GPS Monitoring Technologies and Domestic Violence: An Evaluation Study

The author(s) shown below used Federal funds provided by the U.S.

Department of Justice and prepared the following final report:

Document Title: GPS Monitoring Technologies and Domestic

Violence: An Evaluation Study

Author: Edna Erez, LL.B., Ph.D., Peter R. Ibarra, Ph.D.,

William D. Bales, Ph.D., Oren M. Gur, M.S.

Document No.: 238910

Date Received: June 2012

Award Number: 2007-IJ-CX-0016

This report has not been published by the U.S. Department of Justice.

To provide better customer service, NCJRS has made this Federally-

funded grant final report available electronically in addition to

traditional paper copies.

Opinions or points of view expressed are those

of the author(s) and do not necessarily reflect

the official position or policies of the U.S.

Department of Justice.

GPS Monitoring Technologies and Domestic

Violence: An Evaluation Study

Edna Erez, LL.B., Ph.D., University of Illinois at Chicago

Peter R. Ibarra, Ph.D., University of Illinois at Chicago

William D. Bales, Ph.D., Florida State University

Oren M. Gur, M.S., University of Illinois at Chicago

June 2012

This project was supported by Grant No. 2007-IJ-CX-0016 awarded by the National Institute of

Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are

those of the authors and do not necessarily represent the official position or policies of the US

Department of Justice.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study

Erez, Ibarra, Bales, & Gur

Authors:

Edna Erez, LL.B., Ph.D.

Department of Criminology, Law and Justice

University of Illinois at Chicago

Behavioral Sciences Building (MC 141)

1007 W. Harrison Street

Chicago, IL 60607

(T) 312 996 5262 (F) 312 998 8355

[email protected]

Peter R. Ibarra, Ph.D.

Department of Criminology, Law and Justice

University of Illinois at Chicago

Behavioral Sciences Building (MC 141)

1007 W. Harrison Street

Chicago, IL 60607

(T) 312 413 2475 (F) 312 996 8355

[email protected]

William D. Bales, Ph.D.

College of Criminology and Criminal Justice

Florida State University

Hecht House

634 W. Call Street

Tallahassee, Florida 32306

(T) 850 644 7113 (F) 850 644 9614

wbales@fsu.edu

Oren M. Gur, M.S.

Department of Criminology, Law and Justice

University of Illinois at Chicago

Behavioral Sciences Building (MC 141)

1007 W. Harrison Street

Chicago, IL 60607

(T) 312 996 5290 (F) 312 996 8355

[email protected]

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page i

Erez, Ibarra, Bales, & Gur

Abstract

This study examines the implementation of Global Positioning System (GPS) monitoring

technology in enforcing court mandated “no contact” orders in domestic violence (DV) cases,

particularly those involving intimate partner violence (IPV). The research also addresses the

effectiveness of GPS as a form of pretrial supervision, as compared to other conditions in which

defendants are placed. The project has three components: First, a national web-based survey of

agencies providing pretrial supervision reported on patterns of GPS usage, as well as the

advantages, drawbacks, and costs associated with using GPS for DV cases. The results indicate a

gradual increase in agencies’ use of GPS technology for DV cases since 1996, primarily to

enhance victim safety and defendant supervision.

Second, a quasi-experimental design study of three sites from across the U.S. – referred

to as “Midwest,” “West,” and “South” – examined the impact of GPS technology on DV

defendants’ program violations and re-arrests during the pretrial period (referred to as the “short

term”), and on re-arrests during a one-year follow-up period after case disposition (referred to as

the “long term”). The results indicate that GPS has an impact on the behavior of program

enrollees over both short and long terms. Examination of the short-term impact of GPS

enrollment shows it is associated with practically no contact attempts. Furthermore, defendants

enrolled in GPS monitoring have fewer program violations compared to those placed in

traditional electronic monitoring (EM) that utilizes radio frequency (RF) technology (i.e.,

remotely monitored and under house arrest, but without tracking). GPS tracking seems to

increase defendants’ compliance with program rules compared to those who are monitored but

not tracked.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page ii

Erez, Ibarra, Bales, & Gur

Defendants enrolled in the Midwest GPS program had a lower probability of being re-

arrested for a DV offense during the one-year follow-up period, as compared to defendants who

had been in a non-GPS condition (e.g., in jail, in an RF program, or released on bond without

supervision). In the West site, those placed on GPS had a lower likelihood of arrest for any

criminal violation within the one-year follow-up period. In the South site, no impact deriving

from participation in GPS was observed. The heterogeneity of the defendants who are placed on

GPS at this site, and the different method for generating the South sample of DV defendants,

may account for the absence of GPS impact on arrest in the long term.

An examination of the relationship between GPS and legal outcomes across the three

sites revealed similar conviction rates for defendants on GPS and those who remained in jail

during the pretrial period. Further, a comparison of conviction rates for GPS and RF defendants

at the Midwest site found a significant difference – with GPS defendants being likelier to be

convicted as compared to RF defendants; conviction rates in the Midwest and South sites were

also higher for GPS defendants compared to defendants released on bond without supervision,

suggesting that defendants’ participation in GPS increases the likelihood of conviction. These

findings may be related to the fact that GPS provides victims with relief from contact attempts,

empowering them to participate in the state’s case against the defendant.

The third component of the study is a qualitative investigation conducted at six sites,

entailing in-depth individual and group interviews with stakeholders in domestic violence cases –

victims, defendants and criminal justice personnel. The interviews identified a variety of

approaches to organizing GPS programs, with associated benefits and liabilities. Victims largely

felt that having defendants on GPS during the pretrial period provided relief from the kind of

abuse suffered prior to GPS, although they noted problems and concerns with how agencies and

courts apply GPS technology. Interviews with defendants supported quantitative findings about

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page iii

Erez, Ibarra, Bales, & Gur

the impact of GPS on defendants’ short- and long-term behavior, and found both burdens and

occasional benefits associated with participation. Benefits of GPS enrollment for defendants

included protecting them from false accusations, providing added structure to their lives, and

enabling them to envision futures for themselves without the victim. Burdens pertained to living

with restrictions and becoming transparent, managing issues related to stigma and disclosure of

one’s status as a DV defendant tethered to GPS, and handling the practical issues that emerge

with the technology and equipment. Policy implications highlight the importance of having a

logical connection between defendant attributes and program details, avoiding enrollment in

cases where the GPS has minimal or no value and is imposed for reasons other than protecting

victims or enforcing restraining orders, the need for justice professionals to cultivate

relationships with victims whose abusers are on GPS, and the importance of maintaining an

appropriate balance between victim safety and due process for the defendant.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page iv

Erez, Ibarra, Bales, & Gur

Table of Contents

Abstract ........................................................................................................................................... i

Table of Contents ......................................................................................................................... iv

Acknowledgments ........................................................................................................................ ix

Executive Summary ....................................................................................................................... x

The purpose of the study ........................................................................................................ x

Research design and methods ............................................................................................ xii

Findings ............................................................................................................................... xiii

A. Web-based survey ............................................................................................................. xiii

B. Varieties of GPS for DV programs: Perspectives of justice personnel in six sites ............ xv

C. Quantitative impact study (Midwest, West, South) ........................................................... xvi

D. Victims and defendants: Views and experiences ............................................................. xvii

Study limitations ............................................................................................................... xviii

Conclusions ........................................................................................................................... xx

Chapter 1: Introduction ................................................................................................................1

A. Statement of the problem ..................................................................................................... 1

B. Literature review .................................................................................................................. 2

C. Hypotheses and research questions of the present study .................................................... 12

1. Hypotheses .......................................................................................................................13

2. Research questions ...........................................................................................................13

Chapter 2: Methodology ..............................................................................................................15

A. Web-based survey .............................................................................................................. 15

1. Topics and procedures .....................................................................................................16

B. Quantitative assessment of program effectiveness (the quasi-experimental component) .. 17

1. Participant selection criteria, methods, and sampling ......................................................18

2. Measures and indicators ..................................................................................................22

C. Qualitative assessment of GPS-based monitoring .............................................................. 24

1. Interviewee recruitment and interview procedures ..........................................................26

2. Topics of in-depth interviews ..........................................................................................27

3. Qualitative data analysis ..................................................................................................28

Chapter 3: Findings .....................................................................................................................29

A. Web-based survey .............................................................................................................. 29

1. Types of technologies used and defendants monitored ...................................................30

2. Program start date and location .......................................................................................32

3. Rationale and aims of GPS programs ..............................................................................34

4. Referrals and enrollment ..................................................................................................36

5. Technological attributes and costs ...................................................................................37

6. Current uses and changing program attributes ................................................................41

7. Victim-relevant factors ....................................................................................................44

8. Impact of GPS on defendant behavior and other defendant-relevant factors ..................44

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page v

Erez, Ibarra, Bales, & Gur

B. Varieties of GPS for DV programs: Perspectives of justice personnel in six sites ............ 47

1. Midwest: Crime control and penality ..............................................................................51

2. South: Due process and treatment ...................................................................................53

3. West: Due process and penality .......................................................................................55

4. East: Due process and penality ........................................................................................58

5. Southeast: Treatment and crime control ..........................................................................59

6. Southwest: An ad hoc, make-shift approach ...................................................................60

7. Conclusion .......................................................................................................................61

C. Quantitative impact study (Midwest, West, South) ............................................................ 62

1. Midwest ...........................................................................................................................62

2. West .................................................................................................................................73

3. South ................................................................................................................................83

D. Victims and defendants: Views and experiences ............................................................... 94

1. Victims .............................................................................................................................94

2. Defendants .....................................................................................................................108

Chapter 4: Conclusions .............................................................................................................125

A. Discussion ........................................................................................................................ 125

1. Web-based survey ..........................................................................................................125

2. The quantitative impact study ........................................................................................127

3. Qualitative findings .......................................................................................................138

B. Summary and conclusion ................................................................................................. 144

C. Implications for policy and practice ................................................................................. 147

D. Limitations of the study .................................................................................................... 157

E. Implications for future research ........................................................................................ 161

References ...................................................................................................................................165

Dissemination of research findings...........................................................................................170

Appendices ..................................................................................................................................171

A. Web-based survey ............................................................................................................ 171

B. Midwest coding sheet for GPS ......................................................................................... 185

C. Midwest coding sheet for non-GPS groups ...................................................................... 189

D. Midwest agency intake form ............................................................................................ 193

E. Midwest agency reassessment/termination form .............................................................. 196

F. West coding sheet for GPS ............................................................................................... 199

G. West coding sheet for non-GPS groups ........................................................................... 205

Endnotes .....................................................................................................................................209

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page vi

Erez, Ibarra, Bales, & Gur

List of Tables

Table 3A-1. Percentage of respondents indicating their program uses each technology, by type of

defendant ..................................................................................................................................................... 30

Table 3A-2. Percentage of respondents in the DV subsample utilizing each technology, by type of

defendant ..................................................................................................................................................... 31

Table 3A-3. Percentage of respondents in the DV subsample utilizing each number of technologies, by

type of defendant ......................................................................................................................................... 31

Table 3A-4. Descriptive statistics for number of monthly referrals and enrollments/hook-ups for programs

in the DV subsample ................................................................................................................................... 36

Table 3A-5. Average monthly caseloads for officers using GPS and non-GPS officers in the DV

subsample .................................................................................................................................................... 37

Table 3A-6. Cost of active, hybrid, and passive GPS for programs reporting multiple levels of monitoring

and costs in the DV subsample ................................................................................................................... 39

Table 3C-1. Descriptive statistics of variables with all cases: Midwest site .............................................. 63

Table 3C-2. Descriptive statistics (mean values) of variables across four groups: Midwest site ............... 64

Table 3C-3. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether violated among all cases, dismissed/acquitted cases, and convicted cases: Midwest site ............ 65

Table 3C-4. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: Midwest site ........ 66

Table 3C-5. Logistic regression models of effect of length of time on GPS and long-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: Midwest site ........ 66

Table 3C-6. Logistic regression models of the effect of GPS versus RF on short-term outcome of

violation for all cases, dismissed/acquitted cases, and convicted cases: Midwest site ............................... 68

Table 3C-7. Logistic regression models of the effect of GPS versus RF on short-term outcome of arrest

for all cases and convicted cases: Midwest site .......................................................................................... 68

Table 3C-8. Logistic regression models of the effect of GPS versus RF on long-term outcome of arrest for

all cases, dismissed/acquitted cases, and convicted cases: Midwest site .................................................... 69

Table 3C-9. Logistic regression models of the effect of GPS versus non-GPS on short-term outcome of

arrest for all cases, dismissed/acquitted cases, and convicted cases: Midwest site .................................... 70

Table 3C-10. Logistic regression models of the effect of GPS versus non-GPS on long-term outcome of

arrest for cases, dismissed/acquitted cases, and convicted cases: Midwest site ......................................... 71

Table 3C-11. Logistic regression models of the effect of GPS versus non-GPS on long-term outcome of

arrest for domestic violence for all cases, dismissed/acquitted cases, and convicted cases: Midwest site .71

Table 3C-12. Logistic regression models of the effect of RF versus non-EM on short-term outcome of

arrest for all cases: Midwest site ................................................................................................................. 72

Table 3C-13. Logistic regression models of the effect of RF versus non-EM on long-term outcome of

arrest for all cases, dismissed/acquitted cases, and convicted cases: Midwest site .................................... 73

Table 3C-14. Descriptive statistics of cases placed on GPS: West site ...................................................... 74

Table 3C-15. Descriptive statistics (mean values) of variables across three groups: West site ................. 76

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page vii

Erez, Ibarra, Bales, & Gur

Table 3C-16. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether violated among all cases, dismissed/acquitted cases, and convicted cases: West site .................. 77

Table 3C-17. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: West site .............. 78

Table 3C-18. Logistic regression models of effect of length of time on GPS and long-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: West site .............. 79

Table 3C-19. Descriptive statistics of GPS and non-GPS cases: West site ................................................ 80

Table 3C-20. Logistic regression models of effect of GPS versus no GPS on long-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: West site .............. 82

Table 3C-21. Descriptive statistics of variables with all cases: South site ................................................. 84

Table 3C-22. Descriptive statistics (mean values) of variables across four groups: South site ................. 85

Table 3C-23. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether violated among all cases and convicted cases: South site ............................................................. 86

Table 3C-24. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: South site ............. 87

Table 3C-25. Logistic regression models of effect of length of time on GPS and long-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: South site ............. 88

Table 3C-26. Logistic regression models of the effect of GPS versus RF on short-term outcome of

violation for all cases, dismissed/acquitted cases, and convicted cases: South site .................................... 89

Table 3C-27. Logistic regression models of the effect of GPS versus RF on short-term outcome of re-

arrest for all cases and convicted cases: South site ..................................................................................... 90

Table 3C-28. Logistic regression models of the effect of GPS versus RF on long-term outcome of re-

arrest for all cases, dismissed/acquitted cases, and convicted cases: South site ......................................... 90

Table 3C-29. Logistic regression models of the effect of GPS versus non-GPS on short-term outcome of

re-arrest for all cases, dismissed/acquitted cases, and convicted cases: South site ..................................... 91

Table 3C-30. Logistic regression models of the effect of GPS versus non-GPS on long-term outcome of

re-arrest for all cases, dismissed/acquitted cases, and convicted cases: South site ..................................... 92

Table 3C-31. Logistic regression models of the effect of RF versus non-EM on short-term outcome of re-

arrest for all cases: South site ...................................................................................................................... 93

Table 3C-32. Logistic regression models of the effect of RF versus non-EM on long-term outcome of

arrest for all cases, dismissed/acquitted cases, and convicted cases: South site ......................................... 93

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page viii

Erez, Ibarra, Bales, & Gur

List of Figures

Figure 3A-1. Year in which organization began using GPS technology for tracking domestic violence

(DV) defendants .......................................................................................................................................... 32

Figure 3A-2. Slopegraph of states with programs using EM (left column) and GPS (right column) ........ 33

Figure 3A-3. Percentage of respondents in the DV subsample indicating how important each reason is to

the agency when deciding to start to use GPS for supervising DV pretrial defendants ............................. 34

Figure 3A-4. Percentage of respondents in the DV subsample indicating how important each factor is for

choosing who is placed on GPS when used for supervising DV pretrial defendants ................................. 35

Figure 3A-5. Percentage of respondents in the DV subsample indicating how often defendants are not

enrolled in GPS for DV pretrial programs for each reason ......................................................................... 36

Figure 3A-6. Percentage of programs utilizing each vendor in the DV subsample .................................... 38

Figure 3A-7. Capabilities utilized by programs in the DV subsample ....................................................... 40

Figure 3A-8. Percentage of respondents in the DV subsample indicating how important each reason is to

the agency when using GPS for supervising pretrial defendants ................................................................ 42

Figure 3A-9. Percentage of respondents in the DV subsample indicating to what extent they agree or

disagree with statements regarding using GPS for supervising DV pretrial defendants ............................ 43

Figure 3A-10. Victim-centric features of programs in the DV subsample ................................................. 44

Figure 3A-11. Percentage of respondents in the DV subsample indicating the percentage of defendants

participating in each program while being monitored ................................................................................ 46

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page ix

Erez, Ibarra, Bales, & Gur

Acknowledgments

This study has benefited from the advice, help, and experience of several individuals who

participated in different phases of the project. Amie Schuck (initially a co-PI who withdrew due

to other commitments) developed sections of the data instruments and organized data collection

in two of the three quantitative sites (“West” and “Midwest”). Peggy Conway provided

invaluable advice and suggestions throughout, especially during formative stages but also in

addressing and presenting the issues involved in the use of electronic monitoring in criminal

justice. Norman Lurie and Terry Baumer provided valuable feedback on different iterations of

the grant proposals. Karen Mann helped with management and analyses of the quantitative data

generated from the three sites. Amanda Vasquez and Daniel Downes collected data for the

quantitative study in one site. Daniel Lawrence processed the quantitative data from two sites of

the study and, along with Amanda Vasquez and Laurel Mazar, transcribed interviews. Alana

Gunn helped categorize and interpret qualitative data, and Emma Kalimuthu participated in the

launch of the national web-based survey. We owe much gratitude to agency administrators in the

six sites where we interviewed stakeholders, and particularly to the West, Midwest and South

sites, where we also gathered quantitative data.

Members of the Institutional Review Board (IRB) of the University of Illinois at Chicago

(UIC) merit appreciation for their efforts in the course of approving protocols of this research.

We also benefited from the advice and guidance of grant monitors at the National Institute of

Justice (NIJ): in the first year, Marlene Beckman, and in subsequent years, Laurie Bright, with

Marie Garcia monitoring the project during the last six months through completion. At UIC, Tara

Gordon assisted with administering the grant during its first half. In subsequent years, Tracy

Sikorski of the Office of Social Science Research helped tremendously with fulfilling grant

requirements. The efforts of these individuals facilitated the successful completion of this study.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page x

Erez, Ibarra, Bales, & Gur

Executive Summary

The purpose of the study

This study examines the implementation of Global Positioning System (GPS)-based

monitoring technology in enforcing court mandated “no contact” orders in domestic violence

(DV) cases, particularly those involving intimate partner violence (IPV). The research also

addresses the effectiveness of GPS-based monitoring as a form of pretrial supervision, as

compared to other conditions in which defendants are placed. GPS for DV programs first

emerged in the 1990s and their numbers have since been steadily increasing. GPS for DV

programs are created in response to local demand, court initiative, or state legislation, often

prompted by an incident in which an intimate partner was harmed or killed by an arrestee who

had been released by a court on bond, and are usually directed at bolstering protection orders.

Since 2000, twenty-one states and the District of Columbia have enacted legislation mandating

or recommending that justice agencies employ GPS to protect victims of DV during the pretrial

period; several other states are in the process of considering such legislation.

When used for DV offenses, GPS is typically assigned during the pretrial period, defined

as the justice system phases beginning with arrest and concluding with case disposition. This

period is known to be highly volatile, characterized by heightened danger to the victim and

attempts by the accused to dissuade the victim from participating in the prosecution of the case.

The adoption of GPS tracking to enhance victim safety emerged as a preferred method of

electronic monitoring (EM) over radio frequency (RF)-based approaches that operated primarily

as house arrest programs. When assigned in the context of a DV-related charge, traditional RF

programming was commonly enhanced through the use of a bilateral strategy that augmented

house arrest by placement of a second receiver in the victim’s residence (hereafter “bilateral

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xi

Erez, Ibarra, Bales, & Gur

RF”). The second receiver detected the presence of the defendant wearing a synchronized

transmitter (typically on his ankle), within a range of approximately 500 feet. By synchronizing

the second receiver to the defendant’s transmitter, authorities would be alerted if the defendant

had entered the perimeter of the victim’s residence.

Although bilateral RF was an enhancement to simple house arrest (hereafter “RF”), GPS

represents a further advance in its ability to track a defendant across time and space. Like

bilateral RF, GPS for DV is intended to deter contact attempts by the defendant toward the

victim, but GPS operates on the principle of “geo-fencing” – it can be programmed to establish

multiple and potentially unlimited zones of exclusion (defined as areas where the alleged abuser

may not enter) and inclusion (areas where the alleged abuser is required to remain as per

program specifications). Contact attempts are measured by the defendants’ incursions into

exclusion zones (i.e., areas where the victim is likely to be present), but it does not detect

contacts attempted via other means, such as telephone, email, social media, SMS (i.e., text

messages), chance encounters outside monitored areas, or contacts initiated by the victim (e.g.,

when the victim enters areas where the defendant is present). The advantages of GPS relative to

bilateral RF include its versatility, broadened detection range, capacity for multiple zone

coverage, and, in the version of GPS known as “active,” its unique ability to keep track in real

time of the offender’s whereabouts beyond the immediate range of his and the victim’s

residences (i.e., through logging of the offender’s “GPS points”). Possible drawbacks to using

GPS include greater per diem costs to agencies, greater workload demands on staffs, elevated

legal jeopardy and net-widening effects for defendants, cultivation of false feelings of safety in

victims, and the problems of reliability and accuracy associated with a developing technology.

The purpose of the study was to evaluate the effectiveness of GPS in accomplishing the

aims of victim protection via contact deterrence during the pretrial period. The second aim was

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xii

Erez, Ibarra, Bales, & Gur

to assess whether GPS enrollment is associated with deterrence in the long term, as measured by

lower likelihood of re-arrest during a one-year follow-up period. The study also attempted to

provide an integrated picture of the benefits (direct and indirect), as well as risks and liabilities,

resulting from the use of GPS in DV cases.

Research design and methods

The study is comprised of three parts: First, a web-based survey of pretrial service

agencies in the United States was conducted to learn how electronic monitoring technologies are

used in their operations, with a specific emphasis on applications of GPS to address DV cases.

Agencies known to use GPS were contacted using a national list provided by an EM expert

consultant, Peggy Conway. In addition, two organizations for pretrial agencies, the Pretrial

Justice Institute (PJI) and American Probation and Parole Association (APPA), sent an email

with an embedded link inviting their members to participate in the web-based survey. The survey

received responses from 616 individuals representing agencies in 43 states, the District of

Columbia, and Puerto Rico.

Second, a quasi-experimental design was applied to the analysis of quantitative data

derived from large and established programs, with the aim of assessing GPS systems’

effectiveness in a) deterring defendants’ program violations (including contact violations) while

being monitored, and in b) reducing participants’ re-arrest during both the monitoring period and

a one-year follow-up period. The three sites offer geographic diversity as well as variations in

how agencies approach the use of GPS to monitor defendants’ compliance with “no contact”

orders in DV cases during the pretrial period. In the Midwest site, a total of 2052 defendants

referred to the GPS program over a two-year period (2006-8) comprised the sample; they spent

the pretrial period either on GPS, on RF, in jail, or were bonded out without any form of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xiii

Erez, Ibarra, Bales, & Gur

supervision. In the West site, a total of 1000 defendants who were referred to GPS over sixty-

four months (Oct 2001 – Jan 2007) comprised the sample. These defendants were placed on

GPS, kept in jail, or released on bond without any form of supervision. In the South site, a total

of 604 defendants with DV charges were selected from jail booking and release data over two

years (2008-2009). These defendants spent the pretrial period in one of four conditions: either

GPS or RF-based supervision, on bond release, or in jail. To conduct an analysis of GPS vs. RF

outcomes, data were supplemented with information found in the records of the pretrial agency

that monitored GPS and RF defendants during the pretrial period.

The third prong of the study consisted of in-depth interviews with parties directly or

indirectly involved with programs (victims, offenders, justice personnel, and social service

providers; N = 210) at six agencies – the three included in the quantitative study and three

additional sites, located in different parts of the country and practicing varying approaches to

GPS for DV. The in-depth individual and group interviews investigated participants’ experiences

with GPS and the effects of justice-related practices on victims and defendants.

Findings

A. Web-based survey

The national survey provided a snapshot of current trends in the adoption of GPS and

emerging standards and practices in the targeted implementation of GPS for DV cases. While

most respondents agreed that the desire to keep victims safer was the most important impetus for

their program’s creation, few reported that their programs had objective features to actively

engage victims in the process of increasing their own safety in partnership with the criminal

justice system. The role of the victim is still being defined and aligned with technological efforts

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xiv

Erez, Ibarra, Bales, & Gur

to enhance supervision, ensure client accountability, more effectively protect the public, and

deter additional crimes.

The concern with victims influences who is enrolled in GPS for DV programs.

Practitioners stated their programs were targeting primarily serious offenses for GPS supervision

and cases for which the court had issued an order of protection. A victim request that a defendant

be placed on GPS was also an important consideration for enrollment, as was the defendant’s

prior history of alcohol or drug use. The majority of programs reported successfully enrolling all

referred defendants, but a quarter of the programs reported being able to hook up only half of

those referred to GPS, for a variety of reasons, including defendants not meeting program

requirements, a lack of available GPS equipment, absence of victim cooperation (when the

latter’s consent was required), or the defendants inability to pay program fees. The results also

suggest that the overwhelming majority of programs lack victim-centric features in spite of the

expressed sentiment that GPS was most important as a tool for victims. Programs most often

utilize abilities that allow them to map the defendant’s movements in the community over time,

determine their current location, send the defendant an alert, and establish inclusion and

exclusion zones. Yet the least utilized function is one that allows victims to receive text message

notification when the defendant violates zone restrictions. Such an alert would allow the victims

to take their own precautions without having to rely on law enforcement, a vital option

considering fewer than half of the programs report that law enforcement automatically responds

to defendant violations or alerts.

GPS programs rely on the ability of their officers to effectively supervise defendants, and

a reduced caseload allows GPS officers to scrutinize the information generated by advanced

technologies – on average GPS officers monitor less than half as many clients as non-GPS

officers. Yet, the majority of practitioners agreed with the statement that the GPS cannot prevent

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xv

Erez, Ibarra, Bales, & Gur

a defendant from committing a crime, which could be evidence of the need to shift the object of

practitioner focus onto the victim. Some practitioners also were of the opinion that GPS tracking

empowers victims, or at least that GPS was more empowering to the victim than punitive to the

defendant. However, the majority (70%) also agreed with the statement that victims

misunderstand the capabilities of GPS tracking, and over half agreed that it gives victims a false

sense of security – yet 80% thought it made the general public safer. Only 13% of programs had

ever formally evaluated the effectiveness of GPS tracking for defendants, and only a third

employ special procedures to assess the risk of violent behavior by the defendant during the

pretrial period.

Among programs reporting multiple levels of monitoring (active and passive), the

average cost of GPS to the agency per day is $11.18 for active monitoring, and $6.84 for passive

monitoring. The cost to defendants per day is $8.68 for active monitoring and $6.79 for passive

monitoring. The overall figures reported by a larger national sample of GPS-using agencies is a

$9.80 cost to the agency and $8.80 charged to the defendant. These per diem amounts do not

include the costs of personnel or program administration. On average, DV defendants spend 99.5

days on GPS, with wide variation between programs.

B. Varieties of GPS for DV programs: Perspectives of justice personnel in six sites

The qualitative data gathered from interviews with criminal justice personnel identified a

variety of approaches to organizing GPS for DV programs, each of which can be located on two

continua: first, with respect to crime control vs. due process models of justice, and second, in

relation to penality vs. treatment approaches to defendants. Programs differed in their

restrictions, the scope of investigating defendants’ movements (via “GPS points”), willingness to

violate defendants who did not conform to program rules, levels of fee for participation in the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xvi

Erez, Ibarra, Bales, & Gur

program, and average stay on the program. The different approaches taken were also influenced

by understandings about DV and the need to separate defendants from alleged victims, the

characteristics (including criminal records) of DV defendants who were admitted into the

program, the type of agency, and the depth of agency resources. The different approaches taken

by the six agencies may be helpful in understanding the findings that emerged in the three

quantitative impact study sites.

C. Quantitative impact study (Midwest, West, South)

The quantitative impact study shows that GPS technologies have an impact in the short

term (during the pretrial period): GPS is effective in preventing defendants from (physically)

contacting victims, suggesting that GPS “puts teeth” into restraining orders. A more rigorous

program (at the Midwest site) found a reduced likelihood of program violations (e.g., curfew) by

GPS defendants as compared to RF defendants, while a less restrictive program (at the South

site) found no differences in the likelihood of program violations by GPS defendants as

compared to RF defendants. This finding is especially surprising since the Midwest GPS

defendants, who had a higher mean number of prior arrests compared to their counterparts at the

South site, had a lower likelihood of violating program rules.

In terms of court outcomes, GPS defendants had conviction (and dismissal) rates that

were more similar to rates found among those who remained in jail, as contrasted to rates found

among defendants placed on RF or released on bond without supervision. At the Midwest site,

conviction rates were even higher for GPS defendants than for defendants remaining in jail. It

may be that program-enforced prohibition against contacting the victim (e.g., to persuade her to

renege or not appear in court) results in a higher conviction rate.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xvii

Erez, Ibarra, Bales, & Gur

The impact of enrollment in GPS over the long term (or during a one-year follow-up

period after disposition or completion of a jail sentence) is reflected in a lower likelihood of

arrest for any offense in the West site, and a lower likelihood of arrest for DV offenses in the

Midwest site. The South site did not show any long-term differences attributable to enrollment in

GPS during the one-year follow-up period. The heterogeneity of the defendants who are placed

on GPS at this site, and the different method for generating the South sample of DV defendants,

may account for the absence of GPS impact on arrest in the long term.

D. Victims and defendants: Views and experiences

The qualitative data show that GPS programs can provide victims peace of mind and

relief from harassment and abuse, such that the resumption of a normal life seems more tenable.

Although many victims did not understand how the technology works, most expressed an

awareness of the technology’s limitations in terms of guaranteeing their safety. In some cases

victims were anxious when they saw their alleged abuser moving freely about in settings outside

the exclusion zone(s), and expressed concern that their estranged partners would be able to

manipulate the technology or subvert its capacities and undermine program rules and restrictions.

Victims also discussed problems pertaining to maintaining their privacy during court hearings

related to the defendant’s GPS enrollment (e.g., having to divulge the victim’s new address or

provide information about the children’s babysitter). Concerns were also raised about inadvertent

disclosure of victim locations marked in “unknown zones” due to alerts the monitoring agency

sent to defendants. At the same time, some victims expressed satisfaction at being able to

covertly visit estranged partners enrolled in the GPS program, in some cases even staying

overnight, mentioning that they could come and go “on their own terms.”

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xviii

Erez, Ibarra, Bales, & Gur

Defendants reported having both positive and negative experiences during their time on

GPS, viewing it as “a mixed bag.” Most felt it was far preferable to “sitting in jail” and were

grateful that participation enabled them to maintain their employment. They also appreciated the

fact that GPS shielded them from false accusations that could be (and in some cases were) made

by a vengeful estranged partner. Defendants at a more treatment-oriented site spoke of being

thankful for the various kinds of assistance they received from supervising officers. Defendants

also spoke about using their time in the program, and away from the alleged victim, as an

opportunity to engage in various constructive pursuits, including rebuilding relations with family

members, looking for work, returning to school, and reimagining their lives without the victim

having a part in it.

On the other hand, defendants enrolled in the more rigorous programs found them to be

quite demanding and the personnel extremely inflexible, especially around issues related to

rescheduling “out hours” (i.e., the specific periods that defendants are given to conduct “personal

business”) or accommodating their commuting situation, particularly when reliant on public

transportation. The technology and equipment prompted a number of practical and logistical

concerns, such as difficulty with maintaining an active signal while at work and inadvertently

disclosing one’s status as a monitored subject. Concerns were also raised that GPS participation

could damage or undermine their employment situation or chances of being hired.

Study limitations

This study has several imitations that are common to social science research, particularly

in the criminal justice context. The national web-based survey may suffer from problems of

reliability and generalizability. Because the survey is respondent- rather than agency-based, it is

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xix

Erez, Ibarra, Bales, & Gur

possible that some agencies are overrepresented as a result of more than one employee

responding.

The limitations of the quantitative impact study stem from the research design and the

data that were available to measure key attributes and outcomes. The ideal approach to

examining treatment effects – experimental design with random assignment of individuals to

comparison groups – could not be utilized because referral of DV defendants to GPS programs is

a function of judicial discretion. Therefore, a quasi-experimental design with relevant controls

was used to derive non-GPS and GPS groups that were as equivalent as possible on factors

known to influence the outcomes. Equivalency was accomplished by statistically controlling for

covariates that were observable in the three locations.

The comparability of the GPS and non-GPS groups may have been compromised by

various factors, some related to the socioeconomic status of the defendants (e.g., inability to post

bond or establish a new residence during the pretrial period), others related to program rules and

technological issues (e.g., victim’s consent could not be obtained, GPS signal could not be

sustained at the defendant’s or victim’s residential location). Because SES data were not

available for all groups, SES could not be controlled for in the analysis.

The measures used to create equivalent non-GPS and GPS DV defendant groups were

limited by the data available in the three sites, and the disparate definitions and recording

practices used by the agencies. Had more control variables been available for inclusion in the

statistical models, results of the analyses may have varied. Also, the completeness of the

measures that were used may have been compromised due to missing data. The use of

administrative data to test hypotheses in criminal justice research is accepted practice, however,

even though it may be problematic.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xx

Erez, Ibarra, Bales, & Gur

The outcome measure of recidivism as repeat domestic violence offenses (or crime in

general) through re-arrest may also be problematic; it is likely to miss incidents in which

offenders perpetrate abuse but the offending behavior is not reported or detected – a particularly

common problem in domestic violence cases.

The qualitative component of the study was limited in that it primarily relied on

interviews to document the perspectives and experiences of personnel, victims, and defendants,

and was not complemented with extensive systematic observational fieldwork with each group of

stakeholders. The latter method may provide insights into taken for granted understandings that

may not emerge in the course of being interviewed.

Conclusions

GPS for DV programs usually monitor DV arrestees referred as a condition of pretrial

release, without having been convicted. Thus, defendants may experience these pretrial

supervision programs as a form of punishment without benefit of trial; indeed, the present study

found that, in the three sites, almost half of all GPS clients’ cases were dismissed. This result has

potentially far-reaching implications, both for defendants and for how society responds to

domestic violence. Because GPS programs for DV defendants are increasing, and since they

appear to be effective in accomplishing criminal justice goals, issues emerge for public policy

discussion. First, are GPS for DV programs appropriate for all of those who can technically be

brought under their purview? Second, what is the optimal or suitable approach to take with those

to be subjected to court-imposed liberty restrictions? Third, how should the program incorporate

the victim in the definition of its overall mission and everyday operations? Ideally, deliberations

about how to design pretrial GPS for DV programs will be undertaken with an understanding of

the needs and situations of DV victims, on the one hand, and the rights and interests of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xxi

Erez, Ibarra, Bales, & Gur

defendants, on the other. A balanced approach would consider victims’ welfare (including their

safety) as well as defendants’ rights (to due process).

Agencies that proceed to establish programs without careful consideration of the issues

risk encountering the problem of mismatch (e.g., enrolling “non-hardened” abusers into GPS for

DV programs that are organized to manage high-risk offenders). Conversely, programs designed

to enroll low grade “abusers” (e.g., those facing a first arrest) or those whose charges stem from

a family disturbance that is of a different nature (e.g., mother-daughter fights), may at times

admit offenders who are considered a serious risk into a non-rigorous supervision regime. To the

extent that programs seek to have a consistent set of rules and restrictions by which defendants

must abide, and have a mix of clients across the spectrum of DV arrestees, a series of misfittings

may manifest themselves in logistical terms for both staff and defendants. Defendants may also

experience emotional and psychological effects in such circumstances. Common scenarios in

which logistical problems emerge include the following: having rigorous supervision regimes

that result in officers spending time with low-risk instead of concentrating on high-risk

defendants, subverting defendants’ employment by requiring in-person office visits during

business hours, and having clients obtain employment verification that deters employers from

hiring applicants. The emotional and psychological impact of the program may be found in the

defendant who becomes demoralized when he can no longer work overtime because of program

inflexibility regarding work hours or is repeatedly rejected on a job search due to restrictions on

mobility.

Agencies considering establishing or improving a GPS for DV program implicitly or

explicitly make decisions about the extent to which victims will be central or peripheral to their

operations. How the victim’s role in the program is defined has implications for whether victims

are viewed and treated as an asset or liability. Irrespective of whether victims are central or

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xxii

Erez, Ibarra, Bales, & Gur

peripheral, viewed as assets or burdens, GPS for DV programs should be mindful of possible

discordance between victims’ expectations or understandings and the program’s actual

capabilities and practices. Absent such foresight into victims’ expectations for program

performance – including consideration of how justice system personnel create or influence such

expectations – victims are likely to experience frustration, loss of confidence in the system,

disappointment, fear, a false sense of security, and in the worst-case scenario, the victim’s safety

can be seriously compromised.

An individual client’s risk level and the agency’s program details, rules, and restrictions

should be logically connected. However, because such an approach might ultimately preclude

placement of arrestees who could benefit from the program – or deny victims the indirect but

tangible benefit of defendants’ being on GPS – the alternative is to design GPS for DV programs

that are comprised of graduated degrees of restrictiveness, based on client’s risk levels, whether

these are understood to be fixed or changing. Programs should also strive to establish criteria for

including DV defendants that are derived from an understanding of the dynamics of DV, rather

than having eligibility as a default assignment made in an effort to reduce jail overcrowding, or

as a means of solving indigent defendants’ problems with raising bail. Ad hoc approaches may

address various exigencies encountered by criminal justice agencies, but they ultimately are not a

substitute for developing initiatives that thoughtfully and simultaneously address the problem of

intimate partner violence, the interests of victims, and the rights of the accused.

GPS for DV programs are organized to either make victims central or peripheral to their

focus and operations. The decision to organize a GPS program in either direction will likely

affect the nature of the agency’s relations with the victim, including the quality of the

information that the victim and agency share, and the assistance that the victim receives during

the period when the defendant is enrolled in the program. However, irrespective of whether the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xxiii

Erez, Ibarra, Bales, & Gur

approach is victim-centric, the program should be based on an understanding of the dynamics of

DV, rather than utilizing the GPS program as a way of handling non-DV related problems (e.g.,

jail overcrowding). A DV-focused identity would recognize the importance of accrued expertise

in properly managing DV cases, whether the programs are victim-centric or not.

Maintaining good communication between agency and victims is paramount for all GPS

for DV programs. Personnel should be watchful for possible discrepancies between victims’

expectations for program performance and the program’s actual capabilities and practices.

Programs that are not familiar with victims’ expectations risk engendering a sense of frustration,

fear, and loss of confidence in the system, or the victim can develop a false sense of security.

Providing accurate information to victims about the capabilities and limitations of the GPS

platform in use is critical for victim welfare and safety. Victims who are correctly informed

about the absence of protection may not feel safe, but their actual safety will thereby be

enhanced as they now take precautions that are congruent with how defendants are actually

supervised. As it is not always possible to provide victims with the resources that they should

optimally have, training victims on how to do safety planning is essential.

Solicitation of victim feedback about agency standards and practices is important to a

program’s effectiveness, especially during its learning phase. Incorporating victims’ feedback

empowers them to share individualized concerns, transforming victim input into a catalyst for

agency innovation. In turn, the newfound program flexibility that is thereby encouraged enables

otherwise reluctant victims to access the program, with defendants spending less time in jail as a

byproduct.

Learning from mistakes, misunderstandings, blind spots, and limitations is critical to a

program’s continued improvement. Such improvement entails staying abreast of technological

innovations, becoming familiar with the situations of defendants and victims, and developing

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page xxiv

Erez, Ibarra, Bales, & Gur

greater coordination among all relevant stakeholders (e.g., monitoring agency personnel, judges,

victim advocates, attorneys, shelter workers, and police). Continuous training and program

refinements can address a wide range of issues. Some reforms may entail financial resources,

such as updating technology, but others will require no cost or even result in savings to the

agency (e.g., partnerships with victim service organizations). Promoting greater understanding

among all stakeholders within the local justice system of the purpose and value of GPS for DV

programs may ultimately spearhead community efforts to develop a coordinated response to

domestic violence, with GPS for DV programs being integrated into it.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 1

Erez, Ibarra, Bales, & Gur

Chapter 1: Introduction

A. Statement of the problem

The present study examines the implementation and effectiveness of Global Positioning

System (GPS)-based monitoring technology in programs designed to address domestic violence

(DV)/intimate partner violence (IPV).

Currently, public safety and criminal justice agencies

utilize GPS tracking for various offenders, predominantly those convicted of sex offenses, but

also for those likely to violate their terms of probation and parole (e.g., Bales et al., 2010). Many

jurisdictions also utilize GPS to monitor DV defendants. GPS for DV programs first emerged in

the 1990s, but their numbers have since been steadily increasing. GPS for DV programs are

created in response to local demand, court initiative, or state legislation, often prompted by an

incident in which an intimate partner was harmed or killed by an arrestee who had been released

by a court on bond, and are usually directed at bolstering protection orders. Since 2000, twenty-

one states

and the District of Columbia have enacted legislation mandating or recommending

that justice agencies employ GPS to protect abused women under specific conditions; several

other states are in the process of considering such legislation. The trend is toward increased

adoption of GPS for DV offenses, making the present research a timely endeavor.

For DV programs, GPS is typically used in the context of the post-arrest or pretrial

period, and this is the circumstance in which these kinds of programs were evaluated in the

present project. This study extends a previous investigation of radio frequency (RF)-based

monitoring programs for DV (Erez, Ibarra, and Lurie, 2004; Ibarra and Erez, 2005; Ibarra, 2005;

Erez and Ibarra, 2007), denoted with the term “bilateral” to distinguish it from mere house arrest-

based RF programs (see below). The field of community corrections has since shifted to using

GPS technologies for handling DV cases. GPS tracking promises to enhance victim safety

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 2

Erez, Ibarra, Bales, & Gur

compared to bilateral RF-based EM programs

(hereafter “bilateral RF”). Its advantages over

bilateral RF, in terms of versatility, broadened detection range, and multiple zone coverage, for

example, are demonstrable, as is its unique ability to keep track of the offender’s whereabouts

beyond the home in real time.

GPS monitoring technology operates on the principle of “geo-

fencing” – it can be programmed to establish multiple and potentially unlimited “zones of

exclusion and inclusion”

(Crowe et al., 2002), including the victim’s workplace, house of

worship, and children’s school(s). Bilateral RF monitoring technology ordinarily results in only

one protected “geo-zone” (hereafter “zone”), namely the perimeter of the victim’s home,

typically with a detection range of 500 feet. By contrast, GPS exclusion zones can be

programmed to extend to entire counties, states, and regional areas, which means that the victim,

or law enforcement, can have more time in which to respond to a zone violation by the offender.

RF monitoring facilities are limited in the kinds of determinations they can make of an

offender’s whereabouts – they can confirm

his presence or absence at his and the victim’s

homes, but they cannot determine the offender’s location when he is away from home or his

pattern of movement (see also Conway, 2003).

In situating the present study’s focus on GPS for DV cases at the pretrial stage, this

chapter first reviews the relevant literature and research findings; it then presents hypotheses and

questions derived from a consideration of the state of the field, which the body of the report will

examine based on the empirical data that were collected.

B. Literature review

DV in an intimate partner relationship often involves lengthy and severe abuse

(Rennison, 2003; Tjaden and Thoennes, 2000a; 2000b). Nevertheless, reporting it to authorities

is a step that victims do not take lightly. Typically, women

report it only when they have

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 3

Erez, Ibarra, Bales, & Gur

reached an “enough is enough” tipping point (Fischer and Rose, 1995), after a lengthy period of

victimization during which they often will have been subjected to repeated abuse, such as

intimidation, harassment, stalking, and/or physical assault (Tjaden and Thoennes, 2000a; 2000b).

Taking action against DV by reporting it to the police or attempting separation often places

women at a higher risk of renewed violence. “Separation assault” (Mahoney, 1991) is a well-

known phenomenon in DV dynamics, as is the stalking of estranged partners. Block (2003)

found that three-fourths of female homicide victims and 85 percent of women who experienced

severe but nonfatal violence had left or tried to leave their batterers in the prior year.

Historically, it has proven difficult to protect women in abusive relationships when they

report abuse or leave their abuser. The context and dynamics of DV render the protection of its

victims a continuous challenge to the justice system, one fraught with logistical problems and

contradictions (Jordan, 2004; Worden, 2000). Courts use protection orders to secure the safety of

women who step forward as victims of DV, restricting the abuser from contacting the victim in

person or through other means (e.g., by telephone, email, social media, SMS). Although such

orders can be beneficial, compliance is difficult to enforce (Harrell, 1993; Harrell and Smith,

1996). Protection orders are generally associated with a subsequent decrease in abusive behavior

by batterers (Holt et al., 2002; Holt et. al., 2003), but they can be less effective when the abuser

has a history of violent offenses (Keilitz et al., 1998; Jordan et al., 2010), when the victim is

socially situated in particular ways (such as being unemployed, see Burgess-Proctor, 2003), or

when the offender simply does not respect protective orders (Erez and Belknap, 1998; Jordan,

2004; for a review of the factors that contribute to compliance with protection orders, see

Benitez, McNeil, and Binder, 2010). Ironically, upon receiving protection orders some abusers

become so angered that they seek out their victims for renewed assault (Erez and Belknap, 1998;

Shim and Hwang, 2005).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 4

Erez, Ibarra, Bales, & Gur

Preventing or deterring contact is difficult in DV cases for additional reasons. The

offender is typically familiar with the victim’s routines and social relations, which means that

she often has concerns about being intercepted or ambushed, even when she is away from her

home. He knows where she works, where the children go to school, the stores at which she

shops, the residences of her friends and family relations, her home, cellular, and work telephone

numbers, as well as her travel routes. Knowledge of her routines furnishes the abuser with

numerous opportunities to harass, stalk, intimidate or assault the victim in spite of protection

orders (Erez, Ibarra and Lurie, 2004). A particularly significant concern is that a defendant will

seek out the complaining witness to request or demand that she retract or amend her complaint,

or to intimidate the victim into failing to appear in court. Reporting such incidents is frustrating

for victims when a “he said, she said” situation presents itself without definitive proof of the

protection order violation (Hart, 1993; Erez et al., 2004; Hartley, 2001). A recurrent

complicating factor emerging from the dynamics of DV is that many abused women – still

emotionally or economically attached to the batterer, or sharing parenthood of children – want to

have (non-violent) contact with the offender and are reluctant or ambivalent about facilitating the

state’s prosecution (Dichter et al., 2011). At times, therefore, they will allow their abuser to

contact them despite the presence of a court order barring such association, or even seek him out

at his place (Erez et al., 2004).

Previous research found that in DV cases, courts commonly used bilateral RF during the

post-arrest/pretrial release phase rather than as a condition of sentencing (Erez et al., 2004). A

rationale often cited by court officials for placing offenders on electronic supervision is that it

will strengthen protective orders and “give them teeth;” that is, will force restricted parties to

take them seriously or suffer adverse consequence. The use of bilateral RF in DV cases presumes

that offenders under “no contact” orders, knowing that they cannot approach specific geographic

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 5

Erez, Ibarra, Bales, & Gur

zones without remote detection, are less apt to seek out the victim, notwithstanding a history of

violating such orders without EM. This premise is supported by previous research on bilateral

RF-based programs (Erez et al., 2004), which found that relatively few incursions into the

victim’s home region were attempted by defendants, as per monitoring facilities’ records or

reports from victims participating in EM programs.

Personnel classified the few defendant

incursions into the victim’s home region as non-serious “unintentional” or “informational”

violations. The former were essentially “drive-by penetrations”;

the latter amounted to cases in

which defendants were curious about whether, for example, the woman had a male guest at her

home, or the defendant was testing the sensitivity of the monitoring equipment vis-à-vis the

perimeter around the victim’s address. Over a five-year period in the more active of the two sites

that were studied, only one home region incursion was attempted that actually involved victim

endangerment; the less active site did not report any such contacts over an even longer period of

time. Face-to-face contact violations were the most common type of non-home-based violations

reported by victims in Erez et al. (2004), but few of these face-to-face violations were described

by victims or program staff as “serious.”

Victims in the programs who could contrast

defendant compliance with no-contact orders buttressed by EM and non-EM circumstances

observed a marked difference – before he would routinely ignore the protection order, now he

was rigorously observing it – and they attributed this contact-free period to the effectiveness of

the technology (Erez and Ibarra, 2007).

Furthermore, research found evidence that participation in bilateral RF programs had an

impact on both victims and defendants. Participating women described being empowered and

having an improved quality of life during their involvement with the program, which would

likely have been more challenging in the face of the uprooting effects associated with relocating

to a battered women’s shelter (Erez and Ibarra, 2007). For example, by remaining in their own

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 6

Erez, Ibarra, Bales, & Gur

homes following the offender’s arrest, participating victims (and, if present, their children) were

able to resume a semblance of a “normal life,” without fear of assault, ambush, or vandalism, and

beyond the controlling presence of an abuser, something that most people take for granted, but

which many DV victims cannot (Erez and Ibarra, 2007). Victims who tested police response

times by manually triggering alarms observed that “help” arrived far more quickly than it had in

the past when distress calls were placed through 9-1-1. These shortened response times increased

victim confidence in the system’s ability to protect them. Further, because the victim had to file a

complaint as a prelude to program entrée, she made herself visible to a broad array of persons

whose services she could seek in the event of need. The victim was not restricted to calling a

police dispatcher lacking familiarity with her case, or seeking solace and protection in a personal

support network (although women in the program reported a flowering of their social life during

their tenure). She could mobilize a variety of criminal justice professionals with personal

knowledge of her identity and the intricacies of her abuse history, personnel with whom she

developed rapport and who were viewed as accessible and helpful. During these interactions

between victims and functionaries, the latter encouraged the former to persevere with their case.

Indeed, participating women were observed to be more likely to testify and less likely to recant

or fail to appear for court proceedings than non-participating victims (Ibarra and Erez, 2005),

indicating that participating defendants had less sway with victims than often occurs in DV cases

(Hart, 1993; Worden, 2000). Perhaps as a result of victims’ increased willingness to testify,

dismissal rates were lower for participating than for non-participating defendants (Ibarra and

Erez, 2005).

Although the nature of EM technology is important in that its capacities and limitations

will circumscribe its reach and application, of greater significance is the program in which the

technology is anchored (Erez et al., 2004). Each bilateral RF program is animated by a general

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 7

Erez, Ibarra, Bales, & Gur

outlook and by particular historical origins, both of which define broad purposes that are

reflected in a number of structural and programmatic features, including the professional

orientations of personnel, understanding of DV dynamics, eligibility criteria, expectations for

participants’ conduct while they are enrolled in the program, and directions to personnel on how

to interact with participants and enforce program expectations. Taken together, these various

considerations organize each program’s definition of success (Erez et al., 2004). In addition, EM

programs can be influenced by factors related to the “cycle of violence” identified in the research

literature, and the notion that such EM initiatives might function as an intervention in an abusive

interpersonal relationship (Erez et al., 2004; Ibarra and Erez, 2005; Ibarra, 2005). For example,

one of the sites in the bilateral RF study drew its cases from parties who were both in current and

former relationships, while the second site primarily targeted parties whose relationship had

undergone an irrevocable breach (e.g., living apart, legally separated, in various stages of divorce

proceedings). Personnel at the latter site went to great lengths to be certain that the victim wanted

nothing to do with the defendant before ordering the equipment’s installation, while personnel in

the former site saw much merit in ordering the equipment when the victim was ambivalent about

ending her relationship with the defendant. Because this program enrolled many participants

whose lives were still emotionally, materially, and socially enmeshed, program personnel

approached interactions with male participants with a concern to detect “risk factors” in their

living environments that might incline them to seek contact with the victim, and sought frequent

encounters with them (e.g., though office and home visits) in order to gauge the presence of “red

flags” indicative of elevated danger to the victim (Ibarra, 2005). Furthermore, this program was

far more onerous in the imposition of liberty restrictions on male participants. For instance,

defendants were required to make weekly office visits during which urine screens were

administered, and they were required to submit to surprise home visits. The second site did not

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 8

Erez, Ibarra, Bales, & Gur

have such rules and policies in place, deeming them too intrusive or burdensome for non-

convicted persons, and was far more flexible in setting curfews to accommodate exigencies of

clients’ lifestyles and work schedules, even having a “hands off” attitude toward defendants’

whereabouts while not working, provided they returned home before 11PM (Erez et al., 2004).

Differences in the intensity of supervision appear to be the most salient experiential

dimension for the defendants who participate in bilateral RF programs as controlled parties

(Ibarra and Erez, 2005). Men participating in the more intensively structured program in the Erez

et al. (2004) study, for example, expressed anger and resentment about the “pains” associated

with a pretrial EM program (cf. Payne and Gainey, 1998). Citing their status as defendants, they

reported feeling they were being treated as if they had been found guilty without trial and were

now living under a demoralizing and unjust regime. Their lives became transparent; their

everyday practices open to scrutiny, correction, and sanction. These men reported having been

“tricked” into consenting to enroll, because the extent of supervision was not explained to them

beforehand; many even mentioned they would have preferred to stay in jail, had they known at

the time what awaited them in the bilateral RF program (Ibarra, 2005). Men from the second site,

the overwhelming majority of who were also defendants at the time of their enrollment, did not

voice these criticisms or concerns. Instead, for the latter, EM was more of an inconvenience

rather than an injustice, notwithstanding the fact that clients were required to subsidize the cost

of administering the program by way of a weekly fee, and that they were required to participate

in EM in order to be released on bond (Ibarra and Erez, 2005).

The bilateral RF programs seemed to have an effect on participating men, as measured by

post-program offending behavior documented in records from the more stringent site (Ibarra,

Erez, and Risser, forthcoming) and as indicated by defendants’ self-reported changes in outlook

at both sites (Ibarra and Erez, 2005). The quantitative findings derived from the agency’s

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 9

Erez, Ibarra, Bales, & Gur

records

showed that participation in the RF based program was associated with lower

recidivism rates in the long term (over a two-year follow-up period) than were detention in jail,

conventional EM (i.e., no victim participation), or release on bond without pretrial supervision,

pending disposition (Ibarra et al., forthcoming): Multivariate (MANCOVA) analyses

demonstrated significant differences during the follow-up period (post adjudication/supervision)

in DV arrests by enrollment condition.

Post-hoc analyses revealed significantly fewer post-

supervision arrests during the second year of follow-up for those in the bilateral RF condition

relative to those who had remained in jail during the pretrial phase, as well as those who were

released on bail without conditions or who were in a traditional EM program (without victim

participation). When intimate partner violence variables were combined into a composite, there

were significant differences in post-supervision arrests for the composite in the second year.

Post-hoc analyses revealed that those in the bilateral RF condition demonstrated significantly

fewer post-supervision arrests than those in the jail condition.

In-depth interviews with men who were placed on the program at the more stringent site

shed light on possible reasons for reduced repeat victimization (post-program arrests) by

participants in the bilateral RF program (Ibarra and Erez, 2005). Defendants spoke about feeling

infantilized by having to live with their parents or relatives. They also described feeling

powerless vis-à-vis their female partners as a result of not being able to get in touch with them to

discuss the case or to otherwise reassert an element of control over their fate, while their female

partners, on the other hand, were able to contact the abusers when they so desired and on their

own terms. Defendants at the stringent site recalled the bilateral RF experience as particularly

onerous, as something that they wanted to avoid at all costs, so renewed contact with the victim

was not worth the risk of new entanglement with an EM program (Ibarra and Erez, 2005). On a

more positive note, many men felt that participation in the bilateral RF program provided their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 10

Erez, Ibarra, Bales, & Gur

lives with “structure” that they otherwise lacked, helped them make a “fresh start,” and

impressed upon them that they could function without the victim in their lives (Ibarra and Erez,

2005).

A criticism that has often been made of diversion programs – that they engender forms of

“net widening” (Cohen, 1985; Decker, 1985) – may also apply to pretrial bilateral RF programs,

especially those that entail more extensive liberty restrictions. Net widening refers to the idea

that alternatives to penal strategies represent an extension of penal control by the criminal justice

system over civil society (Austin and Krisberg, 1981; McMahon, 1990). Net widening occurs

when people are brought into the system who would have otherwise avoided or more quickly

exited it, if not for the existence of, for example, a diversion program; through the creation of

new agencies and services that effectively supplement rather than replace the original set of

control mechanisms; and by the creation of special programs that are more intrusive than is the

justice system’s ordinary handling of such cases. The relevant notion here is that individuals

come under justice system scrutiny when they would not be subject to extraordinary supervision

or surveillance while under pretrial release, but for the existence of the EM program, and that

such social control creates possible legal vulnerabilities for those offenders that they would not

face absent the monitoring regimen (Ibarra and Erez, 2005).

There is support for the idea that the more intensive bilateral RF program examined by

Erez et al. (2004) exhibits the kind of “net widening” that critics have observed elsewhere in

diversion programs and other intermediate sanctions (Ibarra and Erez, 2005).

Net widening is

pertinent because DV cases usually have high non-prosecution and dismissal rates relative to

other violent offenses (Fagan, 1995). The net widening thesis would suggest that participants in

bilateral RF programs are likelier to remain entangled in criminal justice processing than non-

participants. Although all DV arrestees are arraigned prior to release, for example, only some of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 11

Erez, Ibarra, Bales, & Gur

them will end up on bilateral RF, and the latter group will tend to have a different experience

with the criminal justice system upon release, owing to the greater transparency of their lives

during the post-arrest period and the interest that the prosecution will take in their cases (Ibarra

and Erez, 2005).

Administrative data analyzed by Ibarra and Erez (2005) support the net-widening thesis.

Dismissal rates for bilateral RF participants with pending cases (i.e., defendants) over a two-year

period were significantly lower than for non-participants: participants had a 14% dismissal rate,

compared to a dismissal rate of 44% for men who remained in jail during the pretrial period

(Ibarra and Erez, 2005). The average number of days that participants spent in the program was

higher (mean = 48, median = 29) compared to the number of days that non-participants spent in

jail (mean = 30, median = 20). In short, bilateral RF participants’ cases were likelier to last

longer and culminate in verdict-based dispositions. The lower dismissal rates for participants

apparently reflected the continued participation of the prosecuting witnesses in the cases, which

in turn resulted in defendants seeking continuances, either to build stronger defenses, or as a way

to wear down complainants’ resolve to remain involved with prosecution efforts (Ibarra and

Erez, 2005). Defendants who did not participate in the bilateral RF program were likelier to have

their cases dismissed and be under the purview of the court for a briefer period of time.

Hence,

bilateral RF defendants (especially in stringent programs) contended with legal liabilities that

comparable DV defendants did not.

Although men historically have been the majority of clients subjected to court ordered

interventions for DV, and hence the likeliest recipients of net widening effects, arrests of women

for domestic violence have been increasing over the past three decades (DeLeon-Granados et al.,

2006). Judges have broadened the focus of EM programs for DV beyond violence between

intimate partners to include such circumstances as assaults on or by the third party in a lover’s

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 12

Erez, Ibarra, Bales, & Gur

triangle, or mother-daughter altercations. In addition, although it happens infrequently, men do

seek out protection orders against women (Durfee, 2011). The implication of these trends is that

women are becoming likelier candidates for EM programs as tethered parties, raising questions

about how women fare in programs designed with male batterers in mind (Ibarra and Erez,

2011).

C. Hypotheses and research questions of the present study

The above review demonstrates that bilateral RF programs can be effective in

accomplishing the aims of victim protection, via contact deterrence and lowered recidivism, and

result in an improved quality of life for the victims as well as their greater participation in

criminal justice proceedings than would otherwise be expected. At the same time, extant research

suggests that these bilateral RF programs produce a complicated scenario for defendants:

Although these programs can, for example, assist offenders in “moving on” and developing a

sense of structure in their lives, they can create everyday hardships ranging from emotional stress

to legal vulnerabilities that stem from the acquired transparency of their ordinary activities.

GPS is technically more sophisticated than bilateral RF. It has superior tracking

capabilities, offers increased customizability, and has greater area coverage. However, it is also

costlier

and creates more streams of information for which supervising agencies can be held

accountable (i.e., GPS will create more work and increase potential liabilities for agencies). It is

also not evident that GPS produces superior victim protection or that it has other ancillary

benefits such as improved quality of life and greater victim participation in proceedings. Indeed,

the increased transparency that results from the use of GPS raises issues about possibly increased

legal liabilities, and the consequent net-widening impact on defendants. Thus, the shift toward

using GPS monitoring technologies within the context of DV cases raises several issues that this

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 13

Erez, Ibarra, Bales, & Gur

study aims to address. In addition to discussing issues related to effectiveness, matters of

practical, legal, psychological, social, and material impact on defendants, victims, and the

criminal justice system as a whole are reported.

In light of this review of extant literature, the current study is designed to examine the

following hypotheses and questions, using quantitative and qualitative data (see “Methodology”

section).

1. Hypotheses

i. Defendants in GPS for DV programs (hereafter GPS defendants) have

lower rates of re-arrest during the pretrial period compared to non-GPS

defendants.

ii. GPS defendants have lower rates of re-arrest during the post-disposition

period compared to non-GPS defendants.

iii. GPS defendants have fewer program violations compared to RF-based

defendants.

iv. Verdict rates are higher and dismissal rates are lower for GPS cases

compared to non-GPS cases.

2. Research questions

i. What safety-relevant effects does GPS monitoring have on victims and

defendants, both in the short and long terms? Is there evidence that clients

enrolled in GPS programs avoid initiating direct contact with their alleged

victims? Do victims develop a “false sense of security”?

ii. What are the advantages and disadvantages of different approaches to

organizing a GPS-based program for DV? Are these different approaches

correlated with various outcomes, including recidivism, client and victim

satisfaction?

iii. What changes in behavior (everyday practices, justice system related

behavior) and outlook (level of confidence in the justice system) are

reported by victims, relative to their behavior and outlook during prior

occasions when reported abuse did not lead to abusers’ participation in a

GPS program?

iv. Are there unique attributes of cases, individuals, or relationships for which

application of GPS may require special considerations (e.g., DV cases

among non-intimate partners; DV cases in which the defendants are

women)?

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 14

Erez, Ibarra, Bales, & Gur

v. Does the shift to GPS platforms represent an advance over (bilateral) RF

in the ability of courts and their proxies to supervise clients within their

purview?

vi. Do the enhanced capacities for tracking offenders result in enhanced

(perceived and objective) safety and quality of life for victims?

vii. How do supervising officers use the increased information that is made

available through GPS technology?

viii. How do various stakeholders (e.g., victims, defendants, and criminal

justice professionals) understand the potentials and limitations of GPS

monitoring?

ix. What are the kinds of problems that repeatedly come up in the

administration of GPS programs for DV?

x. How are defendants personally affected by participation in GPS for DV

programs?

xi. Are there unintended consequences (legal or otherwise) associated with

GPS for DV programs vis-à-vis the parties who are directly or indirectly

involved with it?

xii. What do national patterns reveal about how agencies across the U.S. are

applying GPS technologies in DV cases during pretrial?

xiii. How do practitioners in agencies across the U.S. view GPS for DV (e.g.,

in terms of their caseload and effectiveness)?

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 15

Erez, Ibarra, Bales, & Gur

Chapter 2: Methodology

The study is comprised of three parts: First, we conducted a web-based survey of pretrial

service agencies in the United States to examine how electronic monitoring technologies are

used in their operations, with a specific emphasis on applications of GPS to address DV cases.

Second, to study GPS systems’ effectiveness in deterring recidivism in the short- and long-terms,

we applied a quasi-experimental design to the analysis of quantitative data collected by large and

established programs in three jurisdictions – representing geographical diversity (from the

Southern, Western, and Midwestern U.S.) as well as different program approaches (e.g., levels of

restrictiveness and flexibility) – each using GPS to monitor defendants’ compliance with

protective orders in IPV cases during the pretrial period. Third, questions about the

implementation, effectiveness, and impact of GPS were addressed through in-depth interviews

with parties involved with programs at six cooperating agencies (victims, offenders, justice

personnel, and social service providers).

A. Web-based survey

To situate the quantitative and qualitative findings within a national context, the first

prong of the study sought the opinions and experiences of personnel at agencies using GPS-

based monitoring programs to address DV during pretrial. An EM consultant, Peggy Conway,

provided a list of 278 organizations, which was supplemented by having a graduate assistant

contact state public information officers, departments of corrections, and other relevant

authorities, as well as vendors of the technologies. Trained doctoral students in Criminology,

Law, and Justice began placing phone calls to the agencies in late July 2008 to gather specific

contact information. Following an IRB-approved script, they asked to speak to someone in a

managerial position who could answer questions regarding the agency’s GPS for DV program. A

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 16

Erez, Ibarra, Bales, & Gur

brief screen was used to determine if each agency met the criteria for the online survey. When

appropriate, contact information for the agency representative was obtained and recorded,

including an email address where the survey would be sent in the form of a hyperlink embedded

in the email message. The contact was then emailed a link to a survey administered through the

web-based SurveyMonkey®

data collection site, and invited to participate in the study.

The number of survey respondents had reached 34 by the end of June 2009, when

representatives from two pretrial organizations

offered to help disseminate recruitment

material for the NIJ-funded survey.

With help from the umbrella pretrial organizations, and

follow-up phone calls as necessary, 616 respondents participated in the online survey.

Of the

616 respondents, 393 (63.9%) indicated that their organization uses EM to supervise defendants.

A subset of 149 agencies consistently indicated that they use GPS to supervise DV defendants

and these respondents were directed to questions specifically pertaining to their agency’s use of

GPS for DV.

1. Topics and procedures

The web-based survey posed Likert scale, mutually exclusive, and open-ended questions

aimed at documenting a number of factors related to the implementation of EM programs in

general (yielding 616 respondents) and GPS for DV programs more specifically (resulting in 149

respondents). The questions addressed program history and current attributes, issues of

technology, vendors, and training, respondents’ views about and experiences with programs

utilizing GPS for DV during pretrial, issues pertaining to victims, and defendant-relevant factors.

The survey was available to respondents from February 2009 through November 2009,

and reminders were periodically sent to potential participants who had been contacted by email

but had yet to respond. The web-based SurveyMonkey® data collection site allows survey

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 17

Erez, Ibarra, Bales, & Gur

administrators to download results in multiple formats. Results were downloaded from

SurveyMonkey® in Microsoft Excel (.xls) and comma-separated (.csv) formats, and saved. The

excel spreadsheet was then used to create an SPSS (PASWStatistics 18.0) database, a process

that resulted in over 200 variables. Following data processing and cleaning, descriptive and

bivariate analyses were conducted.

B. Quantitative assessment of program effectiveness (the quasi-experimental component)

Three GPS-using jurisdictions participated in the quantitative component of the study,

referred to as the Midwest, West, and South sites. Agencies participating in this part of the study

were all mature

programs using active

GPS platforms. We selected these agencies after

consulting with a variety of agencies nationwide

about their program design, caseload size, and

data quality and accessibility. The agencies selected for the impact study had been operating a

GPS program for DV defendants for at least three years, and offered access to offender data for a

period of at least one year following defendants’ exit from supervision related to the instant

offense (i.e., the incident that resulted in the charge(s) that qualified them for program

enrollment). The selected sites varied in terms of overriding philosophy, enrollment criteria, and

staffing structure, enabling inferences to be drawn about the impact of variations in how

programs are organized. Because the agencies exercise varying degrees of restrictiveness and

transparency in their supervision practices, site differences may account for some outcome

differences (both in the short and long terms).

The method of data collection in the three sites also varied: In the Midwest and West

sites, data were collected using schedules designed for the current study (see data instruments in

the Appendix), whereas in South, administrative data were extracted from a jail management

system and supplemented with data from the pretrial agency’s record system. As discussed

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 18

Erez, Ibarra, Bales, & Gur

below, the study examines GPS program impact in the short term (during the enrollment period)

and in the long term (during a one-year follow-up period).

1. Participant selection criteria, methods, and sampling

This section describes the rationales and procedures used to construct the comparison

(treatment and control) groups in the quantitative study that examined program effectiveness. At

two of the three sites (West and Midwest), referral by a judge is a predicate for enrollment in a

GPS for DV program. However, due to various legal, technological, and practical or program

design considerations, a substantial number of defendants (57% in West and 73% in Midwest)

who were referred to the GPS program for possible hook-up did not actually enter the GPS

program at any point before the conclusion of their case.

In Midwest, referred defendants may have been assigned to one of three alternative

pretrial conditions: jail, bond without supervision, or RF-based EM. An inability to establish a

separate residence apart from the victim, or obtain a landline telephone, resulted in the defendant

remaining in jail. A successful motion to have the GPS requirement lifted resulted in the

defendant’s placement into RF-based EM or release on bond without supervision. A victim’s

withholding of consent could result in the defendant being reassigned to any one of the three

other groups. Problems with maintaining a GPS signal at either the defendant’s or victim’s

residence could result in the defendant’s reassignment to either RF-based EM or release on bond

without supervision.

In West, referred defendants may have been assigned to one of two alternative pretrial

conditions: jail or bond without supervision (West did not have an RF-based EM program for

DV). Those who posted bond and got their attorney to waive the GPS requirement were released

without supervision. Some of these defendants were of greater means, but others were simply

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 19

Erez, Ibarra, Bales, & Gur

facing more serious charges, and hence hired private legal counsel. An inability to establish a

separate residence apart from the victim resulted in the defendant remaining in jail. Also, in West

if the victim had moved some distance from the jurisdiction (e.g., out of state), the defendant

might have the GPS requirement waived, in which case he would be released without

supervision.

Inability to post bond, a problem endemic to pretrial release, also affected GPS

enrollment in various ways. This hurdle may be mitigated by the existence of GPS programs in

that defendants may be offered, as an inducement, lower or zero bond amounts, provided that

they consent to GPS tracking. Inability to pay program fees was not an issue at the Midwest

agency because it does not charge defendants any fees, but it may have been an issue for some

defendants at the West agency, even though it employs a sliding scale to calculate defendants’

fees (a sliding scale is also used in the determination of GPS and RF program fees in the South

site.) Sliding scales do not necessarily eliminate the burdens that fees may cause a defendant or

his household, especially if they are based on absolute cost to income ratios, and not on the

defendant’s budgetary circumstances.

In sum, because all West and Midwest defendants in the sample were referred for a GPS

evaluation, they were initially perceived as sufficiently dangerous to warrant the extra level of

supervision provided by GPS monitoring, irrespective of their actual eventual

participation. Therefore, the West and Midwest samples were drawn from a database that was

comprised of court referrals, constituted as 1) a treatment group – all defendants who were

referred to a GPS for DV program and enrolled; and 2) a control group – a time ordered random

sample drawn from all referred cases that were not enrolled in a GPS program during the pretrial

period, but who either remained in jail for the duration or posted bond and were released. In

addition, at the Midwest site, some GPS referrals who were not hooked up for reasons cited

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 20

Erez, Ibarra, Bales, & Gur

above were reassigned to a traditional EM (i.e., RF) program (which consists of house arrest with

curfew restrictions, staffed by officers who have general caseloads, rather than DV caseloads).

The patterns of assignment and participation at these sites enabled the construction of a quasi-

experimental design, comparing GPS cases with a matched sample of non-GPS cases. Such an

approach is the best approximation of experimental design in the context of criminal justice.

Cases were drawn over multiple years in both of these sites. In the Midwest site, cases

were drawn from archived judges’ orders (which mandate that the defendant be evaluated for

enrollment into a GPS for DV program) over a two year period (2518 total referrals between July

2006 and July 2008), yielding the following samples: 561 GPS cases (or 100% of defendants

who were referred and hooked up) and 1521 non-GPS cases who comprised a random sample of

78% of defendants who were referred but not hooked up. The non-GPS cases were distributed as

follows: a jail group (N = 549 or 50% of the referrals who remained in jail), a bond group (N =

507 or 56% of the referrals who were bonded out), and an RF-based EM group (N = 465 or 87%

of the referrals who were reassigned to an RF program).

In the West site, cases were drawn from the GPS program's database, into which

information from judges’ orders (including referrals) had been entered, between October 2001

and January 2007, yielding the following samples: 500 GPS cases (or 100% of defendants who

were referred and hooked up) and 500 non-GPS cases (or 75% of defendants [N = 670] who

were referred but not hooked up, i.e., who remained in jail or were bonded out without spending

any time on GPS).

In the third site, the South, the comparison groups were generated using different

procedures that allowed for a non-equivalent control group design by selecting defendants who

had "domestic violence" listed in their charges. In Midwest and West, the basis for equivalence

was a concern with whether the defendant would stay away from the victim, and the potential for

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 21

Erez, Ibarra, Bales, & Gur

him to harm her, as determined by a judge. Judicial decisions resulted in lists of defendants

referred to GPS, from which the Midwest and West comparison groups were drawn.

Generally speaking, DV defendants in South are ordered to GPS or RF based on a

decision by the judge regarding their risk level. The primary difference between RF and GPS

defendants at bond setting appears to be that cases ordered to GPS raise greater concerns for

judges about the defendant trying to contact the victim or her family members. DV defendants

who are considered especially dangerous overall are given high bond amounts that make it

difficult for them to leave jail at all. DV defendants who are seen as low risk may be released on

their own recognizance or with a relatively low bond. Thus, the “treatment group” of GPS, and

the comparison groups of RF, Bond (without EM supervision), and Jail, are viewed by bail

setting judges as posing different levels of risk.

In addition, South defendants who are seen as low risk, but who cannot make a low bond,

may end up in RF or GPS, in lieu of posting a monetary or surety bond. These late entrants are

likely to be reassigned to either GPS or RF as the result of a defense attorney’s motion to have

initial terms of bond modified, or a judge’s independent observation that the defendant has spent

an inordinate amount of time in jail.

To select cases for the evaluation study in South, all jail booking records from January

2002 through June 2011 were extracted. This file included a variety of arrestee demographics,

the date of booking, charging information, bond amount, and other relevant measures.

Additionally, the JMS identification number (JMS Number), which is a unique number assigned

to each booking event, was provided, as was the Offender Control Number, which is unique

number assigned to each new arrestee and included in records associated with all subsequent

booking events for the same individual. The booking file contains a record for each arrest charge

for each JMS Number. In addition, all jail release records from January 2002 to December 2009

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 22

Erez, Ibarra, Bales, & Gur

were extracted. This file contains the same information as the booking file, except that it also

includes the date of jail release and the type of release.

A file containing cases placed on pretrial release supervision was created that included a

variety of variables relating to the arrestees’ placement and removal from pretrial release

supervision. A separate file containing similar data for offenders placed on electronic monitoring

was also provided, which included the date and type of EM placement (Global Positioning

System or Radio Frequency) and the outcome (e.g., completed, violated) of EM enrollment.

The dataset to assess the effectiveness of EM for DV cases in South was constructed as

follows: First, all jail bookings that occurred from August 2007 through December 2009 were

selected in which DV was one of the arrest charges. Second, this file was matched to the jail

release file using the JMS number to capture the first jail release subsequent to the booking date.

Third, this file was matched to the pretrial release monitor supervision dataset to identify cases

placed on supervision. Fourth, this dataset was matched to the monitor dataset that contained

those arrestees placed on electronic monitoring. The resulting dataset was sent to the specialist in

the South site who oversees the agency’s record-keeping to both confirm the accuracy of the data

and to populate missing data based on individual reviews of the case files and electronic records.

The resultant sample of 604 DV defendants in South was distributed across four groups as

follows: GPS = 177, RF = 167, Jail = 42, and Bond without EM = 218.

2. Measures and indicators

As indicated above, the primary aim of the quantitative analysis was to determine the

impact of GPS program participation on defendants over the short and long terms, i.e., during the

pretrial period, and during a one-year follow-up period after the conclusion of supervision related

to the original charges (e.g., after disposition for acquitted and dismissed cases, after jail or

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 23

Erez, Ibarra, Bales, & Gur

prison for convicted cases). Impact was measured in the short term (i.e., during the pretrial

period) by the occurrence of new arrests and charges (for all groups) and program violations (for

those in a GPS or RF program; examples of possible violations include contact attempts, failed

alcohol or drug tests, non-payment of fees, and curfew offenses). In the long term, impact was

measured by the occurrence of new arrests within one year after release from EM supervision or

confinement (i.e., the impact period for acquitted and dismissed cases began with final

disposition; the impact period for sentenced cases with prison or jail terms began following

release from confinement).

Treatment was measured both by exposure to GPS and the number

of days on the program.

Control variables consisted of the following: demographic characteristics (age, gender,

race, Hispanic origin, marital status, etc.), severity of instant offense, offense history, and legal

measures and indicators (conviction for the current offense, type of sentence, time on GPS, etc.);

measures of socioeconomic status were generally not available. The specific measures that were

used in the analysis for each site are discussed in the respective analyses of the data for the three

sites.

The sources (written records, information systems) that were accessed to extract data for

each site analysis are as follows: The Midwest site data set was constituted after consulting EM

agency files, judges’ orders, arrest files (details related to the instant offense), and the

jurisdiction’s computerized records database that collates information pertaining to crimes and

offenders. The West data set was constituted after consulting EM agency files, the jail’s data

base, the misdemeanor court’s information system, LexisNexis for the felony court’s cases, the

state’s corrections information system, and the Federal Bureau of Prisons website. The South

data set was constituted following extensive discussions with jail and community corrections

staff. It included the measures and indicators found in the jail management system (including

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 24

Erez, Ibarra, Bales, & Gur

booking records), and it was complemented by the incorporation of additional sources of data

pertaining to the electronic monitoring and the pretrial program, which was also cross-checked

by the agency’s information system specialist.

The quantitative data from the three sites were processed into statistical packages and

cleaned. The statistical analyses of the data included, for each site, descriptive statistics, and

bivariate and multivariate (logistic regression) analyses. Detailed descriptions of the procedures

and analytic strategies appear in Chapter Three: Findings.

C. Qualitative assessment of GPS-based monitoring

Complementing the web-based survey and quasi-experimental prongs of the study, the

qualitative component included six participating agencies: the West, Midwest, and South sites,

and three additional sites – Southwest, East, and Southeast. The latter were located after

consulting with a national authority on electronic monitoring programs. The additional programs

were spread out geographically and had GPS for DV programs with varying levels of maturity

and comprehensiveness. These were smaller programs and hence did not have the number of

cases required to participate in the impact study. However, they were included to widen the

study’s window on the issues and processes that are associated with GPS for DV programs.

The qualitative component of the study followed an in-depth interview-based research

design. In-depth interviewing is a preferred approach in qualitative research because it enables

investigators to document, in interviewees’ own terms, the issues and problems that concern

them (Rubin and Rubin, 1995). The qualitative data documented stakeholders’ experiences and

perspectives, and enabled an in-depth analysis of how agencies are applying GPS to DV cases.

The in-depth interviews were semi-structured and were conducted with all parties

involved, directly or indirectly, with GPS-based monitoring for DV cases at participating

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 25

Erez, Ibarra, Bales, & Gur

agencies. Such data are essential for shedding light on statistical findings, as well as for revealing

contextual issues and social processes otherwise unobservable through administrative data alone.

Interviewees were drawn from:

a) persons involved in DV incidents, including both former and present defendants (N =

74) and victims (N = 34);

b) social services providers (N = 19) who help victims, including professionals from

private non-profit assistance organizations designed to provide shelter or counseling to

victims of IPV; and

c) justice system practitioners (N = 83), including judges, prosecutors in state or district

attorney’s offices, public defenders, private defense attorneys, and practitioners working

in EM related positions. Among the latter were line staff (e.g., EM or probation officers,

case managers), line staff supervisors (e.g., House Arrest Sergeant, Community Control

Supervisor), agency directors or supervisors (e.g., Director of Pretrial Services, Chief of

EM Unit), and various specialists (e.g., technical support, intake/assessment). Criminal

justice personnel working in victim related services were also interviewed, including

victim advocates or service providers working on the front lines with victims and

professionals holding supervisory positions in agencies providing services to victims.

The first two principal investigators conducted the interviews, which typically lasted

about ninety minutes, although some were shorter than an hour and others lasted longer than two

hours. Focus group interviews were pursued with former victims and offenders/defendants,

usually in pairs or trios, although at times these interviews were done on an individual basis, as

per the interviewee’s preference or turnout. Interviews with personnel were usually done

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 26

Erez, Ibarra, Bales, & Gur

individually or with pairs of co-workers, but in three cases the interviews were conducted with

groups of three or more personnel.

1. Interviewee recruitment and interview procedures

Contact persons at each agency were instrumental in recruiting the individuals who

participated in the interviews. Working from past cases, agency personnel contacted participants

(both victims and defendants) for purposes of recruitment, forwarding an IRB approved

recruitment letter composed by the PIs that explained the purpose of the study and nature of the

interviews. Potential recruits were informed that the interviews were confidential and given

assurances that they were under no obligation to participate, could withdraw from the interview

at any time, and would suffer no harm should they decline to participate. These assurances were

repeated as part of the informed consent process when participants appeared at the conference

room located within the agency that had been reserved for the session. The informed consent

script was read aloud by one of the interviewers, and participating subjects were asked if they

verbally agreed before interviews commenced. Permission to digitally record the interviews was

also sought during the informed consent process. Defendants’ and victims’ names were not

documented in any way. However, in order to be able to summarize the characteristics of the

sample, participants were asked to complete a brief questionnaire on which they could record

their gender, race, age, occupation, and the charges that resulted in the interviewee’s

involvement with the program. Defendants and victims were compensated for their time and

transportation costs, receiving thirty dollars cash for their participation. Because the thirty dollars

in compensation might have been an inducement to financially pressed clients and victims to

participate, the co-PIs emphasized to agency personnel that they should ensure that the sample

was diverse in terms of SES. Accordingly, the resulting sample offered a diverse mix of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 27

Erez, Ibarra, Bales, & Gur

participants in terms of backgrounds representing the spectrum of employment (employed,

unemployed, students).

Contact persons at each agency were also responsible for setting up the initial roster of

justice system personnel and social service providers with whom interviews were conducted,

although the researchers did not confine themselves to this roster of participants (see below). The

contact person forwarded an IRB-approved recruitment letter explaining the purpose and nature

of their participation to potential interviewees. Upon their arrival for the interview (in either the

office of an interviewee or in an agency conference room), justice personnel and social service

providers were given informed consent agreements reiterating that the interviews were

confidential and that that they were under no obligation to participate, could withdraw at any

time, and would suffer no harm should they decline to participate. Permission to digitally record

the interviews was also sought during the informed consent process. The “snowballing”

technique of pursuing interviews with personnel who are mentioned by interviewees as having

expertise in a particular area or on a specific issue provided a means of locating and recruiting

additional justice personnel and social service providers to participate in the study.

2. Topics of in-depth interviews

The topics broached during the interviews were designed to garner the most pertinent

descriptions and evaluations that representatives of each group could offer based on their

expertise and/or experience. Although the topics of the interviews were wide-ranging, the central

focus of the interviews was on a) establishing the legal, organizational, and technological

distinctions and practices that define and structure the program’s working environment, b) the

emotional, psychological, and practical impact that GPS program participation has on defendants

and victims (including the latter’s safety), and c) the quality and intensity of victim support and

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 28

Erez, Ibarra, Bales, & Gur

offender supervision built into the program’s design.

Thus, questions probed the history and

evolution of the interviewees’ encounters with the technology, the technology’s effects on those

who use it, the auspices for social interactions between clients/victims and staff that are

structured into the program, and the interactions between various people working within the local

justice system and/or social service agencies.

3. Qualitative data analysis

In-depth interviews were transcribed into electronic documents and coded manually. The

coded data were the basis for memos on the themes that emerged during interviews for each

category of participant (i.e., memos were developed identifying themes in the interviews with

defendants, victims, judges, etc.). Note was taken of the site at which the interviews were

conducted so as to discern variations in themes by site, resulting in an analysis that was sensitive

to both common themes for each category of participant as well as distinctive themes that are

related to site differences (e.g., variations in program rules, levels of victim-support, intensity

and intrusiveness of supervision, cost basis for participation, implementation of the technology’s

capacities, and compensation for its limitations). The memos were revised and developed into

the various thematic sections of the qualitative analysis component in accord with principles of

constructivist grounded theory (Charmaz, 2003). The latter is concerned with elucidating core

processes of action and interpretation that are evident in the practices, concepts, and distinctions

that are expressed during interview-based accounts or in the context of fieldwork more generally.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 29

Erez, Ibarra, Bales, & Gur

Chapter 3: Findings

This chapter presents findings from all components of the study. First, it reports results

from the national survey of pretrial agencies that use EM technologies in their operations,

focusing on a subset of agencies that employ GPS in the context of DV during the pretrial period.

Second, drawing on interviews with relevant professionals, programs at the six sites that

participated in the qualitative study (three of which also participated in the quantitative study) are

described, highlighting central themes in the programs’ orientations and practices. Third, results

from the quantitative study of the short- and long-term impact of GPS enrollment on participants

at the three sites are reported. Finally, main themes and key issues uncovered during in-depth

interviews with the defendants and victims are described.

A. Web-based survey

The adoption of GPS technology in the U.S. by pretrial agencies, and the place of GPS

technology in their response to DV cases, is documented in the web-based survey. Practitioners

(N = 616) representing 43 states, the District of Columbia, and Puerto Rico, responded to a web-

based survey covering a range of topics pertaining to EM and decisions about whether and how

to adopt and implement GPS.

Almost two-thirds of the respondents reported using EM

technologies to supervise pretrial defendants, and about a quarter use GPS for DV. The majority

of the report focuses on respondents representing programs utilizing GPS for DV during pretrial

(N = 149), hereafter the DV subsample. The survey is helpful in showing that agencies and

programs participating in the qualitative and quantitative components of the study represent

central trends in current usage at the national level.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 30

Erez, Ibarra, Bales, & Gur

1. Types of technologies used and defendants monitored

Respondents in programs using any EM technology (N = 393) indicated the types of

defendants they monitored (see Table 3A-1). Nearly half report using GPS for sexual offenses,

while two in five use GPS for DV or other violent offenses. RF is used most frequently – by one

in three programs – for a violent offense other than DV, while bilateral RF, where both the

defendant and victim receive RF devices (see Erez, Ibarra, & Lurie, 2004), was used by fewer

than one in twenty programs. Nearly one in three programs use Secure Continuous Remote

Alcohol Monitoring (SCRAM) for DWI/DUI defendants during pretrial, over one in five for

other alcohol or drug offenses, and one in six programs for DV defendants. Other forms of

monitoring technologies were used less frequently.

Table 3A-1. Percentage of respondents (N = 393) indicating their program uses each technology, by type of

defendant

GPS

Radio

Frequency

(RF)

Bi-

lateral

SCRAM

Remote

Alcohol

Monitoring

(not SCRAM)

Equipment-

Free

Monitoring

Other

DV or IPV

41.2%

29.3%

4.3%

16.5%

7.6%

3.8%

Other violent offense

40.2%

32.8%

4.3%

11.5%

8.4%

7.6%

3.6%

Sexual offense

48.3%

26.7%

2.8%

6.4%

6.6%

6.4%

3.3%

DWI/DUI

19.3%

28.8%

3.6%

31.6%

17.3%

6.6%

3.1%

Other alcohol/drug offense

23.7%

28.2%

3.3%

20.9%

12.0%

7.1%

4.8%

Juveniles

21.6%

20.6%

2.8%

6.4%

4.1%

5.9%

3.6%

Gang Members

22.4%

22.1%

2.5%

4.6%

4.1%

5.3%

3.6%

Within the DV subsample (N = 149), 87.9% report using GPS for DV (see Table 3A-2).

Programs in the DV subsample also use every other technology more often than programs in the

larger sample (N = 393). Their greater use of technology may indicate that GPS for DV

programs also use GPS and other technologies for other offense types, or that the DV defendant

population includes persons who are facing multiple charges that call for the application of other

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 31

Erez, Ibarra, Bales, & Gur

technologies. As in any self-report survey, it is possible some respondents did not indicate all

applications.

Table 3A-2. Percentage of respondents in the DV subsample (N = 149) utilizing each technology, by type of

defendant

Table 3A-3 reports percentages of programs in the DV subsample using each number of

technologies, by offense type. For example, 40% of programs use one technology to monitor DV

and other violent offense defendants, while more than half of programs use more than one

technology for DV, the highest such percentage for all offense categories. Taken together, Tables

3A-1, 3A-2, and 3A-3 demonstrate the survey reached a large population of agencies utilizing

EM technologies, and that programs represented in the DV subsample specialize in the use of

GPS for DV during pretrial, while employing other EM technologies for other offense types.

Table 3A-3. Percentage of respondents in the DV subsample (N = 149) utilizing each number of technologies,

by type of defendant

Number of technologies used

DV or IPV

5.4%

39.6%

29.5%

14.1%

7.4%

4.0%

0.0%

Other violent offense

14.8%

38.3%

23.5%

12.8%

6.7%

4.0%

0.0%

Sexual offense

11.4%

49.7%

21.5%

8.1%

6.7%

2.7%

0.0%

DWI/DUI

17.4%

36.9%

24.2%

14.1%

5.4%

1.3%

0.7%

Other alcohol/drug offense

22.1%

38.3%

20.8%

10.7%

4.7%

3.4%

0.0%

Juveniles

45.6%

30.2%

14.8%

4.7%

3.4%

0.7%

Gang Members

45.0%

29.5%

12.1%

6.7%

4.0%

2.7%

0.0%

GPS

Radio

Frequency

(RF)

Bi-

lateral

SCRAM

Remote

Alcohol

Monitoring

(not SCRAM)

Equipment-

Free

Monitoring

Other

DV or IPV

87.9%

41.6%

4.7%

22.8%

14.8%

13.4%

5.4%

Other violent offense

74.5%

44.3%

4.7%

16.1%

13.4%

14.1%

3.4%

Sexual offense

81.9%

36.9%

3.4%

6.7%

12.8%

12.1%

3.4%

DWI/DUI

36.2%

38.3%

4.0%

37.6%

26.9%

12.1%

4.7%

Other alcohol/drug offense

40.3%

36.9%

3.4%

27.5%

20.1%

12.8%

6.7%

Juveniles

36.9%

29.5%

3.4%

5.4%

7.4%

8.7%

3.4%

Gang Members

41.6%

30.9%

3.4%

7.4%

8.1%

10.1%

2.0%

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 32

Erez, Ibarra, Bales, & Gur

2. Program start date and location

The DV subsample represents programs operating for varying lengths of time (see Figure

3A-1) located in 43 states, the District of Columbia, and Puerto Rico (see Figure 3A-2).

Figure

3A-1 displays the number of programs beginning to use GPS technology for DV each year, and

the climb in total number of programs using GPS for these purposes to ninety (90) – the total

number of respondents that provided a year. While seven programs had been launched before the

year 2000, 51 started between 2006 and November 2009, with a high of 21 in 2007. The spike in

2007 may be explained by passage of legislation mandating the use of these technologies in

several states and related new funding sources.

Figure 3A-1. Year in which organization began using GPS technology for tracking domestic violence (DV)

defendants (N = 90)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 33

Erez, Ibarra, Bales, & Gur

Figure 3A-2 shows states with programs utilizing any type of EM during pretrial (N =

167), and states with programs utilizing GPS for DV during pretrial (N = 74). Ohio had the

greatest number of responding programs in both categories, followed by Florida, and though

Minnesota had thirteen programs utilizing EM during pretrial, only two used GPS for DV.

Jurisdictionally, the majority of programs (80%, N = 100) serve a county or city, 15% more than

one county, and 5% an entire state. Administratively, 40% of programs (N = 100) are located in a

county probation department, 13% in a sheriff’s department or jail, 18% in a court setting, and

the rest in state probation departments, private or nonprofit organizations, or some other

administrative position.

Figure 3A-2. Slopegraph of states with programs using EM (left, N = 167) and GPS (right, N = 74)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 34

Erez, Ibarra, Bales, & Gur

3. Rationale and aims of GPS programs

The programs designed to supervise and monitor DV pretrial defendants via GPS

emerged due to a convergence of several factors. Figure 3A-3 indicates that 90% of respondents

reported the desire to keep victims safer and provide enhanced supervision to DV defendants as

the most influential reasons for establishing such programs. High profile cases or incidents and

jail overcrowding were also important (over 70% of respondents endorsed these reasons).

Legislation requiring GPS in pretrial of certain DV defendants was also cited as important by

about half of the respondents. However, over a third (40%) of the practitioners (N = 101)

reported there was no legislation in their jurisdiction mandating DV defendants be placed on

GPS, and that there was no policy stipulating that DV defendants not monitored with GPS should

be placed in jail. Only a minority (5%) reported having both legislation mandating placement on

GPS for pretrial DV defendants, and placement in jail if not being monitored with GPS.

Figure 3A-3. Percentage of respondents in the DV subsample indicating how important each reason is to the

agency when deciding to start to use GPS for supervising DV pretrial defendants (N = 116–126)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 35

Erez, Ibarra, Bales, & Gur

Selection criteria developed for choosing who is placed on GPS may also inform the aims

of the program. In the DV subsample, seriousness of offense was most important among seven

criteria for selecting defendants for supervision (see Figure 3A-4). An order of protection was

deemed important by over 90% of respondents, as was the defendant’s prior criminal history. A

victim request that a defendant be placed on GPS was endorsed by about 60% of the

respondents. The least important criteria were the defendant’s income, and the defendant’s

employment status.

Figure 3A-4. Percentage of respondents in the DV subsample indicating how important each factor is for

choosing who is placed on GPS when used for supervising DV pretrial defendants (N = 101–105)

A defendant who is referred to a GPS program for placement may not be enrolled for

various reasons. Figure 3A-5 describes the considerations practitioners report as possibly

disqualifying or preventing defendants from enrollment. A major reason reported is that

defendants do not meet program requirements (e.g., cannot find a separate residence or have a

land line telephone). Inability to pay program fees was another consideration that occasionally

prevented defendants from being enrolled. On the other hand, a lack of GPS equipment or victim

cooperation was rarely reported to be a problem.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 36

Erez, Ibarra, Bales, & Gur

Figure 3A-5. Percentage of respondents in the DV subsample indicating how often defendants are not

enrolled in GPS for DV pretrial programs for each reason (N = 112–116)n

4. Referrals and enrollment

Prevailing practices in pretrial programs using GPS for DV (per data from the DV

subsample, N = 149) are presented below. Over two-thirds of defendants (71%) are referred to

the GPS programs by judicial orders, another portion by the supervising agency (14%) or

prosecutorial requests (10%), with very few referred due to legislation or general policy.

Table

3A-4 indicates that the majority of programs hook up nearly all cases referred for GPS

enrollment. Of programs providing both referral and enrollment rates (N = 98), more than two-

thirds were able to hook up all referrals. Excluding one outlier, the successful enrollment rate

among the remaining programs is over half (53%) of all referrals.

Table 3A-4. Descriptive statistics for number of monthly referrals and enrollments/hook-ups for programs in

the DV subsample (N = 149)

Minimum

Maximum

Mean

Std. Deviation

In an average month, how many defendants

are referred to the GPS program?

101

0.5

150

16.980

25.288

In an average month, of those referred to be

placed on GPS how many are enrolled or

hooked up?

100

0.5

225

15.959

27.802

Note: Outlier excluded.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 37

Erez, Ibarra, Bales, & Gur

The average caseload for officers in the DV subsample monitoring GPS defendants is

depicted in Table 3A-5. On average, non-GPS officers have more than twice as many cases as

GPS officers. GPS officers have smaller caseloads than non-GPS officers in two-thirds of

programs with GPS and non-GPS officers, with non-GPS officers managing three times as many

defendants as GPS officers. While some GPS officers (22.2%) have as many cases as non-GPS

officers, very rarely do GPS officers have larger caseloads than non-GPS officers, and officers in

the few programs with only GPS caseloads average 22.25 defendants. These statistics suggest

that most programs recognize that GPS for DV cases requires careful and labor-intensive

supervision from officers, possibly because of the risk involved to victims or the larger stream of

information produced by GPS tracking.

Table 3A-5. Average monthly caseloads for officers using GPS and non-GPS officers in the DV subsample

5. Technological attributes and costs

Patterns in vendor adoption are depicted in Figure 3A-6, showing that six vendors

dominate the market, with one vendor having almost a quarter of all programs among their

clientele. Nearly 90% of the programs utilizing the two most frequently used vendors were

satisfied with their service provider.

Costs associated with monitoring technology were calculated by considering how much

each unit costs the program per day, per defendant. To calculate mean cost, ranges (e.g., “$5-

$15”) were recoded into average cost (e.g., $10).

The minimum cost per defendant per day was

$0, while the maximum was $18.50, with a mean of $9.80 (N = 85, SD = 3.8). Entries of no costs

likely include pilot programs subsidized by grants, whereas higher costs may reflect a program

Minimum

Maximum

Mean

Std. Deviation

What is the average caseload for officers

who monitor defendants on GPS?

1.5

275

43.833

40.004

What is the average caseload for officers

who do not monitor defendants on GPS?

300

91.805

65.114

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 38

Erez, Ibarra, Bales, & Gur

lacking the volume of units for bulk discounts. Programs indicating a range of costs associated

with different levels of monitoring (i.e., passive, hybrid, or active) are included in Table 3A-6,

which also includes information on how much defendants are charged per day. Acknowledging

sample size limitations, calculating cost to defendants as a percentage of cost to the agency for

each level of technology shows defendants pay over three-quarters of the cost for active

monitoring, over half the cost for hybrid, and nearly the entire cost for passive electronic

monitoring. What seems like a sliding scale may reflect the cost of each technology to the

agency.

Figure 3A-6. Percentage of programs (N = 113) utilizing each vendor (N = 122) in the DV subsample

Advanced technologies are costly, have potential for failure, and inappropriate handling

may damage the equipment. Some programs report zero technology failures or cases of damage

to the equipment, whereas for others the rate of failure and damage is very high (up to 100%).

Vendor A

23%

Vendor B

13%

Vendor C

13%

Vendor D

11%

Vendor E

Vendor F

Vendor G

Vendor H

Vendor I

Other

Multiple Vendors

Unknown

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 39

Erez, Ibarra, Bales, & Gur

Fewer than 20% of GPS units were returned due to failure of the equipment (N = 52), with a

high standard deviation (SD = 27.1). Reported figures likely relate to how long programs have

operated. For example, older programs should have higher rates of failure and damage as devices

age and are used more. Programs operating at full capacity might not send devices for regular

maintenance or repairs, and perhaps contracts with vendors impact these considerations.

Table 3A-6. Cost of active, hybrid, and passive GPS for programs (N = 18) reporting multiple levels of

monitoring and costs in the DV subsample

ACTIVE

HYBRID (On-Demand,

Intermediate)

PASSIVE

Program

Cost to

agency/day

Cost to

defendant/day

Cost to

agency/day

Cost to

defendant/day

Cost to

agency/day

Cost to

defendant/day

$6.50

$5.25

$10.00

$8.00

$10.00

$9.00

$6.50

$8.95

varies

$6.95

varies

$4.45

varies

$8.29

$10.00

$4.41

$10.00

$13.30

$13.50

$12.00

$9.00

$8.95

varies

$4.94

varies

$9.96

$0.00

$5.75

$0.00

$8.95

varies

$6.95

varies

$4.45

varies

$8.29

$10.00

$4.41

$10.00

$15.00

$11.00

$6.50

$5.25

$9.50

$4.50

$12.00

$9.00

$10.00

$6.00

$25.00

$5.00

$16.00

$10.00

$15.00

$12.00

$15.00

AVG/

DAY

$11.18

$8.68

$7.28

$4.00

$6.84

$6.79

a: federally funded

b: varies depending on financial status

c: $50 installation fee; $10/day or 20% of gross salary

d: $10/day or 20% of gross salary

GPS technology has many capabilities for agencies to consider when implementing a

GPS for DV program. The adoption of these capabilities is presented in Figure 3A-7. The

overwhelming majority of respondents report using exclusion and inclusion zones, “real time

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 40

Erez, Ibarra, Bales, & Gur

mapping of defendant’s current location,” and “mapping of defendant’s movement in the

community over time.” Importantly, less than half of respondents report that GPS functionality is

utilized to dispatch law enforcement for any kind of defendant violations, and fewer than one in

three programs send victims text message notification when defendants violate. While three in

four programs share GPS data with law enforcement (e.g., to check an alibi), fewer than half

indicate law enforcement personnel respond to alerts.

Practitioners might become more aware of individual victim and defendant characteristics

and tendencies if programs utilize numerous features available with GPS monitoring. Of the

capabilities described in Figure 3A-7, one in four programs in the DV subsample did not use any

capabilities, yet three in five use five or more.

Figure 3A-7. Capabilities utilized by programs (N = 107-111) in the DV subsample

Core to GPS programs is how they respond to alerts of possible defendant violations.

Programs can opt to have law enforcement respond to alerts and violations. The survey indicates

that in over a third (35.6%) of the programs (N = 101), law enforcement responds to alerts and

violations, in 40% the police responds to neither, in 8.9% there is a response only for violations

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 41

Erez, Ibarra, Bales, & Gur

but not alerts, and in 7.9% of programs law enforcement respond to alerts but not violations.

Lacking a coordinated criminal justice response to violations and alerts makes it less likely that

GPS technology will serve as a preventative tool, rendering data collected by the device useful

only after adverse events have occurred (i.e., confirm suspect was at scene). Programs not reliant

on a law enforcement response may have in-house practitioners investigating violations and

alerts, or do so in cooperation with a vendor or third party call center.

Survey responses about GPS training indicate that equal time is allocated for training on

equipment and the program’s policies – 11.5 hours each. The training period for a minority

(6.4%) of programs consisted of at least 40 hours of training on equipment, and 10.9% of

programs had at least 40 hours on programs and policies. Both programs and vendors provided

training. Practitioners reported satisfaction with both sources of training, but internal training

was associated with greater level of satisfaction, perhaps due to familiarity with local program

issues. The functional components of GPS programs provide practitioners with tools, and

training can help facilitate effective utilization of the technological resources.

6. Current uses and changing program attributes

The ways programs apply and use GPS may change over time, such that the reasons

programs start using GPS may evolve as they continue to rely on the technology. Shifting

expectations and the wisdom an organization acquires in the course of implementing the

technology also influence this process: limitations become apparent, innovations are

incorporated, and a more comprehensive understanding of the implications of using GPS for DV

during pretrial develops. Figure 3A-8 presents how various attributes of the programs are

perceived by the practitioners. It indicates that the greatest degree of agreement was around the

importance of ensuring client accountability, ensuring victim safety, more effectively protecting

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 42

Erez, Ibarra, Bales, & Gur

the public, and deterring additional crimes. The less important factors in continuing to use GPS

were its perceived potential to mitigate pretrial absconding, its utility as a sanctioning tool, and

its ability to relieve jail overcrowding.

Figure 3A-8. Percentage of respondents in the DV subsample indicating how important each reason is to the

agency when using GPS for supervising pretrial defendants (N = 105–126)

The overwhelming agreement among practitioners as to the importance of GPS in

achieving disparate criminal justice goals (e.g., protecting the public and relieving jail

overcrowding) can be contrasted with differences among the DV subsample concerning various

attributes of GPS supervision of DV defendants (see Figure 3A-9). The most agreed upon

statement was that “GPS cannot prevent a client from committing a crime.” Yet, over 90% also

agreed that “exclusion zones deter associations with victims,” and that “GPS data provide

agencies with the opportunity to better supervise their clients, not just ‘monitor’ them.” Further

agreement was reported with statements such as “GPS can eliminate ‘he said/she said’ situations

between client and victim,” and there was consensus about GPS supervision allowing

practitioners to “ask ‘hard questions of their clients” and “proactively help guide clients to make

better decisions.”

There was disagreement with the statements that “GPS is lenient to the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 43

Erez, Ibarra, Bales, & Gur

defendant and they should be in jail” and “GPS tracking increases the number of defendants

being recommitted to jail or prison due to violating the conditions of their release.”

These series of responses imply that practitioners do not feel that GPS is too lenient as a

pretrial option, or that monitoring increases the number of defendants who are sent to jail or

prison for violations. However, there was less agreement among practitioners with the statement

that “GPS is punitive to the defendant.”

Figure 3A-9. Percentage of respondents in the DV subsample indicating to what extent they agree or disagree

with statements regarding using GPS for supervising DV pretrial defendants (N = 95-101)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 44

Erez, Ibarra, Bales, & Gur

7. Victim-relevant factors

Programs may have various degrees of victim-centricity. Figure 3A-10 shows roughly

two thirds of programs contact victims regarding defendant participation in the program and only

10-15% use criteria to qualify victims, report that victims receive training, require victims to sign

documentation acknowledging the capabilities and limitations of the GPS program, and require

victims to participate in order to place defendants in a GPS program. In less than a third of

programs (30%) victims receive text message notification when the defendant violates, meaning

it is the least utilized function.

Figure 3A-10. Victim-centric features of programs in the DV subsample (N = 105-115)

8. Impact of GPS on defendant behavior and other defendant-relevant factors

Figure 3A-9 underscores the idea that GPS monitoring, and its associated restrictions,

enable more extensive supervision of defendants, with nearly 90% of respondents agreeing that

GPS had a positive impact on defendant supervision. Although respondents agree that GPS

monitoring deters defendants from “going to hazardous locations,” (i.e., “bars,” “known drug

areas”), a plurality disagree with the statement that GPS discourages defendants’ “former

associates” from having contact with them.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 45

Erez, Ibarra, Bales, & Gur

One indicator of how defendants adjust their behavior when placed on GPS may be the

degree to which they violate the conditions of their program. The majority (about 60%) indicated

that GPS clients violate conditions of their program less than defendants in non-GPS programs,

and nearly 25% thought they violated about the same. The majority of practitioners reported that,

in an average month, less than 7% of defendants violate the program by entering exclusionary

zones (intentionally or unintentionally).

Using violation rates as measures of programmatic

success can be problematic, however, as low rates may reflect a lack of awareness on the part of

the program about actual violations. Considering victim- or defendant-relevant factors that can

be targeted for improvement may generate alternative measures of success.

According to survey results, defendants who participate in GPS programs are enrolled an

average of 99.5 days (N = 84; SD = 65.6 days), with a minimum of one day and maximum of

365 days, and with clusters at 30 (8.3%), 60 (11.9%), 90 (26.2%), and 180 days (11.9%). The

range in days reflects varied usage of the GPS to monitor defendants. Participating defendants

are charged an average of $8.80 per day (N = 86; SD = 5.1), though 16.3% of defendants cannot

make payments and are exempted from having to do so, while another 16.3% pay $10.00. A few

programs require defendants pay up to $18.50 per day or percentages of gross salary.

Respondents were asked about the responsibilities of defendants and the types of

programming in which they participate (see Figure 3A-11). Regular office visits were the most

frequent expectation for defendants, followed by field visits and alcohol or drug testing. In terms

of treatment programs, substance abuse treatment was more common than anger management

treatment, and mental health counseling, employment assistance, and cognitive interventions

were also reported as being utilized. About one sixth of the respondents did not know whether

specific modalities were available in their jurisdiction.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 46

Erez, Ibarra, Bales, & Gur

Lastly, only a minority (13%) of programs had ever formally evaluated the effectiveness

of GPS tracking for defendants, and only a third employ special procedures to assess the risk of

pretrial violent behavior.

Figure 3A-11. Percentage of respondents in the DV subsample indicating the percentage of defendants

participating in each program while being monitored (N = 96-99)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 47

Erez, Ibarra, Bales, & Gur

B. Varieties of GPS for DV programs: Perspectives of justice personnel in six sites

This section introduces the agencies that participated in the quantitative (West, South,

Midwest) and qualitative (West, South, Midwest, East, Southeast, Southwest) components of the

study, drawing upon in-depth interviews with criminal justice personnel (N = 83) and NGO-

based social service providers (e.g., victim advocates associated with NGO’s; N = 19) to

characterize their distinctive approaches to organizing GPS for DV programs.

Two sets of value dimensions help identify each agency’s approach. First, GPS programs

for DV are distinguishable along the crime control versus due process continuum associated with

criminal justice (Packer, 1968). An emphasis on crime control posits “the repression of criminal

conduct” (Packer, 1968: 48) as the most important function of law enforcement,” and day-to-day

operations are organized on the presumption that offenders within the justice system’s purview

are “probably guilty.” By contrast, an emphasis on due process favors the “doctrine of legal

guilt” over actual guilt (Packer, 1968: 53), recognizing the “possibility of error” entering into

criminal justice procedures and the importance of being mindful about the civil liberties of

offenders (especially non-convicted persons).

Second, the participating GPS programs for DV are distinguishable along the treatment

versus punishment continuum associated with the ends of criminal justice: to rehabilitate or to

sanction. This dimension concerns the extent to which justice responses are oriented toward the

“ideal” of rehabilitation (Garland, 2001) versus administering sanctions. A treatment orientation

results in efforts to discern the offender’s needs and deficiencies – enabling the pursuit of some

kind of intervention or remedial regimen. A punishment orientation entails embedding

restrictions and deprivations in the offender’s everyday life that provide a “taste of punishment,”

impressing upon him

the grievous nature of his past conduct.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 48

Erez, Ibarra, Bales, & Gur

In regard to the crime control versus due process dimension: EM programs have

previously been observed to entail “pains of imprisonment” (Payne and Gainey, 1998) on par

with those that were once described with respect to the hardships of prison life (Sykes, 1958).

Based on the present study’s analysis, however, there is wide variation in the onerousness of

GPS for DV programs. All such programs logically entail one rule: that the defendant stays away

from a named party, i.e., the alleged victim or prosecuting witness. The defendant is also likely

to be required to stay away from the victim’s residence, from any children the two parties have in

common, and any areas that either the victim or her children routinely visit (e.g., the victim’s

workplace and her family’s residence; the children’s school). Such “exclusion zones” can be

defined more or less expansively: More restrictive programs are likelier to broadly define such

zones, possibly resulting in inconveniences to the defendant when the zones interfere with his

paths of travel (e.g., bus routes taken to work); but because the defendant is presumed to be

guilty, he is viewed as capable of causing the victim renewed harm, entailing that the zone be

more broadly defined so that program personnel have more time to respond to any possible

incursions.

Programs with an underlying “crime control” approach are likely to be reactive as well as

proactive. As a result, they erect restrictions that are focused on clients’ lifestyles, routine

activities, and social environments as much as their physical location and patterns of movement.

Restrictions cover a wide range of areas pertinent to defendants’ liberties and sense of well-

being. They establish severe curfew hours and require submission to practices that increase the

transparency of defendant’s lives. These restrictions are coupled with a generalized suspicion

and unaccommodating attitude toward client requests for last minute changes in set routines,

since such requests introduce unknown risks and uncertainty.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 49

Erez, Ibarra, Bales, & Gur

The due process orientation emphasizes the idea that restrictions should veer toward what

is minimally required to accomplish client supervision, allowing the defendant to otherwise lead

life as he would were he under general release on bond: since he is presumed innocent, his

freedoms should be respected. Such an approach is apt to value flexibility in handling exigencies

emerging from clients’ situations – such as promptly accommodating changes in work schedules

or permitting the defendant to call rather than visit a supervising officer for his weekly check-in.

Indeed, flexibility and a willingness to work with the particularities of each defendant's situation,

and not treat them all the same, undermines the uniform standardized goal that would otherwise

be considered the program's core mission: to reinforce a no-contact order. That is, programs with

a more flexible orientation may on occasion allow the defendant to have “non-hostile contact”

with the victim in the case. Programs adopting this approach may be concerned to protect the

defendant's rights by exercising caution about imposing rules that the client will be prone to

violate, especially when those rules are viewed as unreasonable (e.g., tight curfew hours for

defendants whose work entails long-distance trucking) and creating needless legal jeopardy for

him.

The treatment versus penality continuum embodies values expressed in efforts to

organize the offender-supervisor relationship in either a therapeutic or punitive direction. The

agency and the training of its staff are especially determinative of the program’s orientation.

There is no consensus, either nationally or within the sample of six sites that is the focus of the

present study, about what kind of agency is best positioned to run a GPS for DV program. Thus,

programs that are staffed by personnel who have been certified as social workers are likely to

develop a different way of interacting with clients than are programs staffed by personnel who

have law enforcement backgrounds.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 50

Erez, Ibarra, Bales, & Gur

GPS for DV programs focused on rehabilitation are more concerned to establish and

address the offender’s needs, while programs with a penality emphasis are more concerned to

determine the offender’s compliance with court-ordered rules and to administer corresponding

controls and punishments. Thus, a case manager with a penality orientation will listen to a

client’s statements during routine face-to-face meetings with an ear attentive to signs that the

client is involved in activities that run afoul of the court’s rules, or pose a danger to the victim.

Ensuing warnings, reproaches, and threats may lead the client to view the encounter as

investigative (or “Gotcha!” oriented), rather than merely inquisitive – resulting in his becoming

increasingly guarded in future encounters. By contrast, a supervising officer with a treatment or

rehabilitation orientation will likely listen to the client’s statements with a different ear, and

probe for information with a different purpose. The officer with a caseworker perspective may

listen for issues that pertain to the client’s living situation, including whether it addresses his

immediate and long-term needs or its potential effects on his state of mind and behavior. Given

this focus, the officer will seek to encourage openness and trust in the relationship, and in the

spirit of rehabilitation, not penalize clients for “minor infractions,” but rather adopt an attitude of

making the client’s situation “work” so that he is less likely to violate the court’s rules, resulting

in the client’s belief that his caseworker is “trying to help” him.

Crisscrossing the various value elements, and helping to shape and inform the approach

that is likely to make the most sense to each agency, are the characteristics of the population

commonly or likely to be referred to its GPS program. Programs that routinely receive high-risk

clients, or clients with long and varied criminal records, are likelier to develop approaches that

are mindful of contingencies associated with those kinds of clients, as opposed to programs that

have clients with comparatively milder records. The three (quantitative) impact sites (in

Midwest, South, and West) handle different “types” of offender populations, as measured by

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 51

Erez, Ibarra, Bales, & Gur

offense history, and this difference is important in appreciating the approach to supervision that

each agency has adopted.

The remainder of this section presents descriptions of the six agencies to highlight how

the values discussed above are manifested in their GPS for DV programs.

1. Midwest: Crime control and penality

This agency’s program is housed within the Sheriff Department’s Electronic Monitoring

Unit and is led by sworn personnel previously employed in the county’s Probation Department as

“enforcement officers.” Recent arrivals to the unit have backgrounds as patrol or jail officers,

and the emphasis on crime control and penality is a logical extension of their training. Although

officers will say that they do not know whether the defendant

is guilty or innocent of his

charges, they do have access to his case file and a listing of his prior arrests and convictions,

which are likely to be extensive. Midwest GPS enrollees tend to have a relatively high number of

prior arrests (mean = 13.3), as shown in Table 3C-2 (see p. 64). The program’s predicate is that

the client represents a serious threat to the victim and hence rules and restrictions are a vehicle

for doing risk management. As an example, clients’ “free hours” are tightly controlled: When not

at work or at home, the client has five hours per week to conduct “personal business,” thereby

limiting the client’s “opportunity time” for endangering the victim. Clients are also subject to

unannounced surprise home visits, during which armed officers wearing bullet proof vests will

conduct an “in plain view” search of common living areas and the defendant’s sleeping quarters,

a procedure that may at times turn into a full-fledged search. The purpose of such searches

ordinarily is to detect the presence of controlled substances (e.g., drugs or alcohol) or weapons

(e.g., firearms or swords) that might pose heightened risk for the victim.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 52

Erez, Ibarra, Bales, & Gur

Personnel are strict in their enforcement of program rules and requirements, to the point

of appearing unreasonable and indifferent to the clients. Weekly office visits are required of

clients, during which the information exchanged is fairly minuscule (though urine screens will at

times be collected during these visits, and the encounters will be more substantial with clients

who are suspected of engaging in problematic activities). Nevertheless, all clients are required to

appear, often traveling long distances and waiting hours before their visit with their supervisor –

at the expense of their weekly allocated “out hours” and the patience of affected employers

whose workers run late. Inflexibility is also evident in the drawing up of exclusion zones, or the

areas into which the abuser cannot venture: programmed as radii of two miles in circumference,

the officer is unwilling to create zones of irregular (non-circled) shape that can accommodate

travel routes (interstate highways, bus lines, etc.) that may “clip” the zone, resulting in increased

burdens on clients’ commuting plans, especially for those reliant on public transportation.

The rules and restrictions have the merit of highlighting for the officer those clients who

have problems in being compliant, and allow the officer to manage the information stream

provided by GPS tracking logs. Those who perform poorly with respect to rules and restrictions

are likelier to have their patterns of movement studied for indicators of risk to the victim (e.g., is

the client having secret rendezvous with the victim? Is he following her when she is outside of

the exclusion zone?). The officer may also seize upon minor infractions of the rules to warn the

client that whatever limited liberties the client has left may be withdrawn. The client may be

threatened with being put “on lockdown” (or total house arrest), or being sent back to jail, should

he not “get with the program.” Supervisors are sworn law enforcement personnel, thus such

warnings are likely to be viewed by clients as more than empty threats.

The Midwest agency also employs specialized program officers who maintain contact

with the victim in the case (the two women are employed as “victim specialists”). These officers

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 53

Erez, Ibarra, Bales, & Gur

contact victims about the defendants’ proposed assignment to GPS, request and document

victims’ consent to participate,

and announce their availability through a direct telephone

number on an as needed basis, twenty-four hours a day. Victims are also encouraged to furnish

staff with updates about changes in their residential address, the defendant’s circumstances or

other pertinent information, and they are offered escort to court appearances in an effort to

bolster their resolve to stand by the case and not be intimidated by the defendant or his family.

The victim specialists play an important role in sustaining two-way communications between the

victim and the agency, and also serve as a conduit to the broader court system.

2. South: Due process and treatment

South’s GPS for DV program is housed in the pretrial supervision unit of a community

corrections department staffed by a sizable number of probation officers with a social work

orientation. The South approach to supervision is premised in the idea that the clients under

supervision are flawed rather than (probably) guilty. The presumed flaw becomes the basis for

handling the client: as a person with patterns of conduct, feeling, or thinking that result in his

getting into trouble. Clients

enrolled into the GPS program are seen as often being socially

marginalized, lacking social support or social capital, and their offense histories are

comparatively minor (mean prior arrests = 3.8). The supervisory role entails, in part, identifying,

and having the client recognize, the sources of their troubles, so that the client will be “motivated

to change” and be better able to “successfully re-enter” society, as the agency director put it,

adding “GPS is intervention, not punishment.”

Rule violations are not taken at face value. They may have an underlying meaning, one

not associated so much with risk management issues, but with problems in living that the client

is handling unsuccessfully. For example, a client’s persistent problems returning home by curfew

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 54

Erez, Ibarra, Bales, & Gur

may stem from his inability to assert himself to an employer who wants him to work overtime.

Or the client’s repeated attempts to telephone his previous address may be associated with the

client’s difficulty in managing being separated from his children. In short, violating behavior

requires a diagnosis, not an investigation, and the remedy is apt to be a referral to services or an

impromptu counseling session rather than merely violating the client or seeking a bond

revocation. The caseworker’s task is to help the client determine what is keeping him from being

successful; “casework” involves managing the client’s path to reentry as much as it entails

enforcing court rules. From a casework perspective, violations are an opportunity to address

clients’ needs and encourage self-insight. Open and honest dialogue is essential to the process;

clients should feel that they have nothing to hide (including violating conduct). Evidence that

rapport exists between clients and staff is shown in the familiarity and intimacy that they

sometimes display (hugging on arrival or departure). Clients’ comments about how the

caseworker “helps” them deal with various issues, or how the caseworker has become a parental

figure to them, offers more evidence.

Wide variation exists in how restrictive or demanding “life on the box” is for clients,

illustrating the importance accorded to flexibility and the practice of customizing program

participation. Exclusion zones are irregularly shaped and can be of diverse ranges, from as few

as 500 feet to as much as a few miles, but can also be drawn up to encompass entire cities or

other jurisdictions (i.e., when the victim lives in another city, county or state). Some clients have

early curfews, others have very late curfews and, in a handful of cases, no curfew at all. Some

clients may be directed to attend mental health related treatment, others to drug counseling or

anger management and batterer intervention programs. The program does not create a universal

template of rules for clients to follow but rather takes guidance from the presiding judge or

magistrate who referred the client to the monitoring program and formulated the terms of pretrial

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 55

Erez, Ibarra, Bales, & Gur

release. Caseworkers may add new foci to the case management as needs are discovered, such as

helping a client who develops an interest in obtaining a GED. Some judges allow clients to have

limited, “non-hostile” interaction with victims (e.g., to manage shared business assets or discuss

parental issues), and even permit a smaller number of clients to live with their victims

(essentially undermining the idea that GPS tracking should be used to shield the victim from the

defendant).

Much supervision occurs by telephone. Clients call in advance to announce that they will

return home late or leave home early. Supervisors call clients when the transmitter’s signal is

weak, lost, or out of range. Defendants are encouraged to call if they have equipment problems,

and to inform their supervising officer if they have difficulties conforming to rules or if

hardships arise. Some clients call without prompting, just to check-in informally. The telephone-

based interaction becomes a basis for reinforcing themes that emerge during office visits. Home

visits are rare, conducted by an unarmed field officer, who rarely enters the client’s home

premises unless he has to inspect equipment for technical problems. Transmission issues

typically prompt home visits, whether related to signal strength or battery or power supply

problems. Home visits do not involve searches or efforts to detect risk factors in the residence

(officers in South do not have arrest powers). In short, supervision is not invasive, but instead

has therapeutic and welfare aims. Intrusive approaches are believed to undermine the rapport and

open dialogue essential to a productive client-supervisor relationship.

3. West: Due process and penality

The West’s due process approach positions the monitoring agency as a neutral collector

of information that may be used at a later date in judicial proceedings. GPS “brings data to the

table,” as the agency director stated. It is no longer a matter of “he said, she said” where claims

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 56

Erez, Ibarra, Bales, & Gur

of abuse and unlawful contact are concerned. Given the tracking information that is compiled by

GPS technology and put at the DA’s, defense attorney’s, and judge’s disposal, the defendant’s

case can be helped or hurt (and the victim’s credibility can be enhanced or undermined) by the

GPS points logged by the technology. Despite the fact that defendants entered into the GPS

program are viewed by staff as “high risk” – more than one in four face charges pertaining to

violation of a DV-related restraining order, and the mean number of prior arrests for GPS

supervisees is 4.6 – caseworkers are loath to pry into the everyday doings or mental states of

their clients. Inadvertent admissions by supervisees make caseworkers legally responsible for

relaying any divulged information to the DA, thus transforming themselves into witnesses for the

prosecution, which is not how they define their role: as a neutral collector of information.

The supervising officer’s role entails neither proactively catching defendants in the act,

nor rehabilitating defendants. Although personnel will notify the police if the victim zone is

breached and the offender refrains from exiting the area, West agency supervisors claim to

“never” mention the word “protection” when the GPS program is described to victims. Victims

are reminded that they are responsible for their own safety and should remain watchful of their

environment; victims should have “peace of mind,” but not a “false sense of security.” The GPS

program is described to victims as an information tool that will play a role in the legal outcome if

required, and not as an entity mobilized to act proactively or reactively with regard to the

victim’s safety. Indeed, the program has over the years reduced the routine contact it has with

victims, who at the time of the study were being serviced by victim advocates at a different

location – in the prosecutor’s office.

The rehabilitation emphasis of South and the proactive enforcement approach of Midwest

presume intensive, face-to-face casework, which is not practiced here. Instead West favors a

“teamwork” approach in which caseworkers “share” supervisees, resulting in defendants seeing

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 57

Erez, Ibarra, Bales, & Gur

different caseworkers from visit to visit. Although the information stream documented by GPS

tracking is important to supervision, defendants are not under curfew restrictions and need not

return home daily to the same residence; many of the supervisees are understood to be self-

employed and thus without regularly set schedules. Exclusion zones are typically set for

comparatively narrow radii, on average about 500 feet (though they can be smaller still if the

defendant lives closer to the victim) because it is believed that defendants should not be

burdened by exclusion zones that cause hardships for the non-convicted. This is especially an

issue because this agency sets multiple exclusion zones for each victim (e.g., victim’s home,

work place, victim’s parents’ home, places of worship, etc.): If the zones are sufficiently large,

entire swaths of the metropolitan area’s transportation grid could become inaccessible.

The flexibility that is provided by the absence of curfew restrictions and residential

inclusion zones, and the accommodating approach that is evinced by relatively small victim

zones, is consistent with a due process approach in that the program’s strictures are minimally

intrusive and require few behavioral modifications on the defendant’s part, but victims are rattled

by the sense that their estranged partner moves about unencumbered despite their GPS

enrollment. Although defendants are often subject to urine testing – as per a judge’s order – there

are no routine home visits, and in-person office visit requirements are relaxed over time, so that

weekly visits can become once or twice a month events, and supervisees can check in via

telephone if needed. Any violations discovered – whether in the form of positive urine screens,

victim zone incursions, or equipment tampers – are forwarded to the DA. Staff in the district

attorney’s office note that they rarely act on violation reports before a case is disposed, however,

except when they might seek leverage during plea bargaining.

The most onerous aspect of the program is cost of participation: on average ten dollars a

day, though as high as sixteen dollars a day (fees are on a sliding scale that can go down to zero

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 58

Erez, Ibarra, Bales, & Gur

if the supervisee has a documented inability to pay). The EM program is financially self-

sufficient, relying on defendants paying fees on a weekly basis, which may also affect the

program’s reach and intrusiveness: because the program’s survival depends on defendants

submitting fees, their continued ability to earn a living is important. A significant portion of the

population has an immigrant background (almost half of the GPS defendants have Hispanic or

Latino ancestry), and administrators speculated that the experience of being on GPS and paying

sizable per diem fees can have the effect of teaching men from cultures with traditions of

machismo about the seriousness with which violence against women is addressed in the U.S.

4. East: Due process and penality

East, a suburban agency within a major metropolitan area of the country, takes an

approach similar to that found in West, emphasizing collecting information over controlling

crime, and enforcing program rules instead of developing rehabilitative courses of action.

East’s approach is embedded within a court culture that practices an evidentiary-based model of

prosecution. Justice personnel at this site emphasize that the state should not elevate risk for the

victim of DV by having her testify against the defendant. Instead, it is important for the state to

proceed with the case independent of whether the victim participates as a witness for the

prosecution, which entails collecting information that can be leveraged into the state’s case.

There is no expectation that defendants ought to pay the costs associated with GPS.

A hybrid of passive and active GPS is used by the monitoring agency: active during

business hours, passive during non-business hours. Defendants’ GPS points logged during non-

business hours are downloaded the following morning by a dedicated monitoring technology

specialist, who reviews logs for any indications of technical or moving violations, which are then

forwarded to caseworkers who decide what fixes are needed (e.g., a new battery in the case of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 59

Erez, Ibarra, Bales, & Gur

power supply issues), or whether to issue warnings to clients or report violations to the

prosecutor’s office. East’s approach is clearly not focused on either proactive or reactive

responses to victim endangerment, but rather is organized to document any possible program

violations (including contact attempts) and to treat them as matters for the courts to adjudicate,

especially if they should occur beyond business hours. Consistent with an evidence-based

approach to prosecution, East’s personnel are more oriented toward preventing victim tampering

than toward victim safety per se. In addition, supervising officers believe that GPS provides a

useful pretext for having face-to-face encounters with defendants, during which information

pertinent to the case might be gleaned, and which can be more important and accurate than

anything that might be learned directly through GPS tracking.

5. Southeast: Treatment and crime control

Staffed by sheriff’s deputies, and hence inclined toward a crime control orientation, this

agency also adopts a treatment approach, owing to a) the small caseload that the program has to

manage, b) a specialization in domestic violence cases among the deputies who do the casework,

and c) an outlook typically associated with small towns – one marked by police officers who are

familiar with the families who reside in the area.

Accordingly, officers can be both rigorous

about investigating the movements of defendants

who are enrolled in the GPS for DV program,

and also very focused in trying to “help” the supervisees.

This program’s motto is, “social work and law enforcement go together.” Deputies might

spend an hour studying each defendant’s GPS points from the previous day – noting how much

time he stayed in one location, the patterns in his driving speeds – to determine whether anything

suspicious is happening (e.g., are there signs of stalking behavior?). A “founded” suspicion

might provide the basis for a sting operation: a deputy’s vehicle might show up at those locations

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 60

Erez, Ibarra, Bales, & Gur

where a defendant reliably goes during his non-curfew hours, catching him by surprise (officers

have full powers of arrest, including in relation to program violations). This kind of intensive

“detective” work is supplemented by a touch of the therapeutic approach seen in South:

encouraging defendants to rethink ingrained patterns of thought and action, directing them to

pursue new opportunities, or make a fresh start. Deputies see clients as risks to be managed, and

yet they also believe clients are fundamentally malleable and can be guided in a more non-

violent direction. In the short term, however, personnel are focused on preventing contact that

might result in injury, or in the defendant’s persuading the victim to withdraw or amend her

complaint. The simultaneous focus on treatment and crime control is also seen in the approach

taken by the district attorney in this jurisdiction. On the one hand, the DA sees GPS as a

“homicide prevention tool.” On the other hand, GPS provides the DA with an opportunity to

interact with reluctant victim-participants, who are required to meet with her when they wish to

request removal of the GPS requirement. On such occasions, the DA will explain the

ramifications of this request to the victim, correcting “purposefully inaccurate information” she

receives from her alleged abuser, with the intent of empowering her to follow up with her

complaint.

6. Southwest: An ad hoc, make-shift approach

This agency developed its GPS for DV program as a result of jail overcrowding that

occurred in the wake of a massive swelling of the county’s population over the past twenty years.

Housed within the jail, the monitoring unit was established as a way of allowing non-violent

inmates to be released so as to ease pressures that stem from overcrowding. Eventually, judges

came to the view that some defendants facing DV charges could be allowed into the program,

enabling them to maintain their employment and uphold their familial responsibilities, while

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 61

Erez, Ibarra, Bales, & Gur

keeping them under some degree of supervision. Supervisory personnel would rather not have

defendants with any violent charges on their case load, and, indeed, violent offenses were

broadly considered a disqualifying condition, but these clients were referred to them by judges

and so had to be accepted for supervision.

The Southwest program is non-programmatic: it has an ad hoc quality, lacking an explicit

philosophy. Treatment and rehabilitation elements mix with penality emphases; defendants

are

as likely to be met with a due process sensibility as they are with one focused on crime control.

The caseworkers who staff the program and the administrators who oversee it have a background

as guards in the county jail. They serve in the GPS program for a two-year stint, after which their

assignment rotates them back to a position among the jailers. This kind of perennial rotation does

not permit the formation of the kind of institutional memory on which systematic and integrative

policies and practices can be established, or clearly understood and conveyed across successive

cohorts of staff: as personnel shift, approaches shift. Judges describe the current program as a

make-do, but note that it has the potential to become a more systematic and integrative response

to the problem of domestic violence.

7. Conclusion

This review of the varieties of GPS programs and their divergent approaches to

monitoring alleged abusers is important to keep in mind in contextualizing the findings from the

quantitative component of the present study. The Midwest, West, and South agencies discussed

above participated in this part of the research project, to which the report now turns.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 62

Erez, Ibarra, Bales, & Gur

C. Quantitative impact study (Midwest, West, South)

1. Midwest

a. Description of the domestic violence cases

This section presents findings relating to the impact of GPS or RF

monitoring on short-

term outcomes of violations and re-arrest and the long-term outcome of re-arrest, as well as

whether the duration on GPS has an effect on these outcomes. The examination of these issues is

also conducted separately for those cases that resulted in a dismissal or acquittal by the court

versus those in which the individuals arrested for DV were convicted by the court. Table 3C-1

provides descriptive statistics of the variables used in this analysis based on the entire sample. Of

the total sample of 2,052 DV cases, 25.9% were tethered to a GPS device, 22.6% were placed on

RF, and 51.5% were not placed on either form of electronic surveillance. The defendants placed

on GPS for whom data were available (n = 416) wore the device for an average of 58.82 days,

with a median of 36 days, and a range of one to 517 days. Of the 1,813 cases in which the

judicial disposition information was available, 40.7% were convicted and 59.3% were dismissed

(only six defendants were acquitted). The three outcomes are distributed in the following manner

among the cases for which these data were available: 13.4% of the cases violated the conditions

of their supervision and 3.3% were re-arrested in the short-term (i.e., during their supervision

period). Second, nearly three-fifths (59.3%) of the DV defendants were re-arrested in the long-

term period (i.e., the one-year follow-up period).

Table 3C-1 shows that males comprise 92.2% of the cases and the racial/ethnicity groups

are as follows; Whites = 37.5%, Hispanics = 2.1%, other races = 0.2%, and Blacks comprise the

largest proportion of the total population at 60.0%. The average age of the DV defendants was

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 63

Erez, Ibarra, Bales, & Gur

32.6 years. Just over one-fifth of defendants (22.4%) were arrested for a felony crime; the

average number of prior arrests was 15.4, and the average number of prior convictions was 12.9.

Table 3C-1. Descriptive statistics of variables with all cases: Midwest site

Mean

Number of Days on GPS

416

58.820

67.673

Convicted Cases

1,813

.407

.491

Short-Term Outcome – Violation

2,052

.134

.341

Short-Term Outcome – Arrest

2,052

.033

.178

Long-Term Outcome – Arrest

2,052

.593

.491

Male

1,915

.922

.269

White

1,879

.375

.484

Black

1,879

.600

.490

Hispanic

1,879

.021

.143

Other Races

1,879

.002

.052

Age at DV Incident

1,905

32.550

10.387

Number Prior Arrests

2,051

15.372

15.064

Number Prior Convictions

2,030

12.957

12.716

Current Crime – Felony

1,979

.224

.417

Global Positioning System (GPS)

2,052

.259

.438

Radio Frequency (RF)

2,052

.226

.419

Non-Electronic Monitoring

2,052

.515

.500

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models

presented in subsequent tables include only cases that had no missing data on all of the variables in each model.

Table 3C-2 provides mean comparisons across four groups of interest in the analysis:

those placed on GPS, RF, jailed, and bonded (i.e., without EM supervision). Some salient

differences across these four groups include the following: GPS cases are the most likely to

result in a conviction (55.6%), while defendants bonded out of jail are least likely to be convicted

(22.5%). Short-term violations are much more likely to occur among RF cases (41.1%)

compared to GPS cases (15.8%) or bond cases (0.0%).

GPS cases are also the most likely to be

arrested in the short term (8.9%).

In contrast, the likelihood of cases being arrested in the long

term is very similar for GPS cases (58.6%), RF (57.0%), and those released on bond (57.6%) and

somewhat higher among jail cases (63.6%).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 64

Erez, Ibarra, Bales, & Gur

Females are more likely to be in the GPS sample (10.3%) than in any of the other three

groups, and Blacks are likelier to be in the Jail sample (69.3%) than in any other group. The

average age is essentially the same across the four groups (32.4 to 32.8 years). The group most

likely to have been arrested for a felony crime is the Jail group (30.5%), while the Bond group is

comprised of only 14.6% felons. The Jail group also has more serious criminal records as

measured by prior arrests (mean = 18.7) and convictions (mean = 15.4). Comparisons in these

prior record measures across the other three groups indicate relative consistency in the average

number of arrests and convictions.

Table 3C-2. Descriptive statistics (mean values) of variables across four groups: Midwest site

GPS

(N=531)

(N=465)

Jail

(N=549)

Bond

(N=507)

Convicted Cases***

.556

.298

.433

.225

Short-Term Outcome –Violation***

.158

.411

N/A

Short-Term Outcome – Arrest***

.089

.002

.007

.030

Long-Term Outcome – Arrest

.586

.570

.636

.576

Male

.897

.921

.944

.923

White***

.430

.410

.285

.385

Black***

.532

.564

.693

.597

Hispanic

.028

.018

.022

.016

Other Races***

.000

.009

.000

.002

Age at DV Incident

32.362

32.400

32.812

32.547

Number Prior Arrests

13.275

14.426

18.749

14.773

Number Prior Convictions

11.281

12.457

15.418

12.484

Current Crime – Felony**

.217

.222

.305

.146

*** Indicates significant Chi-Square statistics in crosstabular analysis, p<.001.

** Indicates significant Chi-Square statistics in crosstabular analysis, p<.01.

- T-tests of mean differences across all of the permutations of the four groups found statistically significant differences (p<.001)

in the number of prior arrests and convictions for GPS vs. Jail, RF vs. Jail, and Jail vs. Bond.

N/A indicates not applicable.

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models presented

in subsequent tables include only cases that had no missing data on all of the variables in each model. The number of

cases reflects the maximum number of cases that were used in one or more of the individual variables.

Tables 3C-3, 3C-4, and 3C-5 present Logistic Regression models to address the question

of whether the duration on GPS is associated with the short-term outcomes of a violation or re-

arrest, and the long-term outcome of re-arrest among all DV defendants in the Midwest site that

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 65

Erez, Ibarra, Bales, & Gur

were electronically monitored using GPS, as well as among two subgroups – those that were

dismissed or acquitted and those in which the DV arrest resulted in a conviction. Because DV

cases tend to have a relatively high rate of dismissals (victims are reluctant to testify against their

abuser due to emotional or/and economic ties, or are intimidated from following through with the

complaint), an examination of the impact of GPS is conducted separately for those cases which

resulted in a dismissal or acquittal by the court versus those in which the individuals arrested for

DV were convicted by the court. The findings presented in Table 3C-3 demonstrate that the

length of time on GPS has no empirical link to whether domestic violence cases were more or

less likely to violate the conditions of supervision. This finding is consistent for all DV cases and

for the subgroups dismissed/acquitted and convicted when modeled separately. Identical

conclusions are derived from the logistic models predicting re-arrest in both the short and long

term presented in Tables 3C-4 and 3C-5. The conclusion drawn from this analysis is that, in the

Midwest site, increasing the dosage of GPS surveillance has no impact on the likelihood of DV

cases failing on supervision through a violation or a new arrest.

Table 3C-3. Logistic regression models of effect of length of time on GPS and short-term outcome of whether

violated among all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

.000

1.000

-.003

.997

-.001

.999

Male

.252

1.164

.061

1.063

1.016

2.761

Black

.328

1.388

-.748

.473

.957*

2.604

Age at DV Incident

-.022

.978

-.148*

.863

-.010

.990

Number of Prior Arrests

.024

1.024

.094**

1.098

.002

1.002

Current Crime – Felony

-.058

.944

.338

1.403

-.468

.626

Intercept

-1.806*

.735

-2.536*

Model Chi-square

5.829

17.989**

7.394

R-Square

.030

.245

.067

355

149

176

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

The variables Other Races and Hispanic were excluded from the analysis because the model could not properly converge with

their inclusions.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 66

Erez, Ibarra, Bales, & Gur

Table 3C-4. Logistic regression models of effect of length of time on GPS and short-term outcome of whether

re-arrested among all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

.003

1.003

.003

1.003

.000

1.000

Male

-.141

.868

-.977

.377

-.386

.680

Black

.046

1.047

-1.025

.359

.513

1.670

Age at DV Incident

-.003

.997

-.002

.998

.007

1.007

Number of Prior Arrests

.030*

1.031

.033

1.033

.041*

1.042

Current Crime – Felony

1.540***

4.666

.534

1.706

2.327***

10.25

Intercept

-3.074***

-1.892

-3.265**

Model Chi-square

28.620

6.454

33.394

R-Square

.155

.099

.302

316

149

176

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

The variables Other Races and Hispanic were excluded from the analysis because the model could not properly converge with

their inclusions.

Table 3C-5. Logistic regression models of effect of length of time on GPS and long-term outcome of whether

re-arrested among all cases, dismissed/acquitted cases, and convicted cases: Midwest site

* p<.05, **p<.01, ***p<.001.

Reference categories: White, Crime.

The variables Hispanic, Other Races, and Number of prior convictions were excluded from the analysis because the models could

not properly converge with their inclusion.

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

-0.001

0.999

0.003

1.003

-0.002

0.998

Male

-0.136

0.873

0.75

2.117

-0.97

0.379

Black

0.354

1.424

0.581

1.787

0.197

1.218

Age at DV Incident

-.042***

0.959

-0.02

0.98

-.042*

0.959

Number of Prior Arrests

.037**

1.038

0.028

1.028

.044*

1.045

Current Crime – Felony

-.663*

0.515

-0.759

0.468

-0.271

0.762

Intercept

1.503**

-0.107

2.223**

Model Chi-square

27.317***

11.894

17.022**

R-Square

0.1

0.105

0.124

355

149

176

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 67

Erez, Ibarra, Bales, & Gur

b. Effect of GPS vs. RF on outcomes

Tables 3C-6, 3C-7 and 3C-8 present Logistic Regression models to address the relative

effectiveness of GPS technology (i.e., being tracked with house arrest during curfew hours) to

the RF method (i.e., being only on house arrest during curfew hours) in producing positive short-

and long-term outcomes among all cases, and then separately for cases that were dismissed or

acquitted by the courts versus those resulting in a criminal conviction. In terms of the effect of

GPS in reducing the likelihood of violations in the short term relative to RF, the model shows a

significant reduction in violations among GPS cases (B = -1.455, p<.001) when controlling for

offender demographic characteristics, current crime type, and prior record. The odds ratio of .233

indicates that DV defendants placed on GPS exhibit a reduced odds of a violation in the short

term that is 76.7% less than comparable RF cases. Comparable findings in terms of GPS being

more effective than RF are found when the cases are distinguished by the judicial outcome of

dismissal/acquittal and conviction; however, the GPS effect is stronger among

dismissed/acquitted cases (B = -2.077, p<.001) than the convicted cases (B = -1.077, p<.001).

The effect of GPS surveillance relative to the less restrictive form of electronic

monitoring (RF) on short-term arrest outcomes is quite different. Table 3C-7 shows that GPS

cases are significantly more likely to be arrested in the short term than RF cases (B = 4.107,

p<.001) for all cases and among convicted cases only (B = 3.486, p<.001). A different result

occurs when the outcome measure is the likelihood of long-term arrest. Table 3C-8 shows there

is no statistically significant effect of being placed on GPS surveillance relative to RF for all

cases (B = .268, n.s.), dismissed/acquitted cases (B = .148, n.s.), and convicted cases (B = .435,

n.s.).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 68

Erez, Ibarra, Bales, & Gur

Table 3C-6. Logistic regression models of the effect of GPS versus RF on short-term outcome of violation for

all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

-1.455***

0.233

-2.077***

0.125

-1.077***

0.341

Male

-0.094

0.91

-0.456

0.634

0.903

2.466

Black

.374*

1.453

-0.188

0.829

1.133***

3.104

Hispanic

-0.379

1.46

-0.378

0.72

0.734

2.083

Age at DV Incident

-.043***

0.958

-.072***

0.93

-0.021

0.979

Number Prior Arrests

.019**

1.019

.037***

1.038

0.003

1.003

Current Crime – Felony

-0.001

0.999

-0.132

0.876

-0.049

0.952

Intercept

0.556

1.917

-1.257

Model Chi-square

105.833***

95.015***

34.453

R-Square

0.175

0.262

0.155

814

463

306

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

Table 3C-7. Logistic regression models of the effect of GPS versus RF on short-term outcome of arrest for

all cases and convicted cases: Midwest site

All Cases

Convicted

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

4.107***

60.762

3.486***

32.657

Male

-.165

.848

-.429

.651

Black

-.065

.938

.288

1.333

Age at DV Incident

.005

1.005

.011

1.011

Number Prior Arrests

.028*

1.029

.044*

1.006

Current Crime – Felony

1.484***

4.409

2.164***

8.710

Intercept

-7.148***

-6.693***

Model Chi-square

83.077***

55.605***

R-Square

.299

.362

816

308

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the outcome variable with values of 1 (arrested) to calculate a model for the dismissed or acquitted

cases. Also, too few cases were in the Other Race variable to include in the models. Including the number of prior arrests and

prior convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the

model. However, the substantive findings were the same regardless of which prior record measure was included in the model.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 69

Erez, Ibarra, Bales, & Gur

Table 3C-8. Logistic regression models of the effect of GPS versus RF on long-term outcome of arrest for all

cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

.269

1.309

.148

1.360

.435

1.544

Male

-.154

.857

.502

1.652

-1.014*

.363

Black

.227

1.255

.275

1.316

.119

1.127

Hispanic

-.485

.616

-.346

.708

.041

1.042

Age at DV Incident

-.054***

.948

-.050***

.952

-.047***

.954

Number Prior Arrests

.049***

1.050

.050***

1.051

.056***

1.057

Current Crime – Felony

-.592**

.553

-.675**

.509

-.345

.708

Intercept

1.482***

.873

1.734

Model Chi-square

87.638***

56.904***

31.468

R-Square

.138

.157

.131

814

463

306

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

c. Effect of GPS vs. non-GPS on outcomes

This section compares short- and long-term outcomes of DV defendants placed on GPS

compared to Non-GPS (RF or bonded out). Table 3C-9 provides results on the effect of GPS on

whether DV defendants are arrested in the short term among all cases, and those who were

dismissed or acquitted versus those who were convicted. Consistently the models demonstrate

that DV cases placed on this stringent form of electronic surveillance are significantly more

likely to be arrested in the short term (all cases - B = 2.003, p<.001; dismissed/acquitted - B =

1.735, p<.001; acquitted cases - B = 1.939, p<.001). By contrast, Table 3C-10 shows that when

the outcome measure is long-term arrest, the findings indicate that DV defendants placed on GPS

are no more likely to be re-arrested than Non-GPS cases (all cases - B = .212, n.s.;

dismissed/acquitted - B = .116, n.s.; acquitted cases - B = .261, n.s.).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 70

Erez, Ibarra, Bales, & Gur

Of particular interest in this study was whether GPS surveillance has an impact on the

likelihood of domestic violence defendants committing further acts of violence against domestic

partners in the long term relative to those DV defendants not placed on GPS. The models in Table

3C-11 address this question among all cases, those who had their cases dismissed or acquitted, and

those who were convicted by the court. For all cases, the first model demonstrates that DV

defendants placed on this stringent form of electronic surveillance are significantly less likely to be

arrested for a subsequent domestic violence crime in the long term (B = -.368, p<.05). The odds ratio

of .692 indicates that the odds of DV defendants placed on GPS to commit future domestic violence

in the long term are 30.8% less than comparable DV defendants not placed on GPS. While the effects

of GPS surveillance on the likelihood of future domestic violence are not statistically significant

when examining dismissed/acquitted and convicted cases separately, they remain in the same

direction of lower probabilities of future DV among GPS cases than non-GPS (dismissed/acquitted -

B = -.166, n.s.; acquitted cases - B = -.368, n.s).

Table 3C-9. Logistic regression models of the effect of GPS versus non-GPS on short-term outcome of arrest

for all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

2.003***

7.411

1.735***

5.667

1.939***

6.954

Male

-.052

.949

-.733

.480

-.024

.976

Black

-.177

.838

-.731

.481

.101

1.106

Age at DV Incident

-.008

.992

.001

1.001

-.016

.984

Number Prior Arrests

.024*

1.024

.022

1.022

.036

1.036

Current Crime – Felony

1.088***

2.967

.408

1.504

1.645***

5.180

Intercept

-4.419***

-3.732***

-4.173***

Model Chi-square

67.272***

21.378**

45.523***

R-Square

.172

.121

.245

1,310

823

412

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

Note: Models in which the Non-GPS control group did not include RF cases resulted in substantively the same findings in terms

of the effect of GPS.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 71

Erez, Ibarra, Bales, & Gur

Table 3C-10. Logistic regression models of the effect of GPS versus non-GPS on long-term outcome of arrest

for all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

.212

1.236

.116

1.123

.261

1.298

Male

.094

1.098

.416

1.516

-.509

.601

Black

.374***

1.454

.372**

1.551

.311

1.365

Hispanic

-.508

.601

-.727

.483

-.118

.889

Age at DV Incident

-.057***

.944

-.052***

.949

-.054***

.947

Number Prior Arrests

.041***

1.042

-.038***

1.039

.048***

1.049

Current Crime – Felony

-.330**

.719

-.250

.779

-.350

.704

Intercept

1.397***

1.049**

1.682***

Model Chi-square

201.376***

114.099***

67.745***

R-Square

.143

.135

.140

1,804

1,095

612

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

Note: Models in which the Non-GPS control group did not include RF cases resulted in substantively the same findings in terms

of the effect of GPS.

Table 3C-11. Logistic regression models of the effect of GPS versus non-GPS on long-term outcome of arrest

for domestic violence for all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Global Positioning System (GPS)

-.368*

.692

-.166

.847

-.368

.692

Male

.215

1.239

.246

1.278

.076

1.079

Black

.127

1.135

.052

1.053

.083

1.086

Hispanic

-.935

.393

-1.015

.362

-.557

.573

Age at DV Incident

-.023**

.977

-.031***

.970

-.002

.996

Number Prior Arrests

.013**

1.013

.010

1.010

.020**

1.020

Current Crime – Felony

-.047

.954

.072

1.075

-.248

.780

Intercept

1.343***

-.989*

-2.057**

Model Chi-square

27.450***

16.053*

12.419

R-Square

.026

.025

.037

1,804

1,095

612

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

Note: Models in which the Non-GPS control group did not include RF cases resulted in substantively the same findings in terms

of the effect of GPS. In fact, the effect of GPS on the likelihood of future domestic violence arrest was greater when RF cases

were excluded for the entire sample (B=-.525,p<.01,odds ratio = 0.592).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 72

Erez, Ibarra, Bales, & Gur

d. Effect of RF vs. non-EM on outcomes

This final section compares the effect of using RF for DV cases relative to those not

placed on any form of electronic surveillance (hereafter non-EM). Table 3C-12 indicates that RF

defendants are significantly less likely to be arrested in the short term relative to Non-EM

defendants (B = -2.130, p<.05). The odds ratio of .119 indicates that comparable RF defendants

have an 88.1% less odds of being arrested in the short term than comparable non-EM defendants.

Turning to the effect of RF on the long-term outcome of arrest, the models in Table 3C-

13 demonstrate that there is no significant influence of RF on the likelihood of arrest after

controlling for several relevant factors. This finding is consistent among all of the cases (B = -

.092, n.s.), those that were dismissed or acquitted (B = -.014, n.s.), and those that were convicted

by the court (B = -.246, n.s.).

Table 3C-12. Logistic regression models of the effect of RF versus non-EM on short-term outcome of arrest

for all cases: Midwest site

All Cases

Odds

Ratio

Radio Frequency (RF)

-2.130*

.119

Male

.007

1.007

Black

-.896

.408

Age at DV Incident

-.028

.972

Number Prior Arrests

.009

1.009

Current Crime – Felony

-.376

.687

Intercept

-2.738*

Model Chi-square

12.048

R-Square

.066

1,441

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model. Too few

cases were re-arrested when examining dismissed/acquitted cases and convicted cases separately to calculate these models.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 73

Erez, Ibarra, Bales, & Gur

Table 3C-13. Logistic regression models of the effect of RF versus non-EM on long-term outcome of arrest for

all cases, dismissed/acquitted cases, and convicted cases: Midwest site

All Cases

Dismissed/Acquitted

Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Radio Frequency (RF)

-.092

.912

-.014

.986

-.246

.782

Male

.159

1.172

.301

1.352

-.217

.805

Black

.376**

1.456

.344*

1.411

.298

1.348

Hispanic

-.457

.633

-.846

.429

-.035

.966

Age at DV Incident

-.060***

.941

-.057***

.945

-.057***

.944

Number Prior Arrests

.041***

1.042

.040***

1.041

.048***

1.049

Current Crime – Felony

-.234

.791

-.158

.854

-.332

.718

Intercept

1.439***

1.289***

1.588**

Model Chi-square

175.763***

107.287***

53.238***

R-Square

.155

.147

.154

1,441

940

434

* p<.05, **p<.01, ***p<.001.

Reference categories: White.

Too few cases existed in the Other Race variable to include in the model. Including the number of prior arrests and prior

convictions in the model resulted in high levels of multicollinearity, therefore, prior convictions were not included in the model.

However, the substantive findings were the same regardless of which prior record measure was included in the model.

2. West

a. Description of the domestic violence cases

This section presents findings relating to the impact of placing DV offenders on GPS

monitoring on short term (i.e., the pretrial/predisposition period) outcomes of violations and re-

arrest, and the long term (i.e., one-year follow-up period) outcome of re-arrest. We also

examined whether the duration of being monitored by the GPS (the “dosage” of GPS) has an

effect on re-arrest in the long term.

The examination of the duration effects of GPS includes 446 GPS cases in the West site.

Table 3C-14 provides descriptive statistics of the variables used in this analysis. The number of

days on GPS is measured as a continuous variable. There are 317 cases with complete data on

this measure and, on average, DV cases were tethered to a GPS tracking device for 132 days,

with a median of 79 days (the range was from 1 to 1,171 days). Of the 441 cases in which

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 74

Erez, Ibarra, Bales, & Gur

information about the judicial disposition was available, 57.1% were convicted and 42.9% were

dismissed or acquitted. The three outcomes are distributed in the following manner among the

cases in which these data were available: 15.2% of the GPS cases violated the conditions of their

supervision and 19.5% were re-arrested in the short-term. Second, just over one-fourth (26.1%)

of the DV cases on GPS were re-arrested in the long-term period.

Table 3C-14. Descriptive statistics of cases placed on GPS: West site

Mean

Number Days on GPS

317

132.429

174.744

Convicted Cases

441

.571

.495

Short-Term Outcome – Violation

448

.152

.359

Short-Term Outcome – Re-Arrest

447

.195

.396

Long-Term Outcome – Re-Arrest

379

.261

.440

Male

448

.955

.207

White

441

.272

.446

Black

441

.243

.429

Hispanic

441

.460

.499

Other Races

441

.025

.156

Age at DV Incident

446

32.204

9.496

Victim Younger Than Defendant

368

.595

.492

Number Prior Arrests

449

4.601

6.682

Number of Address Changes

450

.409

.909

GPS Due to Violation of Court Order

428

.283

.451

Crime – Simple Assault

446

.451

.498

Crime – Assault/Battery

446

.150

.358

Crime – Burglary/Trespass

446

.076

.266

Crime – Harass/Stalk/Intimidation

446

.054

.226

Crime – Murder/Kidnapping

446

.078

.269

Crime – Arson/Property Damage

446

.020

.141

Crime – Violate Restraining Order

446

.170

.376

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models presented in

Tables 2, 3, and 4 include only cases that had no missing data on all of the variables in each model.

Table 3C-15 provides mean comparisons across three groups of interest in the analysis:

those placed on GPS, remaining in jail, or bonded out of jail. Some salient differences across

these three groups include the following: Bond cases are the most likely type of case to result in

a conviction (71.2%), while those remaining in jail are the least likely to be convicted (56.0%).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 75

Erez, Ibarra, Bales, & Gur

Males are somewhat less likely to be in the bond sample (90.4%) than any of the other two

groups. However, none of these differences are statistically significant. Blacks make up a higher

percentage in the Jail group (39.7%) than any other group, and are the least likely to be part of

the GPS group (24.3%), and the differences are statistically significant (p<.001). The average

age differs significantly across the three samples with the jail sample as the oldest (34.0 years),

GPS cases 32.2 years, and Bond cases 30.8 years. No significant differences were found across

the three groups relating to age differences between victim and offender, or the number of

address changes. In terms of the measure of prior criminal record, the jail and bond groups were

quite similar (mean number of prior arrests: Jail = .340, Bond = .352). However, the GPS group

had significantly higher prior arrests on average (.451, p<.001). Significant differences were

found across the three groups in terms of the current offense types of simple assault,

assault/battery, burglary/trespassing, and violating a restraining order. These results indicate that

while the three samples are comparable on several of the covariates measured, controlling for

these measures to determine the unique effect of GPS on the outcomes to derive comparable

groups is essential.

To examine the unique effect of the duration on GPS, several control variables were

included in the multivariate analysis presented in Tables 3C-16, 3C-17, and 3C-18, which is

summarized in this section. Males comprise 95.5% of the cases and the racial/ethnicity groups

are as follows: Whites = 27.2%, Blacks = 24.3%, other races = 2.5%; Hispanics comprise the

largest proportion of the total population at 46.0%. The average age of the DV defendants was

32.2 years and in 59.5% of the cases, the victim was younger than the defendant. As a measure

of community stability, the number of times the defendant changed addresses was captured and

the average was .409 times. The defendants in this sample had an average of 4.6 arrests prior to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 76

Erez, Ibarra, Bales, & Gur

their instant DV arrest. More than one-quarter of defendants (28.3%) were placed on GPS as a

result of violating a court issued restraining order (in this site, any violation of restraining orders

result in placement on the GPS assignment). In terms of the most serious charge the DV

defendants were arrested for, simple assault had the highest prevalence at 45.1%, followed by

violating a restraining order (17.0%), and assault and battery (15.0%).

Table 3C-15. Descriptive statistics (mean values) of variables across three groups: West site

GPS

(N=639)

Jail

(N=434)

Bond

(N=56)

Convicted Cases

.571

.560

.712

Short-Term Outcome – Violation

.261

N/A

Male

.955

.960

.904

White

.272

.216

.185

Black***

.243

.397

.315

Hispanic*

.460

.376

.444

Other Races

.025

.012

.056

Age at DV Incident

32.204

34.028

30.811

Victim Younger Than Defendant

.595

.557

.612

Number Prior Arrests

4.601

6.672

5.444

Number of Address Changes

.451

.340

.352

GPS Due to Violation of Court Order

.150

N/A

Crime – Simple Assault***

.076

.340

.352

Crime – Assault/Battery***

.054

.306

.352

Crime – Burglary/Trespass*

.078

.037

.019

Crime – Harass/Stalk/Intimidation

.020

.044

.093

Crime – Murder/Kidnapping

.170

.090

.092

Crime – Arson/Property Damage

.571

.009

N/A

Crime – Violate Restraining Order*

.261

.141

.037

*** Indicates significant Chi-Square statistics in crosstabular analysis, p<.001.

* Indicates significant Chi-Square statistics in crosstabular analysis, p<.01.

- T-tests of mean differences across all of the permutations of the three groups found statistically significant differences for age

at DV incident GPS vs. Jail (p<.01) and Jail vs. Bond (p<.05) and number of prior arrests for GPS vs. Jail (p<.001).

N/A indicates not applicable.

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models presented in

subsequent tables include only cases that had no missing data on all of the variables in each model. The number of cases reflect

the maximum number of cases that were used in one or more of the individual variables.

Tables 3C-16, 3C-17, and 3C-18 present Logistic Regression models to address the

question of whether the duration on GPS is associated with the short-term outcomes of violations

or re-arrests, and the long-term outcome of re-arrests among GPS group defendants, both overall

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 77

Erez, Ibarra, Bales, & Gur

and by case outcome (i.e., convicted versus dismissed/acquitted). Binary logistic regression

analyses are used throughout this section given that the dependent variable (violation or re-

arrest) is a dichotomous variable (Allison, 1991; DeMaris, 1992; Menard, 1995). Checks for

multicollinearity were conducted on all of the logistic models using ordinary least squares

regression and assessing the tolerance statistics for each variable as recommended by Allison

(1991) to ensure the coefficients are not unstable. All of the tolerance levels were above .55 and

the vast majority exceeded .70, well above the .40 threshold suggestive of a multicollinearity

problem (Allison, 1991; Menard, 1995).

Table 3C-16. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether violated among all cases, dismissed/acquitted cases, and convicted cases: West site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

.000

1.000

.004

.996

-.000

1.000

Male

----

Black

.044

1.045

1.264

3.540

-2.142

.117

Hispanic

-.445

.641

.395

1.484

-1.511

.221

Other Races

-.231

.794

.973

2.645

----

Age at DV Incident

-.006

.995

-.020

.980

.022

1.023

Victim Younger Than Defendant

.269

1.309

.497

1.644

-.395

.673

Number Prior Arrests

.041

1.041

.030

1.031

.115

1.121

Number of Address Changes

.141

1.151

.116

1.123

.130

1.139

GPS Due to Violation of Court Order

.607

1.836

1.267

3.551

.560

1.751

Crime – Assault/Battery

-.413

.662

.591

1.805

----

Crime – Burglary/Trespass

.546

1.725

-.016

.985

1.392

4.021

Crime – Harass/Stalk/Intimidation

.129

1.137

----

2.540

12.677

Crime – Murder/Kidnapping

-.884

.413

.639

1.894

----

Crime – Violate Restraining Order

-.707

.493

-1.202

.301

-1.131

.877

Intercept

-1.861*

-2.504

-2.025

Model Chi-square

9.565

8.605

16.777

R-Square

.072

.121

.258

229

132

* p<.05, **p<.01, ***p<.001.

Reference categories: White, Crime – Simple Assault.

---- Indicates variables excluded from the analysis because the model could not properly converge with their inclusions.

Additionally, the variable Crime – Arson/Property Damage was excluded from the analysis in both models for the same reason.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 78

Erez, Ibarra, Bales, & Gur

The findings presented in Table 3C-16 demonstrate that the length of time on GPS has no

empirical link to whether DV defendants were more or less likely to violate the conditions of

supervision. This finding is consistent for all DV cases and for the subgroups dismissed/acquitted

and convicted when modeled separately. Identical conclusions are derived from the logistic

models predicting re-arrest in both the short and long term presented in Tables 3C-17 and 3C-18.

The conclusion drawn from this analysis is that, in the West site, increasing the dosage of GPS

surveillance has no impact on the likelihood of DV cases failing on supervision through a

violation or a new arrest.

Table 3C-17. Logistic regression models of effect of length of time on GPS and short-term outcome of

whether re-arrested among all cases, dismissed/acquitted cases, and convicted cases: West site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

.001

1.001

.001

1.001

.001

1.001

Male

-.034

.967

-1.351

.259

----

Black

-.284

.753

-.407

.666

-.108

.897

Hispanic

-.793

.453

-2.037**

.130

.054

1.055

Other Races

-.485

.616

-1.195

.303

----

Age at DV Incident

-.044

.957

-.084*

.919

-.011

.989

Victim Younger Than Defendant

.104

1.109

.500

1.649

-.069

.933

Number Prior Arrests

.022

1.022

.060

1.062

.001

1.001

Number of Address Changes

.420*

1.522

.938*

2.555

.194

1.213

GPS Due to Violation of Court Order

.843

2.323

1.183

3.265

.820

2.271

Crime – Assault/Battery

-.849

.428

-.715

.489

-.970

.379

Crime – Burglary/Trespass

-1.289

.275

----

-1.051

.350

Crime – Harass/Stalk/Intimidation

.370

1.447

.943

2.567

.105

1.111

Crime – Murder/Kidnapping

.578

1.782

.671

1.956

.456

1.578

Crime – Violate Restraining Order

-.466

.628

-.500

.607

-.482

.617

Intercept

-.065

1.897

-1.199

Model Chi-square

23.305

24.343

7.460

R-Square

.152

.276

.111

228

131

* p<.05, **p<.01, ***p<.001.

Reference categories: White, Crime – Simple Assault.

---- Indicates variables excluded from the analysis because the model could not properly converge with their inclusions.

Additionally, the variable Crime – Arson/Property Damage was excluded from the analysis in both models for the same reason.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 79

Erez, Ibarra, Bales, & Gur

Table 3C-18. Logistic regression models of effect of length of time on GPS and long-term outcome of whether

re-arrested among all cases, dismissed/acquitted cases, and convicted cases: West site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

Number Days on GPS

-.001

.999

-.002

.998

-.001

.999

Male

1.168

3.214

-2.048

.129

1.168

3.214

Black

1.667

5.295

-1.435*

.238

1.667

5.295

Hispanic

1.045

2.843

-.559

.572

1.045

2.843

Other Races

----

-1.581

.206

----

Age at DV Incident

-.049

.952

-.077**

.926

-.049

.952

Victim Younger Than Defendant

.574

1.775

.066

1.068

.574

1.775

Number Prior Arrests

.021

1.021

.102**

1.108

.021

1.021

Number of Address Changes

-1.011

.364

-.318

.728

-1.011

.364

GPS Due to Violation of Court Order

1.913

6.776

.613

1.846

1.913*

6.776

Crime – Assault/Battery

.686

1.985

.530

1.698

.686

1.985

Crime – Burglary/Trespass

.728

2.071

1.798

6.038

.728

2.071

Crime – Harass/Stalk/Intimidation

----

-2.210

.110

----

Crime – Murder/Kidnapping

1.766

5.845

1.796

6.026

1.766

5.845

Crime – Violate Restraining Order

-.973

.378

-.218

.804

-.973

.378

Intercept

-2.562

3.493*

-2.562

Model Chi-square

16.599

27.275*

16.599

R-Square

.288

.267

.288

208

131

* p<.05, **p<.01, ***p<.001.

Reference categories: White, Crime – Simple Assault.

---- Indicates variables excluded from the analysis because the model could not properly converge with their inclusions.

Additionally, the variable Crime – Arson/Property Damage was excluded from the analysis in both models for the same reason.

Tables 3C-19 and 3C-20 present the analysis addressing the question of whether those

arrested for DV offenses (DV defendants whose cases were dismissed/acquitted, or convicted)

who are placed on GPS are less likely to be re-arrested in the long-term (one-year follow-up

period) than comparable individuals who are not subjected to this form of intense electronic

surveillance. Table 3C-19 presents descriptive statistics within the GPS and non-GPS samples

relating to whether cases were convicted, their likelihood of being arrested in the long-term, and

a variety of control variables, including demographic characteristics of the DV defendants,

victim/defendant age differences, prior arrest history, and the type of current charge. The GPS

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 80

Erez, Ibarra, Bales, & Gur

and non-GPS cases have similar judicial outcomes in terms of the percent that are convicted

(GPS = 57.1%; non-GPS = 57.8%).

Table 3C-19. Descriptive statistics of GPS and non-GPS cases: West site

GPS Cases Non-GPS Cases

Mean

Difference:

GPS vs.

Non-GPS

Convicted Cases

441

.571

.495

498

.578

.494

-.007

Long-Term Outcome – Re-Arrest

379

.261

.440

437

.350

.478

-.089**

Male

448

.955

.207

493

.953

.211

.002

White

441

.272

.446

492

.213

.410

.059*

Black

441

.243

.429

492

.386

.487

.143***

Hispanic

441

.460

.499

492

.384

.487

.076*

Other Races

441

.025

.156

492

.016

.127

.009

Age at DV Incident

446

32.204

9.496

499

33.697

9.704

-1.493*

Victim Younger Than Defendant

368

.595

.492

421

.565

.496

.030

Number Prior Arrests

449

4.601

6.682

500

6.766

7.851

-2.165***

Crime – Simple Assault

446

.451

.498

498

.345

.476

.106***

Crime – Assault/Battery

446

.150

.358

498

.309

.462

-.159***

Crime – Burglary/Trespass

446

.076

.266

498

.034

.182

.042**

Crime – Harass/Stalk/Intimidation

446

.054

.226

498

.050

.219

.004

Crime – Murder/Kidnapping

446

.078

.269

498

.092

.290

-.014

Crime – Arson/Property Damage

446

.020

.141

498

.008

.089

.012

Crime – Violate Restraining Order

446

.170

.376

498

.127

.333

.043

t-test mean differences: *p<.05, **p<.01, ***p<.001.

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models presented later

include only cases that had no missing data on all of the variables.

On a bi-variate level, GPS cases are less likely to be re-arrested in the long term (26.1%)

compared with non-GPS cases (35.0%), and the mean difference of .089 is statistically

significant (p<.01). In regard to the demographic characteristics of the GPS and non-GPS

defendants, the two groups are essentially the same in terms of the percentage that are male

(95.5% vs. 95.3%), while the former group is significantly more likely to be White (diff = .059,

p<.05), Hispanic (diff = .076, p<.05), and younger (diff = -1.493, p<.05). In contrast, the non-

GPS defendant is significantly more likely to be Black than the GPS one (38.6% vs. 24.3%,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 81

Erez, Ibarra, Bales, & Gur

p<.001). No significant difference in the two groups was found in terms of whether the victim

was younger than the defendant. In contrast to non-GPS cases, individuals placed on the GPS

have significantly lower numbers of prior arrests (diff = -2.165, p<.001). Among the seven

different crime types examined, the data show that DV cases are more likely to involve simple

assault (diff = .106, p<.001) and burglary or trespass (diff = .042, p<.01), but are less likely to

involve an assault or battery (diff = -.159, p<.001). The comparisons across the GPS and non-

GPS cases on several covariates clearly indicate the need to control for these factors that may

influence long term re-arrest outcomes independent of GPS surveillance to quantify the unique

GPS effect.

Table 3C-20 presents logistic models of the predictors of long-term re-arrest

for all

cases (n = 642), those cases dismissed or acquitted (n = 311), and those resulting in a conviction

(n = 330). The first model presented, utilizing all of the available cases, indicates that males are

significantly less likely to be re-arrested (beta = -1.073, p<.01), as are younger offenders (beta =

-.045, p<.001), and those for which the crime involved harassment, stalking, or intimidation

(beta = -1.064, p<.05). In contrast, it was found that the individuals with a greater number of

prior arrests were significantly more likely to be re-arrested (beta = .074, p<.001). The primary

variable of interest, GPS surveillance, was statistically significant and indicates that, after

controlling for several relevant covariates, offenders who were placed on GPS were less likely to

be re-arrested in the long-term (beta = -.514, p<.001). The magnitude of the effect of GPS

monitoring among domestic violence cases is quantified by the odds ratio of .598, which

demonstrates that DV cases placed on GPS have a 40.2% lower odds of being re-arrested than

equivalent cases not electronically tethered. The models created among the two groups of

acquitted/dismissed cases and those convicted provide similar conclusions. Specifically, for

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 82

Erez, Ibarra, Bales, & Gur

those DV cases that were dismissed or acquitted by the courts, the effect of GPS on long-term re-

arrest probabilities is significant and negative (beta = -.572, p<.05) and the odds ratio of .564

indicates a 43.6% lower odds of re-arrest for those placed on GPS versus comparable individuals

without the GPS monitoring. Among cases in which a conviction resulted, similar GPS effects

emerge with the odds of re-arrest being 46.1% less than those who were not placed on this form

of electronic monitoring (beta = -.618, p<.05).

Table 3C-20. Logistic regression models of effect of GPS versus no GPS on long-term outcome of whether re-

arrested among all cases, dismissed/acquitted cases, and convicted cases: West site

All Cases Dismissed/Acquitted Convicted

Odds

Ratio

Odds

Ratio

Odds

Ratio

GPS

-.514**

.598

-.572*

.564

-.618*

.539

Male

-1.073**

.342

-.992

.371

-.826

.438

Black

.094

1.098

-.856*

.425

1.484**

4.412

Hispanic

.119

1.126

-.894

.538

1.277**

3.585

Other Races

.075

1.077

-.051

.409

1.063

2.896

Age at DV Incident

-.045***

.956

-.051**

.950

-.046**

.955

Victim Younger Than Defendant

.141

1.152

.405

1.499

-.144

.866

Number Prior Arrests

.074***

1.077

.087***

1.091

.059**

1.061

Crime – Assault/Battery

-.086

.917

.052

1.054

-.270

.975

Crime – Burglary/Trespass

.205

1.227

.415

1.515

-.025

.290

Crime – Harass/Stalk/Intimidation

-1.064*

.345

-1.159

.314

-1.239

.543

Crime – Murder/Kidnapping

-.219

.809

.283

1.327

-.610

1.327

Crime – Violate Restraining Order

.076

1.079

-.066

.936

.469

1.599

Intercept

1.327*

2.030*

-2.562

Model Chi-square

71.481**

43.454***

48.009*

R-Square

.148

.179

.197

642

311

330

* p<.05, **p<.01, ***p<.001.

Reference categories: White, Crime – Simple Assault.

---- Indicates variables excluded from the analysis because the model could not properly converge with their inclusions.

Additionally, the variable Crime – Arson/Property Damage was excluded from the analysis in both models for the same reason.

Taken as a whole, the results in Table 3C-19 indicate a compelling finding, i.e., that the

use of electronic monitoring through the application of GPS technology among individuals

charged with DV results in a significant reduction in the likelihood that re-arrest will occur in the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 83

Erez, Ibarra, Bales, & Gur

future. Additionally, this reduction in the likelihood of re-arrest among the GPS group exists for

those who were dismissed/acquitted and those convicted by the courts. In sum, the results in the

West site indicate that being on the GPS (rather than the length of time on GPS) is associated

with reduction of future violations or re-arrest.

The West site did not have an RF comparison group, and the Bond group was too small

to permit a reliable statistical analysis, so an examination of the short-term effect of GPS vs. non-

GPS was not performed; only the long-term effect comparing GPS to non-GPS is reported for

West.

3. South

a. Description of the domestic violence cases

This section presents findings relating to the impact of tethering DV defendants to GPS

or RF devices on short-term outcomes of violations and re-arrest, and the long-term outcome of

re-arrest. Additionally, the question of whether the duration on GPS has an effect on these

outcomes is addressed. An examination of these issues is conducted separately for those cases

that resulted in a dismissal or acquittal by the court versus those in which the individuals arrested

on DV charges who were ultimately convicted. Table 3C-21 provides descriptive statistics of the

variables used in this analysis based on the entire sample of cases in the South jurisdiction. Of

the total sample of domestic violence cases, 29.3% were tethered to a GPS device, 27.6% were

placed on RF, and 43% were not placed on either form of electronic surveillance. The defendants

placed on GPS wore the device for an average of 127 days, the median was 102 days, and the

range was from 7 to 508 days.

Of the 533 defendants for which the judicial disposition

information was available, 46.7% were convicted and 53.3% were dismissed or acquitted of the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 84

Erez, Ibarra, Bales, & Gur

charges filed against them. The three outcomes are distributed in the following manner among

the cases in which these data were available: 6.8% of the cases violated the conditions of their

supervision, 5.6% were re-arrested in the short-term (i.e., during their supervision period), and

7.8% of the DV defendants were re-arrested in the long-term period (i.e., the one-year follow-up

period after their acquittal/dismissal or completion of their sentence, if convicted).

Table 3C-21 shows that males comprise 83.6% of the cases and the racial/ethnicity group

distribution is as follows; Whites = 51.3%, Blacks = 48.7%, and Hispanics = 10.8%. The average

age of the DV defendants was 33.3 years. In terms of prior criminal involvement, this sample

averaged 3.28 prior arrest of any kind and .341 DV arrests. The type of DV crime charged in the

current incident most commonly involved an alleged battery (80.8%) followed by an assault

(11.9%) and some other form of DV (7.3%).

Table 3C-21. Descriptive statistics of variables with all cases: South site

Note: Cases that had missing data on each variable are not included in these statistics. The multivariate models

presented in subsequent tables include only cases that had no missing data on all of the variables in each model.

Mean

Number of Days on GPS

102

127.667

114.603

Convicted Cases

533

.467

.499

Short-Term Outcome – Violation

413

.068

.252

Short-Term Outcome – Arrest

413

.056

.230

Long-Term Outcome – Arrest

592

.078

.268

Male

604

.836

.370

White

604

.513

.500

Black

604

.487

.500

Hispanic

604

.108

.310

Age at DV Incident

604

33.321

10.526

Number Prior Arrests

604

3.281

4.564

Number Prior Arrests for DV

604

.341

.674

Current Crime – DV Assault

604

.119

.324

Current Crime – DV Battery

604

.808

.394

Current Crime – DV Other

604

.073

.260

Current Crime – Weapon Used

604

.157

.364

Global Positioning System (GPS)

604

.293

.456

Radio Frequency (RF)

604

.276

.448

Non-Electronic Monitoring

604

.430

.496

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

GPS and DV: Evaluation Study Page 85

Erez, Ibarra, Bales, & Gur

Table 3C-22 provides mean comparisons across four groups of interest in the analysis,

those placed on GPS or RF, remaining in jail, or bonded out of jail (without EM supervision).

Some salient differences across these four groups include the following: GPS and Jail cases are

most likely to result in a conviction (GPS = 58.7%, Jail = 58.8%) while those placed on RF are

the least likely to be convicted (48.0%). Violations in the short term are most likely to occur

among GPS cases (13.0%). GPS defendants are also the most likely to be arrested in the short

term (9.8%). In contrast, the likelihood of defendants being arrested in the long term is very

similar for GPS (10.9%) and RF (9.3%) cases, and those released from jail with an electronic

tether are only modestly less likely to be arrested in the long term (7.1%).

Table 3C-22. Descriptive statistics (mean values) of variables across four groups: South site

GPS

(N=177)

(N=167)

Jail

(N=42)

Bond

(N=218)

Convicted Cases***

.587

.480

.588

.353

Short-Term Outcome – Violation***

.130

.097

.027

Short-Term Outcome – Re-Arrest

.098

.058

----

.037

Long-Term Outcome – Re-Arrest

.109

.093

.071

.042

Male***

.927

.886

.857

.720

Female

.073

.114

.143

.280

White

.508

.443

.548

.564

Black

.492

.557

.452

.438

Hispanic

.136

.108

.071

.092

Age at DV Incident

33.655

31.802

34.452

33.995

Number Prior Arrests*

3.870

3.479

5.857

2.156

Number Prior Arrests for DV*

.401

.425

.333

.229

Current Crime – DV Assault**

.181

.120

.023

.087

Current Crime – DV Battery***

.684

.814

.857

.894

Current Crime – DV Other***

.136

.066

.120

.018

Current Crime – Weapon Used**

.243

.143

.048

.119

T-tests of mean differences across the following groups are significant at the p<.05 level. Age – RF/Bond. Number of prior

arrests – GPS/Jail, GPS/Bond, RF/Jail, RF/Bond, Jail/Bond. Prior arrests for domestic violence – GPS/Bond and RF/Bond.