Smart Policing in Action 1: Findings and Accomplishments from the - - PowerPoint PPT Presentation

1 This project was supported by Grant No. 2013-DP-BX-K006 awarded by the Bureau of Justice Assistance. The Bureau of Justice Assistance is a component of the Office of Justice Programs, which also includes the Bureau of Justice Statistics, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, the Office for Victims of Crime, and the SMART Office. Points of view or opinions in this document are those of the author and do not necessarily represent the official position or policies of the U.S. Department of Justice.

Phoenix SPI, Port St. Lucie SPI, BJA, and CNA March 4, 2015

Smart Policing in Action 1: Findings and Accomplishments from the Smart Policing Initiative (SPI)

2 This project was supported by Grant No. 2011-DB-BX-0018 awarded by the Bureau of Justice Assistance. The Bureau of Justice Assistance is a component of the Office of Justice Programs, which also includes the Bureau of Justice Statistics, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, the Office for Victims of Crime, and the SMART Office. Points of view or opinions in this document are those of the author and do not necessarily represent the official position or policies of the U.S. Department of Justice.

Charles Katz, Mike Kurtenbach, David Choate, Justin Ready

March 4, 2015

Evaluating the Impact of Officer Worn Body Cameras in the Phoenix Police Department

3

City Manager Task Force

• Created in April 2010 to address residents’

concerns about Phoenix Police Department interactions with the community

• Developed 34 recommendations designed to

increase community access to, communication with, and confidence in the Phoenix Police Department

• One recommendation called for a pilot program

involving the deployment of dashboard cameras

4

Targeted Problems

• Violence in general has declined in Phoenix, but

domestic violence has remained problematic

– Approximately 40,000 incidents of domestic violence are dispatched per year – Domestic violence is one of the top five call types

• Shift in relationship with residents

– Police community relations are complex in some communities – High-profile events involving police-resident encounters have and continue to occur in these same communities

5

The Technology

• Selected Vievu

– Self-contained device worn on the torso

• Size of a pager

– Docking station – Uploaded to Phoenix Police Department servers

6

Quasi-Experimental Design

• Repeated measures

from the sources below

– Police/court data – Administrative records – Officer self-report surveys – Meta-data from cameras – Interviews with

• fficers

7

October December January March April July October June 2012 2013 2014 Area 82 17.4 10.8 32.4 31.4 75.7 75.7 66.7 61.8 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

Percent Agree

Equipment Is Easy to Use

Implementation Date

Key takeaway: After implementation, officers found the cameras easier to use than they expected.

8

October December January March April July October June 2012 2013 2014 Area 82 11.4 9.1 8.9 2.3 5.3 0.0 0.0 2.9 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

Percent Agree

Incident Reports: Less Time Spent On Paperwork

Implementation Date

Key takeaway: Camera implementation did not decrease the time officers spent on paperwork.

9

October December January March April July October June 2012 2013 2014 Area 82 20.0 18.2 34.4 26.5 21.6 18.4 27.0 23.5 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

Percent Agree

Easy to Download Data

Implementation Date

Key takeaway: After implementation, officers were slightly less likely to agree that downloading data was easy.

10

October December January March April July October June 2012 2013 2014 Area 82 24.2 22.7 15.9 19.0 17.9 8.6 8.1 32.4 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

Percent Agree

Cameras Should Be Expanded to Other Departments

Implementation Date

Key takeaway: Officers were more likely to agree camera use should be expanded into other departments after several months of implementation.

11

Productivity: Mean Numbers of Arrests

0.08 0.11 0.12 0.12 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Target Comparison Pre-test Post-test

% change 42.6 % change 14.9

12

Percentage Change in Complaints Before and After Body Worn Cameras

• 22.5

10.6 45.1

• 30
• 20
• 10

10 20 30 40 50 Target Comprison City wide

13

Percentage of Complaints That Are Unfounded

45.0 74.2 59.1 82.2 55.0 56.5 10 20 30 40 50 60 70 80 90 Target Comparison City wide % change 64.9 % change 39.1 % change 3.5

14

Use of Body Camera Evidence in Court for Domestic Violence Offenses

• Investigator use
• Evidence storage (information technology)
• Prosecutor tracking and review
• Court liaison officer

15

Domestic Violence (DV) Case Flow Pre & Post Camera Deployment

Pre-Test Case Post-Test Comparison Post-Test Camera n % n % n % Number of DV-Related Contacts a 878 100.0 933 100.0 252 100.0 Cases Initiated 369 42.0 320 34.3 103 40.9 Charges Filed 333 37.9 243 26.0 90 37.7 Case Furthered (Not Dismissed) 131 14.9 58 6.2 32 12.7 Plead Guilty 27 3.1 11 1.2 11 4.4 Guilty at Trial 25 2.8 9 0.9 11 4.4

a The number of contacts is derived from the DV pocket cards, which included data on 2,063 unique incidents

from January 1, 2012, through July 31, 2014, from the Maryvale Precinct.

16

Case Processing Time

Number of Days to Process Case to Disposition (N=795) † Pre-Test Case Post-Test Comparison Post-Test Camera mean SD n mean SD n mean SD n All Completed Cases * 95.8 124.30 340 43.5 77.50 266 78.1 105.10 92 Dismissed * 65.3 91.00 201 38.2 67.80 184 56.1 65.90 58 Plead Guilty * 167.7 157.57 104 71.3 100.44 46 131.9 156.40 21 Trial 74.4 90.61 27 114.2 125.06 11 105.5 126.07 11 * Significant at p < 0.05 † Original values ranged from 0 to 756. Values above the 98th percentile of 438 days (n=16) were truncated to 438 to control for outlier cases.

17

October December January March April July October June 2012 2013 2014 Area 82 52.8 38.6 35.6 55.8 40.5 8.3 14.3 32.4 10 20 30 40 50 60 70 80 90 100

Percent Agree

Easier To Prosecute DV Offenders

Implementation Date

18

Conclusions

• Decrease in complaints
• Increase in unfounded incidents
• Increase in arrests (+/-)
• Prosecution of domestic violence

Strengths

• Officer resistance
• Information technology costs
• Increase time spent on officer paper work
• Prosecutor capacity
• Redaction

Challenges

E xpe rime nta l T e st o f Offe nde r-Ba se d Po lic e Re spo nse in L

• ng -T

e rm Pro pe rty Crime Ho t Spo ts

Po rt St. L uc ie , F L Po lic e De pa rtme nt F unde d b y Bure a u o f Justic e Assista nc e , Sma rt Po lic ing I nitia tive

Pre se nte d b y:

• Dr. Ra c he l Sa nto s, Asso c ia te Pro fe sso r

Sc ho o l o f Crimino lo g y a nd Crimina l Justic e F lo rida Atla ntic Unive rsity L

• t. Ro b e rto Sa nto s, PhD

Pro fe ssio na l Sta nda rds Po rt St. L uc ie , F L Po lic e De pa rtme nt

T his pro je c t wa s suppo rte d b y Gra nt No . 2012-DB-BX-0002 a wa rde d b y the Bure a u o f Justic e Assista nc e . T he Bure a u o f Justic e Assista nc e is a c o mpo ne nt

• f the Offic e o f Justic e Pro g ra ms, whic h also inc lude s the Bure a u o f Justic e Sta tistic s, the Na tio nal I

nstitute o f Justic e , the Offic e o f Juve nile Justic e a nd De linq ue nc y Pre ve ntio n, the Offic e fo r Vic tims o f Crime , a nd the SMART Offic e . Po ints o f vie w o r o pinio ns in this do c ume nt a re tho se o f the a utho r a nd do no t ne c e ssa rily re pre se nt the o ffic ia l po sitio n o r po lic ie s o f the U.S. De pa rtme nt o f Justic e .

Pro je c t Ove rvie w

• 2012-2014 Sma rt Po lic e I

nitia tive (SPI ) g ra nt fro m Bure a u o f Justic e Assista nc e (BJA)

• E

xpa nd Stra tifie d Po lic ing imple me nta tio n into the a g e nc y to o ffe nde r-b a se d stra te g ie s

• Co nduc ting e vide nc e -b a se d re se a rc h: Hig h

q ua lity re se a rc h to te st e ffe c tive ne ss o f a g ive n stra te g y

• SPI

T e a m

K e y Hypo the sis

• I

f the o ffe nde r-b a se d re spo nse s a re imple me nte d fo r

• ffe nde rs who live in hig h c rime a re a s,
• T

he re will b e a re duc tio n o f c rime in tho se a re a s

• Sinc e the o ffe nde rs a re like ly c o mmitting so me o f

the ir c rime s ne a r whe re the y live .

Pro je c t Ove rvie w

• Ra ndo mize d c o ntro lle d tria l (RCT

)

• Offe nde r-b a se d stra te g ie s imple me nte d in lo ng -te rm

ho t spo ts o f pro pe rty c rime

• Applic a tio n to o the r a g e nc ie s:

~Use simple a nd re a listic a na lysis o f o ffe nde rs ~Re spo nse s a lso re a listic fo r sub urb a n a re a s with a ve ra g e le ve ls

• f c rime a nd “a ve ra g e o ffe nde rs”

Blo c k Ra ndo miza tio n

• Blo c ke d o n o ffe nde rs pe r c rime in e a c h ho t spo t
• Ma rc h 2012 – Ja nua ry 2013
• Offe nde r da ta

~Only tho se living in PSL ~Arre ste d fo r b urg la ry in PSL

, St. L uc ie Co unty, F

• rt Pie rc e , a nd

Ma rtin Co unty in pa st two ye a rs

~Adults a nd juve nile s c urre ntly o n fe lo ny pro b a tio n

• Crime type s

~Re side ntia l b urg la ry ~Re side ntia l the ft fro m ve hic le

Ho t spo ts (48) 24 T re a tme nt a nd 24 Co ntro l (df=48-6-1=41)

L

• w o ffe nde r/ c rime (12)

6 tre a tme nt 6 c o ntro l Me dium o ffe nde r/ c rime (26) 13 tre a tme nt 13 c o ntro l Hig h o ffe nde r/ c rime (10) 5 tre a tme nt 5 c o ntro l

Blo c k Ra ndo miza tio n

Re side ntia l Burg la ry a nd T he ft fro m Ve hic le Ho t Spo ts Ra ndo m Assig nme nt Re sults

I nitia l Gro up Co mpa riso n

T r e atme nt Gr

• up

N = 24 Contr

• l Gr
• up

N = 24

Me a n

• St. De v

Me a n

• St. De v

T

• Va lue

P-Va lue

Cr ime Pe r Offe nde r

1.63 1.16 1.40 0.70

• 1.07

0.29

Squar e Mile s

0.58 0.23 0.73 0.44

• 1.43

0.16

Population

3,026.58 1,068.05 3,471.38 1,395.98

• 1.24

0.22

Housing

1,239.13 465.15 1,485.58 594.55

• 1.60

0.12

Housing De nsity

2,304.99 1,071.67 2,350.06 927.48

• 0.16

0.88

Prio ritiza tio n o f Offe nde rs

Prio rity 1 Prio rity 2 Prio rity 3 Arre ste d fo r b urg la ry o f a re side nc e o r a c o nve ya nc e F e lo ny pro b a tio n a nd ha d a prio r a rre st fo r b urg la ry F e lo ny pro b a tio n fo r drug

• ffe nse s

Prio rity o f Offe nde rs Co nta c te d

Prio rity 1 42% Prio rity 2 11% Prio rity 3 47% 151 Offe nde rs

NCI C/ F CI C De pa rtme nt o f Mo to r Ve hic le s

E xte rna l L a w E nfo rc e me nt Da ta (DOC, DJJ, e tc .) I nte rna l Po lic e Da ta (F I s, re po rts, e tc .) Ope n So urc e (so c ia l me dia , c o unty re c o rds, Go o g le )

Ana lyze Da ta Co mple te Crimina l Pro file

Da ta fo r Crimina l Re sume

Offe nde r-Ba se d Re spo nse s

• I

nitia l re spo nse b e g ins b a se d o n initia l a na lysis:

~Ve rify o ffe nde rs re side nc e ~Ve rify c o rre c tio na l sta tus

• I

n de pth re spo nse s b a se d o n c rimina l re sume

~Re spo nse s va ry b y na ture o f o ffe nde r’ s a c tivity ~Wo rk with o the r units (e .g ., g a ng unit, na rc o tic s) a nd divisio ns

(e .g ., pa tro l a nd CI D) in the a g e nc y

~Wo rk with sta te a tto rne y, judg e s, pro b a tio n, pa ro le , pa re nts,

• ffe nde r

Co ntinuo us Re spo nse

• Curfe w c he c ks o n o ffe nde rs with sa nc tio ns
• F
• llo w up with pro b a tio n o ffic e rs
• F

a c e -to -fa c e c o nta c t with o ffe nde rs

~Re fe rra ls to pro g ra ms (e .g ., sc ho o l, jo b s) ~F

a mily inte ra c tio n

~Disc uss stre sso rs tha t ma y le a d to re -o ffe nding

• Re spo nse b a rrie rs

~Offe nde rs le a ve ho t spo t ~Offe nde rs with no c urre nt sa nc tio ns

Co nta c t Outc o me T ype

F a c e to fa c e a t ho me 50% F a c e to fa c e fa mily 15% No c o nta c t ma de 18% I nc a rc e ra tio n fo llo w up 12% Arre st 3% Co nta c t with PO 1% Othe r 1%

T

• ta l Numb e r = 1,143

Pa ire d T

• te st

p < .001 Me a n 10.33 SD 4.10

T re a tme nt Gro up (N=24)

Offe nde r- Ba se d Re spo nse s

Co ntro l Gro up (N=24)

I nde pe nde nt T

• te st

p = .66 Me a n 11.00 SD 6.05 Me a n 6.29 SD 4.01 Me a n 8.13 SD 5.40 I nde pe nde nt T

• te st

p = .19 Pa ire d T

• te st

p = .02

E xpe rime nta l I mpa c t Re sults

No rma l Po lic e Re spo nse

Crime Co unts b y Mo nth

5 10 15 20 25 30 35 40 T re a tme nt (39% Cha ng e ) Co ntro l (28% Cha ng e ) 22.6% L e ss Crime in T re a tme nt Ho t Spo ts

Offe nde r-Ba se d Po lic e Re spo nse in L

• ng -T

e rm Pro pe rty Crime Ho t Spo ts E xpe rime nt: Re sults fro m Offe nde rs

Po rt St. L uc ie , F L Po lic e De pa rtme nt F unde d By Bure a u o f Justic e Assista nc e , Sma rt Po lic ing I nitia tive

• Dr. Ra c he l Sa nto s, Asso c ia te Pro fe sso r

Sc ho o l o f Crimino lo g y a nd Crimina l Justic e F lo rida Atla ntic Unive rsity

Pre se nte d b y:

L

• t. Ro b e rto Sa nto s, PhD

Pro fe ssio na l Sta nda rds Po rt St. L uc ie , F L Po lic e De pa rtme nt

T his pro je c t wa s suppo rte d b y Gra nt No . 2012-DB-BX-0002 a wa rde d b y the Bure a u o f Justic e Assista nc e . T he Bure a u o f Justic e Assista nc e is a c o mpo ne nt

• f the Offic e o f Justic e Pro g ra ms, whic h also inc lude s the Bure a u o f Justic e Sta tistic s, the Na tio nal I

nstitute o f Justic e , the Offic e o f Juve nile Justic e a nd De linq ue nc y Pre ve ntio n, the Offic e fo r Vic tims o f Crime , a nd the SMART Offic e . Po ints o f vie w o r o pinio ns in this do c ume nt a re tho se o f the a utho r a nd do no t ne c e ssa rily re pre se nt the o ffic ia l po sitio n o r po lic ie s o f the U.S. De pa rtme nt o f Justic e .

Offe nde r I nte rvie ws

• I

n ho me inte rvie ws with o ffe nde rs a nd fa mily me mb e rs

• Purpo sive sa mple
• Co nduc te d PSL

PD pro je c t dire c to r

• Co nta c t ma de simila r to pro g ra m c o nta c ts
• Pro g ra m de te c tive s no t pre se nt
• I

nte rvie ws fo c use d o n a c c o unta b ility o f de te c tive s/ a g e nc y; impa c t o f pro g ra m o n o ffe nde rs (e .g ., c rime s, a ttitude , fa mily, a nd o ve ra ll life )

Offe nde r De mo g ra phic s

Rac e Inte r vie we e s (N=34) All Offe nde r s (N=151) White

17 (50%) 105 (70%)

Blac k

15 (44%) 41 (27%)

Hispanic

2 (6%) 5 (3%)

Age Unde r 18

7 (20%) 18 (12%)

18 to 25 ye ar s

21 (62%) 72 (48%)

26 to 35 ye ar s

2 (6%) 34(22%)

36 and ove r

4 (12%) 27 (18%)

Se x Male

32 (94%) 133 (88%)

F e male

2 (6%) 18 (12%)

F a mily Me mb e rs I nte rvie we d

• 29 fa mily me mb e rs inte rvie we d

~11mo the rs ~14 fa the rs ~4 spo use s

Re sults: Pe rc e ptio n o f De te c tive s

How do you fe e l a bout how the de te c tive s ha ve inte ra c te d with you (or your son/ da ug hte r/ spouse ) in the ir c onta c ts?

• Appre c ia tive , like d the de te c tive s (re spe c tful)
• Unde rsto o d why the de te c tive s we re c o nta c ting the m

Quo te s:

~I

t fe lt the y we re he lping ; the y we re ve ry re spe c tful.

~T

he pro g ra m is g o o d a nd the de te c tive s a c te d like the y c a re d.

~At e a se a nd c o mfo rta b le with the c o nve rsa tio n. ~I

like d the de te c tive . I t se e me d like the y c a re d a nd wa nte d me to sta y o ut o f tro ub le .

~Do ing the ir jo b ; I

'm o k with tha t.

Re sults: Crimina l Ac tivity

Did the c onta c t with the de te c tive s ma de you think twic e a bout c ommitting a ny ne w c rime s? Influe nc e you to stop a ssoc ia ting with pe ople you thoug ht would g e t you in trouble ?

• Ma de the m “think twic e ” a b o ut c o mmitting c rime
• Sto ppe d g o ing o ut a s muc h with “b a d” pe o ple

Quo te s:

~He ll, ye a h [while la ug hing ]! ~T

he tho ug ht wa s a lwa ys in the b a c k o f the he a d tha t the y wo uld c he c k o n me .

~T

ho ug ht o f it a fe w time s b ut didn't.

~I

t ma de me think twic e b e c a use the ne ig hb o rs wo uld se e .

~He [husb a nd] is no t g o ing o ut a s muc h a nd no t ha ng ing o ut

with c e rta in pe o ple .

Re sults: E ffe c t o n Pro b a tio n

Did the de te c tive s influe nc e you to follow pr

• bation or

supe r vision mor e c lose ly?

• T
• o k sa nc tio ns mo re se rio usly
• Offe nde rs with sa nc tio ns mo re re spo nsive to the pro g ra m

Quo te s:

~I

t wa s a stra in to g e t he r to fo llo w pro b a tio n. T he de te c tive s he lpe d ma ke he r liste n to me [husb a nd] mo re .

~He [so n] a lwa ys fo llo we d b ut this g a ve him a n e xtra push. ~T

he y ma de him unde rsta nd ho w se rio us it wa s.

~T

he de te c tive s ne ve r ma de it ro utine , so yo u ne ve r kne w whe n the y wo uld c o me b y ve rsus pro b a tio n o ffic e r who a lwa ys c o me s b y the sa me time a nd da y.

Re sults: E ffe c t o n F a mily Re la tio nships

Is your re la tionship with your fa mily (son/ da ug hte r/ spouse ) be tte r now tha n a ye a r a g o?

• Re la tio nships a nd c o mmunic a tio n impro ve d
• De te c tive s we re a dditio na l “suppo rt” fo r the fa mily

Quo te s:

~I

'm ho me mo re o fte n a nd c o mmunic a te mo re with my pa re nts.

~He ’ s sta ying o ut o f tro ub le ; b e ing a b e tte r pe rso n; no t

c o mmitting c rime s a nymo re .

~He liste ns to me mo re ; we ha ve b e tte r c o mmunic a tio n. ~We ha ve mo re c o mmunic a tio n a nd g o o ut a s a fa mily. ~He ’ s spe nding mo re time a t ho me .

Re sults: Ove ra ll

• Surprising : Co nsiste ntly po sitive pe rc e ptio ns a nd impa c t
• n fa milie s
• Co nsiste nt with pro c e dura l justic e re se a rc h
• Pro g ra m mo st e ffe c tive with o ffe nde rs with sa nc tio ns
• De te c tive s’ pe rso na litie s a nd inte ra c tio ns ve ry impo rta nt
• De te c tive s pe rc e ive d diffe re ntly a nd mo re influe ntia l

tha n pro b a tio n o ffic e rs Quo te s:

~T

he inte ra c tio n wa s ve ry impo rta nt; I t ha s to b e ve ry re la xe d inte ra c tio n.

~I

t wa s impo rta nt to ha ve the sa me de te c tive to b uild a re la tio nship.

~T

he de te c tive s a c te d like the y we re ve ry c o nc e rne d a b o ut

• ur fa mily's we ll-b e ing .

~I

re c o mme nd to ke e p the visits ra ndo m.

~Ha ving trust with the o ne de te c tive ve rsus the re g ula r c o ps.

Co ntrib utio ns a nd Susta ina b ility

• Hig he st q ua lity re se a rc h
• E

xa mina tio n o f pro pe rty c rime

• Do ne in a typic a l sub urb a n c ity with a ve ra g e c rime
• Re duc tio n o f c rime
• Po sitive a ttitude s a nd impa c t o n o ffe nde rs
• E

xa mple syste ma tic imple me nta tio n fo r o the r a g e nc ie s

~Que stio ns~

46 This project was supported by Grant No. 2013-DP-BX-K006 awarded by the Bureau of Justice Assistance. The Bureau of Justice Assistance is a component of the Office of Justice Programs, which also includes the Bureau of Justice Statistics, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, the Office for Victims of Crime, and the SMART Office. Points of view or opinions in this document are those of the author and do not necessarily represent the official position or policies of the U.S. Department of Justice.

Zoë Thorkildsen March 4, 2015

Raising the Bar for Evidence in Evidence- Based Policing

Using the Maryland Scale of Scientific Methods in SPI

47

Introduction

• Research and evaluation in criminal justice

varies considerably in methodological rigor

– Pre-post comparisons → sophisticated regression techniques → randomized controlled trials

• When determining “what works” in criminal

justice interventions, must consider methodological rigor

– Applies both when considering impact and lack of impact

48

Maryland Scale of Scientific Methods

• “Preventing Crime: What Works, What Doesn't,

What's Promising

– 1997 report to Congress – Lawrence Sherman, Denise C. Gottfredson, Doris L. MacKenzie, John Eck, Peter Reuter, and Shawn D. Bushway

• Established the Maryland Scale of Scientific Methods

– A 1 to 5 scale to assess methodological rigor of analysis of programs and interventions – Key factors are:

• the level of control over other variables
• the potential for measurement error, and
• the statistical power of the analysis

49

Maryland Scale of Scientific Methods

• Correlation between a crime prevention program and a measure of

crime or crime risk factors at a single point in time.

Level 1

• Temporal sequence between the program and the crime or risk
• utcome clearly observed, or the presence of a comparison group

without demonstrated comparability to the treatment group.

Level 2

• A comparison between two or more comparable units of analysis,
• ne with and one without the program.

Level 3

• Comparison between multiple units with and without the program,

control for other factors, or using comparison units that evidence

• nly minor differences.

Level 4

• Random assignment and analysis of comparable units to program

and comparison groups.

Level 5

From: Sherman et al. “Preventing Crime: What Works, What Doesn’t, What’s Promising.” National Institute of Justice, Research in Brief, July 1998.

50

Boston SPI (Phase I) and the Maryland Scale

• Intervention: Safe Street Teams

– Hot spots policing using POP strategies

• Methodology: Retrospective quasi-experimental

– Propensity score matching techniques – Growth curve regression model examining violent crime over time

• Results: Significant reduction in violent crime in

treatment v. control hot spots

• Maryland Scale: Level 4

– Multiple units of analysis – Controlled for additional factors via PSM

51

Boston SPI

Safe Street Teams in action

52

Glendale SPI (Phase I) and the Maryland Scale

• Intervention: Reducing convenience store theft

– CPTED principles at stores – Youth outreach discouraging theft – Targeted surveillance and patrol by officers

• Methodology: Difference-in-differences

– Treatment stores v. control stores pre-post

• Results: Significant reduction in CFS at 5 of 6

targeted stores, overall CFS reduction of 42%

• Maryland Scale: Level 4

– Multiple units – Control stores are demonstrably similar to treatment group

• Revised analysis using interrupted time series, also

Level 4

53

Glendale SPI

BEFORE AFTER

54

Los Angeles SPI and the Maryland Scale

• Intervention: Operation LASER

– Offender-based strategies: Crime Intelligence Detail – Place-based strategies: hot spot policing, POP

• Methodology: Quasi-experimental design

– Interrupted time series – Treatment v. control hot spots

• Results: Significant reduction in violent crime,

average of 5.4 fewer crimes per month

• Maryland Scale: Level 4

– Pre-post comparison of crime trends controlling for existing trends – Includes comparison areas

55

Los Angeles SPI

50 100 150 20 40 60 80

Intervention

Month Gun-Related Crimes Predicted Crime

Jan 2006 to Jun 2012

Gun-Related Crimes in Newton

56

Palm Beach SPI and the Maryland Scale

• Intervention: Community outreach

– Improving police legitimacy – Reducing victimization in immigrant community

• Methodology: Pre-post comparisons

– Community survey results – Robbery crime and arrest rates

• Results: Improved community perception of police;

initial spike in robberies then 22% decrease (not significant)

• Maryland Scale: Level 2

– Pre-post comparison – Control area designated but implementation problems

57

Palm Beach SPI

Engaging Police in Immigrant Communities (EPIC) project

58

Philadelphia SPI and the Maryland Scale

• Intervention: Comparison of three methods

– Foot patrol – POP – Offender-focused strategies

• Methodology: Randomized, controlled trials

– 81 hot spots were deliberately assigned into three methods – Within each 27 hot spot group, treatment v. control was randomly assigned

• Results: Offender-focused strategies associated with a

significant, 22% decrease in violent crime

• Maryland Scale: Level 5

– Randomized controlled trials

59

Philadelphia SPI

60

Conclusions

• Methodological considerations must be

incorporated during planning phases

• Randomized, controlled trials are the “golden

standard” but not always appropriate or possible

• Effective use of analysis techniques can

improve methodological rigor

61

Q & A