleved from
hLLpX//www.uscourLs.gov/fedprob/SepLember_2006/assessmenL.hLml
Pay, C. (2001). An exploraLory LesL of 8ralLhwalLe's relnLegraLlve shamlng Lheory. Iootooloofo
keseotcbolooctlmeoooJouellopoeocy,oJ8(2), 132-133.
Peckscher, C., & Adler, . S. (Lds.). (2006). 1beofltmoosooocollobototlveocommoolty.o
kecoosttoctloqottostolootbeokoowleJqeoecooomy. Cxford, uoXCxford unlverslLy ress.
Penggeler, S. W., Schoenwald, S. o., 8orduln, C. M., 8owland, M. u., & Cunnlngham, . 8.
(1998). MoltlsystemlcotteotmeotoofoootlsoclolobebovlotoloocbllJteooooJooJolesceots. new ?ork,
n?XCullford.
Plgglns, P., & Sllverman, o. (1999). Motlvotloqobebovlotocbooqeoomooqollllclt-Jtoqoobosets.o
keseotcboooocootloqeocyomoooqemeotolotetveotloos. WashlngLon, uCXAmerlcan sychologlcal
AssoclaLlon.
Pughes, 1., & Wllson, u. !. (2003). keeottyotteoJsolootbeouolteJo5totes.olomotesotetotoloqotootbeo
commooltyooftetosetvloqotlmeolooptlsoo. WashlngLon, uCXu.S. ueparLmenL of !usLlce, 8ureau of
!usLlce SLaLlsLlcs.
Pughes, 1., Wllson, u., & 8eck, A. (2001). 1teoJsoloostoteopotole,o1990-2000. WashlngLon, uCX
u.S. ueparLmenL of !usLlce, 8ureau of !usLlce SLaLlsLlcs.
LalounLaln, 8., Schaufer, 8., SLrlckland, S., 8afery, W., 8romage, C., Lee, C., & Clbson, S. (2008).
xomloloqotbeowotkoofostoteocootts,o2007. Wllllamsburg, vAXnaLlonal CenLer for SLaLe CourLs.
Landenberger, n. A., & Llpsey, M. W. (2003). 1he poslLlve effecLs of cognlLlve-behavloral
programs for offendersXA meLa-analysls of facLors assoclaLed wlLh effecLlve LreaLmenL. Iootoolo
ofoxpetlmeotoloctlmlooloqy,o1, 431-476.
Langan, ., & Levln, u. (2002). keclJlvlsmoofoptlsooetsoteleoseJoloo199X. WashlngLon, uCX
u.S. ueparLmenL of !usLlce, 8ureau of !usLlce SLaLlsLlcs.
Larson, C., & LalasLo, l. (1989). 1eomwotk.owbotomostoqootlqbt/wbotocoooqoowtooq. newbury
ark, CAXSage ubllcaLlons.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
64
LaLessa, L., Cullen, l. 1., & Cendreau, . (2002). 8eyond professlonal quackeryXrofesslonallsm
and Lhe posslblllLy of effecLlve LreaLmenL. leJetololtobotloo, 66(2), 43-49.
Lewls, M. (2003). Mooeyboll.o1beoottoofowlooloqoooooofoltoqome. new ?ork, n?XW.W. norLon &
Company, lnc.
Llnd, L. A., & 1yler, 1. 8. (1988). 1beosoclolopsycboloqyoofoptoceJotolojostlce. new ?ork, n?X
Sprlnger.
Llpsey, M. W., & Cullen, l. 1. (2007). 1he effecLlveness of correcLlonal rehablllLaLlonXA revlew of
sysLemaLlc revlews. AoooolokevlewoofolowoooJo5oclolo5cleoce,oJ, 297-320.
Llpsey, M. W., & Landenberger, n. (2006). CognlLlve-behavloral lnLervenLlons. ln 8. Welsh and
u. . larrlngLon (Lds.), lteveotloqoctlme.owbotowotksofotocbllJteo,ooffeoJets,ovlctlms,oooJo
ploces. uordrechL, neLherlandsXSprlnger.
Llpsey, M. W., Landenberger, n.A., & Wllson, S. !. (2007). ffectsoofocoqoltlve-bebovlotolo
ptoqtomsofotoctlmloolooffeoJets (Campbell CollaboraLlon SysLemaLlc 8evlews 2007-06)
[ul documenL]. 8eLrleved from www.campbellcollaboraLlon.org/llb/download/143/
Lowenkamp, C., & LaLessa, L. (2004). undersLandlng Lhe rlsk prlnclpleXPow and why
correcLlonal lnLervenLlons can harm low-rlsk offenders. 1oplcsoloocommooltyocottectloos,
Aoooololssoe 200X.oAssessmeotolssoesofotoMoooqets. WashlngLon, uCXu.S. ueparLmenL of
!usLlce, naLlonal lnsLlLuLe of CorrecLlons.
Lowenkamp C. 1., LaLessa L. !., & Polslnger, A. M. (2006). 1he rlsk prlnclple ln acLlonXWhaL have
we learned from 13,676 offenders and 97 correcLlonal programs? ctlmeoooJouellopoeocy,o52,
77-93.
Lowenkamp, C. 1., LaLessa, L. !., & SmlLh, . (2006). uoes correcLlonal program quallLy really
maLLer? 1he lmpacL of adherlng Lo Lhe prlnclples of correcLlonal lnLervenLlon. ctlmlooloqyoooJo
lobllcolollcy,o5,o201-220.
Lowenkamp, C., ealer, !., SmlLh, ., & LaLessa, L. (2007). Adherlng Lo Lhe rlsk and need
prlnclplesXuoes lL maLLer for supervlslon-based programs? leJetololtobotloo,o70(3). 8eLrleved
from hLLpX//www.uscourLs.gov/fedprob/uecember_2006/adherlng.hLml
Macuuffle, !. ., & Pelper, S. (2006). CollaboraLlon ln supply chalnsXWlLh and wlLhouL LrusL. ln
C. Peckscher and . S. Adler (Lds.), 1beofltmoosooocollobototlveocommoolty.okecoosttoctloqottosto
lootbeokoowleJqeoecooomy (pp. 417-466). new ?ork, n?XCxford unlverslLy ress.
Macoenzle, u. L. (1997). Crlmlnal [usLlce and crlme prevenLlon. ln L. W. Sherman, u.
CoLLfredson, u. Macoenzle, !. Lck, . 8euLer, & S. 8ushway (Lds.), wbotowotks,owbotoJoesot,o
wbotsoptomlsloq!oAotepottotootbeouolteJo5totesocooqtess. WashlngLon, uCXu.S. ueparLmenL of
!usLlce, naLlonal lnsLlLuLe of !usLlce.
Macoenzle, u. L., 8rame, 8., Mcuowall, u., & Souryal, C. (1993). 8ooL camp prlsons and
recldlvlsm ln elghL sLaLes. ctlmlooloqy, JJ, 327-337.
Macoenzle, u. L., Wllson, u. 8., & older, S.8. (2001). LffecLs of correcLlonal booLcamps on
offendlng. 1beoAooolsoofotbeoAmetlcoooAcoJemyoofololltlcoloooJo5oclolo5cleoce, 578, 126-143.
63
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
Marlowe, u. 8., lesLlnger, u. S., Arabla, . L., uugosh, o. L., 8enasuLLl, o. M., CrofL, !. 8., &
Mcoay, !. 8. (2008). AdapLlve lnLervenLlons ln drug courLXA plloL experlmenL. ctlmlooloIostlceo
kevlew,oJJ, 343-360.
Marlowe, u. 8., lesLlnger, u. S., Lee, . A., uugosh, o. L., & 8enasuLLl, o. M. (2006). MaLchlng
[udlclal supervlslon Lo cllenLs' rlsk sLaLus ln drug courL. ctlmeoooJouellopoeocy,o52, 32-76.
MaLsueda, 8. L., oreager, u. A., & Pulzlnga, u. (2006). ueLerrlng dellnquenLsXA raLlonal cholce
model of LhefL and vlolence. Ametlcooo5ocloloqlcolokevlew, 71, 93-122.
McCulre, !. (2001). WhaL works ln correcLlonal lnLervenLlon? Lvldence and pracLlcal
lmpllcaLlons. ln C. 8ernfeld, u. . larrlngLon, & A. Lescheld (Lds.), OffeoJetotebobllltotloooloo
ptoctlce.olmplemeotloqoooJoevolootloqoeffectlveoptoqtoms. ChlchesLer, uoX!ohn Wlley & Sons.
McCulre, !. (2002). lnLegraLlng flndlngs from research revlews. ln !. McCulre (Ld.), OffeoJeto
tebobllltotloooooJotteotmeot.offectlveoptoctlceoooJopollclesotooteJoceote-offeoJloq. ChlchesLer,
uoX!ohn Wlley & Sons.
Meyers, 8. !, Mlller, W. 8., SmlLh, !. L., & 1onnlgan, S. (2002). A randomlzed Lrlal of Lwo
meLhods for engaglng LreaLmenL refuslng drug users Lhrough concerned slgnlflcanL oLhers.
IootooloofocoosoltloqoooJocllolcololsycboloqy, 70(3), 1182-1183.
Meyers, 8. !., & SmlLh, !. L. (1997). CeLLlng off Lhe fenceXrocedures Lo engage LreaLmenL-
reslsLanL drlnkers. Iootooloofo5obstooceoAboseo1teotmeot, 1X, 467-472.
nagln, u. S. (1998). Crlmlnal deLerrence research aL Lhe ouLseL of Lhe LwenLy-flrsL cenLury.
ln M. 1onry's (Ld.), ctlmeoooJoIostlce.oAokevlewoofokeseotcb, 2J, 1-42.
naLlonal AssoclaLlon of Crlmlnal uefense Lawyers. (2009). Mlootoctlmes,omosslveowoste.o
1beotettlbleotolloofoAmetlcosobtokeoomlsJemeoootocootts. WashlngLon, uCXAuLhor.
naLlonal 8esearch Councll, CommlLLee on CommunlLy Supervlslon and ueslsLance from Crlme.
(2007). lotole,oJeslstooceoftom ctlme,oooJocommooltyoloteqtotloo. WashlngLon, uCX1he
naLlonal Academles ress.
nlchols, !., & 8oss, P. L. (1990). LffecLlveness of legal sancLlons ln deallng wlLh drlnklng drlvers.
Alcobol,outoqs,oooJoutlvloqo6(2), 33-60.
C'Connor, 1., & erryclear, M. (2002). rlson rellglon ln acLlon and lLs lnfluence on offender
rehablllLaLlon.oIootooloofoOffeoJetokebobllltotloo, J5(3/4), 11-33.
Csborne, u., & Caebler, 1. (1992). keloveotloqoqovetomeot.onowotbeoeottepteoeotlolospltltolso
ttoosfotmloqotbeopobllcosectot. 8eadlng, MAXAddlson-Wesley.
aLernosLer, 8. (1989). ueclslons Lo parLlclpaLe ln and deslsL from four Lypes of common
dellnquencyXueLerrence and Lhe raLlonal cholce perspecLlve. lowoooJo5ocletyokevlew,
23(1), 7-40.
eLers, 1. !., & AusLln, n. (1986). Aoposslooofotoexcelleoce.o1beoleoJetsblpoJlffeteoce. new ?ork,
n?XWarner 8ooks.
eLers, 1. !., & WaLerman, 8. P. (2004). looseotcboofoexcelleoce.olessoosoftomoAmetlcosobestotooo
compooles. new ?ork, n?XParperColllns 8uslness.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
66
eLersllla, !. (1999). A decade of experlmenLlng wlLh lnLermedlaLe sancLlonsXWhaL have we
learned? letspectlves, 2J(1), 39-44.
eLersllla, !. (2004). WhaL works ln prlsoner reenLry? 8evlewlng and quesLlonlng Lhe evldence.
leJetololtobotloo, 68(2), 3-9.
eLersllla, !. (2007). Lmploy behavloral conLracLlng for 'earned dlscharge' parole. ctlmlooloqyo
ooJolobllcolollcy, 6(4), 807-814.
eLersllla, !., & 1urner, S. (1993). lnLenslve probaLlon and parole. ln M. 1onry (Ld.), ctlmeoooJo
jostlce.oAotevlewoofotbeoteseotcb,ovolomeo17. Chlcago, lLXunlverslLy of Chlcago ress.
lquero, A. 8., & ogarsky, C. (2002). 8eyond SLafford and Warr's reconcepLuallzaLlon of
deLerrenceXersonal and vlcarlous experlences, lmpulslvlLy, and offendlng behavlor. Iootooloofo
keseotcbolooctlmeoooJouellopoeocy, J9(2), 133-186.
ogarsky, C. (2002). ldenLlfylng deLerrable offendersXlmpllcaLlons for deLerrence research.
IostlceoOoottetly, 19(3), 431-432.
ogarskl, C. (2007). ueLerrence and lndlvldual dlfferences among convlcLed offenders. Iootoolo
ofoOoootltotlveoctlmlooloqy, 2J(1), 39-74.
orLer, M. L. (1983). competltlveooJvootoqe. new ?ork, n?Xlree ress.
8hlne, L. (1993). keclolmloqooffeoJetooccoootoblllty.olotetmeJloteosooctloosofotoptobotloooooJo
potoleovlolotots. Laurel, MuXAmerlcan CorrecLlonal AssoclaLlon.
8ogers, 8. W., Welllns, 8. S., & Conner, u. 8. (2003). 8ollJloqocompetltlveooJvootoqeobyo
moxlmlzloqobomoootesootceololtlotlveso[ul documenL]. lLLsburgh, AXuevelopmenL
ulmenslons lnLernaLlonal. 8eLrleved from
hLLpX//www.ddlworld.com/pdf/ddl_reallzaLlon_whlLepaper.pdf
Senge, . (2006). 1beoflftboJlsclplloe.o1beoottoooJoptoctlceoofotbeoleotoloqootqoolzotlooo(2nd ed.).
new ?ork, n?Xuoubleday.
Shaplro, C., & SchwarLz, M. (2001). Comlng homeX8ulldlng on famlly connecLlons.
cottectloosoMoooqemeotoOoottetly,o5(3), 32-61.
Sheppard, !. M., & Paas, S. (2003). coopetotlooottockloqostoJy.oAptllo200JoopJote. ClnclnnaLl,
ChloX1he Councll for MarkeLlng and Cplnlon 8esearch (CMC8).
SmlLh, ., Cendreau, ., & SwarLz, o. (2009). valldaLlng Lhe prlnclples of effecLlve lnLervenLlonX
A sysLemaLlc revlew of Lhe conLrlbuLlons of meLa-analysls ln Lhe fleld of correcLlons. vlctlmsoooJo
OffeoJets, X, 148-169.
SmlLh, ., Coggln, C., & Cendreau, . (2002). 1beoeffectsoofoptlsoooseoteocesoooJolotetmeJloteo
sooctloosooooteclJlvlsm.oCeoetoloeffectsoooJoloJlvlJooloJlffeteoceso[ul documenL]. 8eLrleved
from hLLpX//ww2.ps-sp.gc.ca/publlcaLlons/correcLlons/200201_Cendreau_e.pdf
Spelman, W. (1994). ctlmloololocopocltotloo.onew ?ork, n?Xlenum ress.
67
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
SLafford, M. C., & Warr, M. (1993). A reconcepLuallzaLlon of general and speclflc deLerrence.
IootooloofokeseotcbolooctlmeoooJouellopoeocy, J0, 123-133.
SLanford CraduaLe School of 8uslness. (2008). lostltoteofotoneoltbcoteolmptovemeot.o
1beocompolqootoosoveo100,000ollves (CaseXL-13, !anuary 21, 2008).
SLroker, 8. (2006). llveoteosoosowbyojoJqesosboolJobecomeomoteolovolveJolooestobllsbloq,o
leoJloq,oooJopottlclpotloqoooocollobototlve,opollcy-focoseJoteoms. Sllver Sprlng, MarylandX
CenLer for LffecLlve ubllc ollcy.
1axman, l., Soule, u., & Celb, A. (1999). CraduaLed sancLlonsXSLepplng lnLo accounLable
sysLems and offenders. ltlsoooIootool, 79(2), 182-204.
1aylor, S., & MarLln, C. (2006). SLaLe and local agencles parLner Lo manage vlolaLlons of
supervlslon ln Cregon. 1oplcsoloocommooltyocottectloos.offectlvelyoMoooqloqovlolotlooso
ooJokevocotloos,oAoooololssoe,o2006. WashlngLon, uCXu.S. ueparLmenL of !usLlce, naLlonal
lnsLlLuLe of CorrecLlons.
1onry, M. (1996). 5eoteocloqomottets.onew ?ork, n?XCxford unlverslLy ress.
1onry, M. (1997). lotetmeJloteosooctloosolooseoteocloqoqolJelloes. WashlngLon, uCX
u.S ueparLmenL of !usLlce, naLlonal lnsLlLuLe of !usLlce.
1yler, 1. 8. (2000). Soclal [usLlceXCuLcome and procedure. lotetootloooloIootooloofo
lsycboloqy, J5(2), 117-123.
1yler, 1. 8. (2007). rocedural [usLlce and Lhe courLs. coottokevlew, XX, 26-31.
1yler, 1. 8., & Puo, ?. !. (2002). 1tostolootbeolow.oocootoqloqopobllcocoopetotlooowltbotbeo
pollceoooJocootts.onew ?ork, n?X8ussell Sage loundaLlon.
umbrelL, M. S. (1998). 8esLoraLlve [usLlce Lhrough vlcLlm-offender medlaLlonXA mulLl-slLe
assessmenL. westetooctlmlooloqyokevlew, 1(1). 8eLrleved from
hLLpX//wcr.sonoma.edu/v1n1/umbrelL.hLml
vannosLrand, M. (2003). AssessloqotlskoomooqoptettloloJefeoJootsoloovltqlolo.o1beovltqloloo
ltettloloklskoAssessmeotolosttomeot. 8lchmond, vAXvlrglnla ueparLmenL of Crlmlnal !usLlce
Servlces.
vannosLrand, M., & oeebler, C. (2009). ltettlolotlskoossessmeotolootbeoleJetolocoottofototbeo
potposeoofoexpooJloqotbeooseoofooltetootlvesotooJeteotloo. WashlngLon, uCXu.S. ueparLmenL of
!usLlce, Cfflce of Lhe lederal ueLenLlon 1rusLee. 8eLrleved from
hLLpX//www.nlclc.org/Llbrary/023738
von Plrsch, A. (1993). ceosoteoooJosooctloos. Cxford, uoXCxford unlverslLy ress.
Warren, 8. (2007). vlJeoce-boseJoptoctlceotooteJoceoteclJlvlsm.olmpllcotloosofotostoteo
joJlclotles. WashlngLon, uCXu.S. ueparLmenL of !usLlce, naLlonal lnsLlLuLe of CorrecLlons.
Wllson, u., 8ouffard, L. A., & Macoenzle, u. L. (2003). CuanLlLaLlve revlew of sLrucLured, group-
orlenLed, cognlLlve-behavloral programs for offenders. ctlmlooloIostlceoooJo8ebovlot, J2(2),
172-204.
A

F
r
a
m
e
w
!
r
k

f
!
r

E
v
i
d
e
n
c
e
-
B
a
s
e
d

D
e
c
i
s
i
!
n

M
a
k
i
n
g

i
n

L
!
c
a
l

C
r
i
m
i
n
a
l

J
u
s
t
i
c
e

S
y
s
t
e
m
s

|

3
r
d

E
d
i
t
i
!
n

|

0
4
/
1
6
/
2
0
1
0
68
Wllson, u. 8., Macoenzle, u. L, & MlLchell, l. n. (2003). ffectsoofocottectlooolobootocompso
ooooffeoJloqo(Campbell SysLemaLlc 8evlews 2003X1) [ul documenL]. 8eLrleved from
www.campbellcollaboraLlon.org/llb/download/3/
Wllson, u, MlLchell, C., & Macoenzle, u. (2006). SysLemaLlc revlew of drug courL effecLs on
recldlvlsm. Iootooloofoxpetlmeotoloctlmlooloqy, 2(4), 439-487.
WlnLerfleld, L., Coggeshall, M., & Parrell, A. (2003). uevelopmeotoofooooempltlcolly-boseJotlsko
ossessmeotolosttomeot. WashlngLon, uCX1he urban lnsLlLuLe.
WSl. (2004). CuLcome evaluaLlon of WashlngLon SLaLe's research-based programs for
[uvenlle offenders. Clympla, WAXWashlngLon SLaLe lnsLlLuLe for ubllc ollcy.

Vous aimerez peut-être aussi