.".""""Llact of Systemwide Drug Testing I
Oregon
Corporation ~.
:: ',
Institute of Justice Cooperative'Agreement:'#91-DD-CX-KOS7 ......
Report:', "
", '
,'
"
::
Acknowledgements
The authors would like to thank the Multnomah Department of Community Corrections, especially Tamara Holden, Cary Harkaway, and the branch chiefs. Their active assistance with this evaluation is deeply appreciated. Dr. Winifred Reed of the National Institute of Justice provided wonderful support and guidance throughout the project. We would also like to acknowledge the project team for their help on this project. Dr. Mark Kleiman, Dr. David Cavanagh and Jenny Rudolph of BOTEC for their leadership and direction; Bud Frank and Dick Damon for their help with field work; Andrew Chalsma for his help with data preparation and analysis; Margot Barnert, Andrew Hirshey, and Rose Marie Yu for file coding and data entry; and Peggie Arvidson for her skill and patience in producing this fmal document Very special thanks go to' Dr. Philip Wirtz, Professor of Management Science at George Washington University, for his expert advice on the hierarchical linear modeling. Thanks gang!
,.
Prepared under Grant No. 91-DD-CX-K057. Points of view or opinions in this document are those of the authors, and do not necessarily represent the official position or policies of the U.S. Department of Justice
T ABLE OF CONTENTS
......... .- ..... ,. ..................................... .
1
Drug Testing in the Criminal Justice System ...... ' .. . . . . . . . . . . . . . . . . . . . . . . ..
4
Drug Use and Crime .......................................... Systemwide Drug Testing ................................... : ..
4 9
Design and Methods ......................,.................. . .......
14
Introduction
The Evaluation Design ........... , . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 15 The Survey of Probation and Parole Officers ......................... 16 The Sample ................................................ 18 Data Collection Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 20 The Conceptual Model and its Operationalization ...................... 22 The Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 27 The Effects of DTE on Rearrest During Pretrial Release . . . . . . . . . . . . .'. . . . . . . . . .. 28 The Effects of DTE on Correction Officer Caseload Outcomes .. . . . . . . . . . . . . . . . .. 34
..,
Implications ...................................................... 47 4
References ....................................................... SO APPENDIX A. Results of the Survey of Multnorriah County Corrections Officers
/'
11
55
LIST OF TABLES AND EXHmITS
Exhibit A.
Evaluation Framework: . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23
Exhibit B.
Data for Impact Analysis . ..................................... . .. . '. . .. 24
Table 1.
Description of Sample of Pretrial Clients ......... . .................... .. ~ . . . 29
Table 2.
DTE Violations for Tested Pretrial Clients ..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Table 3.
Logistic Regression Odds Ratios of the Effects of DTE Participation on Rearrests of Clients on Pretrial Release .......................... 32
Table 4.
Regression Coefficients of the Effects of DTE Participation and Drug Problem Severity on Rearrests of Clients on Pretrial Release ............... 33
Table 5.
Description of Sample of Clients ........................................
38
Table 6.
Client Criminal History and Risk ..................... • .....•............
38
Table 7.
Arrests Predicted by Officer Use of DTB. Client Characteristics. and Officer Willingness to Request a Revocation Hearing .... . ... . . . . . . . . . . . . . . . . . . .. 41
Table 8.
Arrests Predicted by Officer Use of DTE. Client Characteristics.,and Officer Informal Sanctioning of More than Two Positive Drug Tests . . . . . . . . . . . . . . .. 42 "--"
Table 9.
Technical Violations Predicted by Officer Use of DTE. Client Characteristics. and Officer Willingness to Request a Revocation Hearing .......................... 43
Table lO.
Technical Violations Predicted by Officer Use of DTE. Client Characteristics. and Officer Informal Sanctioning of More than Two Positive Drug Tests ......... . ....
~ .
44
Table 11.
Total Violations Predicted by Officer Use of DTE. Client Charactenstics,and Officer Willingness to Request Revocation ....................................... 45
Table 12. ·
Total Violations Predicted by Officer Use of DTE. Client Characteristics. and Officer Informal Sanction of More than Two Positive Drug Tests ....................... 46 Officer DTE Use by Sanctioning for Positive Tests of Failure
Table A-I.
to
Appear for Test. . . . . .. 56
J'
Table A-2.
Actions Probation and Parole Officers Would Take . . . . . . . . . . . . . . . . . . . . . . . . . . ..
,~
III
57
INTRODUCTION
This impact evaluation assesses the effects of systemwide drug testing based on the experience of the Multnomah County Drug Testing and Evaluation Program (DTE). The primary policy question addressed by this analysis is whether DTE reduces criminal activity and increases compliance with court orders among defendants or offenders released under the supervision of the courts. Systemwide testing programs have comprehensive objectives that include assessing the type and level of drug use, evaluating the need for treatment, and establishing effective controls which permit drug-involved offenders to be released while protecting public safety and minimizing the risk of non-compliance with court orders or appearance at scheduled
.
,
hearings. Used in conjunction with supervised. pretrial release, probation or parole, these programs are intended to support comprehensive client management in which testing is used to monitor arrestees or offenders released into the community. Administratively, systemwide programs link: multiple public and private agencies--such as pretrial services, jails, treatment agencies, and probation and parole offices through coordinated procedures or central administration by a unit created for this purpose. The Drug Testing and Evaluation Program (DTE) was funded as an 18-month demonstration program by the Bureau of Justice Assistance. The program, operated by the Multnomah County Department of COmnlunity Corrections, provides regular random drug tests designed to monitor compliance with release conditions and progress in treatment programs and signal court supervisors when intervention is needed. DTE services also include standardized evaluations of the severity of substance abuse and need for treatment to 1
facilitate early intervention and appropriate referrals to treatment. There are two main divisions to the Multnomah DTE Program: pretrial and postsentence. DTE's post-sentence programs serve many more clients than the pre-trial program, and consist of four components: ,
i
I
1.
drug testing and evaluation of probation and parole clients;
2.
drug testing of clients in women offender programs;
3.
drug testing of at-risk women offenders in residential drug treatment and in the community; and
4.
drug testing in the Community Service Forest Project, a 30-bed work program for sentenced male felons oper~ted by the Multnomah County Department of Community Corrections in partnership with the US Forest Service.
This impact evaluation examines the effects of drug testing"on the criminal activity and compliance with court-ordered conditions of release for two groups of DTE clients: (1) those on pretrial release, and (2) those on regular probation or parole (number 1 on list above). Evaluation of the pretrial component of DTE is based on an experimental design which compares a treatment group randomly selected from eligible pretrial clients to a control group of pretrial clients who were not selected for DTE services. The random assignment allows the assumption that differences between the groups are not systematic and an analysis ,.
that compares differences in individual client outcomes in the two groups. The evaluation of the probation and parole component of DTE is based on a crosssectional comparison of caseload outcomes for corrections officers who differ in use of DTE. The design takes advantage of the naturally occurring variation across officers in DTE
.2
utilization and sanctioning of positive drug tests or failure to appear for testing, while avoiding the risk of using a clearly non-comparable comparison group of clients not enrolled in DTE. Hierarchical linear models are used to compare rates of arrest and technical violations across officers, controlling for characteristics related to risk of violations among their clients, and examining the interaction between the use of DTE and officer sanctioning for drug testing violations. The following section describes the research that supports the potential utility of drug testing programs, the key components of systemwide drug testing, and the Multnomah County DTE program. Section 3 presents the conceptual framework that guides the evaluation and a description of study methodology. Sections 4 and 5 present the fmdings on the impact of DTE during pretrial and during probation and .parole, followed by a discussion of implications in Section 6.
,.
3
DRUG TESTING IN THE CRIMINAL JUSTICE SYSTEM
Drug testing programs have been developed in response to evidence of extensive drug use among the criminal population and close connections between drug use and criminal activity. Drug testing within the criminal justice system recognizes the potential benefits to reducing or, at the very least, monitoring drug use among those accused or convicted of a crime while in the community and under supervision of the court. The Multnomah County DTE program, described below, was designed as a comprehensive, systemwide drug testing program for offenders on community release, both before and after sentencing. Additional details on Multnomah's DTEprogram are provided in the report on the process evaluation prepared as part of this study.l The results from this evaluation should be generalized to programs similar to those described in our process evaluation and not to programs that differ in the population tested, the frequency of tests or the use of sanctions.
Drug Use and Crime An extraordinary proportion of crime can be attributed to drug dependent offenders
(Chaiken 1986; Gropper1985; Inciardi 1979; Johnson, Goldstein, Preble, Schmeidler, Lipton, Sprunt, and Miller 1985). . The National Institute of Justice's Drug Use Forecasting (DUF) program consistently finds that a majority of arrestees in major cities test positive for drugs at J'
the time of their arrest (NIJ 1990). Substance abusers, especially offenders who use heroin and cocaine, have been found to exhibit extremely high crime rates (Ball, Rosen, Flueck, and
Cavanagh. David. 1994. A Process Evaluation of the Multnomah County Drug Testing and Evaluation Program. Report to the National Institute of Justice under NU Cooperative Agreement #91-DD-CX-K057. Cambridge. MA: BOTEC Corporation. February.
4
Nurco 1981; Ball, Corty, Petroski, Bond, and Tommasello 1986; Chaiken and Chaiken 1983', Collins, Hubbard, and Rachal 198.5; Johnson, et al. 1985; McGlothlin, Anglin, and Wilson 1977). As the severity of drug abuse increases among users, the frequency and severity of their criminal behavior rises dramatically (Chaiken 1986; Chaiken and Chaiken 1982; Collins. HUbbard, and Rachal 1985; Speckart and Anglin 1986a, b). Heavy illegal drug use also seems to be one of the most important factors which distinguish criminals with persistently high frequencies of violent criminal acts from other types of offenders (Chaiken and Chaiken
I
1982; Rolph and Chaiken 1987; Dembo et ale 1990; Visher 1990). Drug abuse is linked to crime through: (1) the psychopharmacological effects of the drug which lead to crimes conunitted while under the influence; (2) econoriucally compulsive crimes committed to support drug consumption; and (3) systemic crime associated with drug. ~
.-
transactions and marketing (Goldstein 1985). In the underground economy, non-using dealers engage in, and are victimized by, violent crime as part of their business (Falkin, Wexler, and Lipton 1992; Brounstein, Hatry, Altshuler, and Blair 1989). They are joined in drug selling by users who traffic and engage in property crimes to generate income for drug consumption. User-sellers may lure others not only into drug use, but also into criminal behavior in anticipation of large profits (Chaiken and Chaiken 1982; Goldstein 1985; Johnson, et ale 1985). Research also supports the folk wisdom that heavy drug users resort to other crimes in order to
su{>POrt their drug habit (Chaiken and Johnson 1988) and use drugs as part of the
crime committing process (Johnson, et ale 1985). Drug use plays a major role in recividism. The vast majority of the nation's prisoners,
5
more than 85 percent, are recidivists, and about three-quarters have histories of drug use (Innes 1988). Criminal offenders who are regular users of hard drugs or of multiple drugs are typically at high risk of recidivism after release from prison (Chaiken and Chaiken 1982; Innes 1986; Wexler, Lipton, and Johnson 1988).
In California, drug possession or use is a
contributing factor in 64 percent of the cases in which parolees are returned to prison for parole violations (Blue Ribbon Commission 1990). In general, drug-dependent criminals lead lifestyles characterized by self-destructive and antisocial behaviors; they also have problems related to the absence of job training, dependence on others, and frequent conflict with criminal justice authorities (Collins, Hubbard, and Rachal 1985; Wexler, Lip~on, and Johnson 1988). Controlling the drug use of arrested or convicted offenders
by the criminal justice
system is key to interrupting the criminal careers of drug-involved offenders. Studies have found that the threat of sanctions for drug use appears to enable many offenders to at least temporarily desist from the use of illegal drugs (Carver 1986; BOTEC Analysis Corporation 1987; BOTEC Analysis Corporation 1990). Drug testing has been found to lower drug-using arrestees' re-arrest rates and rates of failure to appear for hearings (Toborg, Bellassi, Yezer, and Trost 1989). Prison drug treatment programs have been found to be at least moderately effective at weaning participants from illegal drug use and are highly cost-beneficial in terms of the crime they prevent (Chaiken 1989; National Institute on Drug Abuse 1988; American Correctional Association 1981). Since 1984, Washington, D.C. has conducted urinalyses of arrestees and defendants at the time of arrest and, for those testing positive, during the period of pretrial release. Smaller
6
experimental programs have been implemented in other jurisdictions (Wish, Toborg, and Bellassi 1987; Dembo et aL 1990). During pretrial release, regular, random tests are conducted, and those who test positive are penalized with jail time. The rationale for these programs is that offenders who continue to use drugs are expected to be less likely to appear in court and more likely to engage in criminal activity than offenders who never used drugs or abstain from drugs. Evaluation of the DC drug testing program indicated that drug tests conducted shortly
(1992) found that the urinalysis results predicted recidivism independently of other predictors of risk and were more effective in predicting recidivism among"some groups of offenders, e.g., novice offenders and employed offenders. than others. Despite the relationship of prearraignment drug test results to behavior while on release, subjecting these offenders to drug testing during pretrial release did not result in lower rates of arrest or failure to appear than the rates for offenders assigned to drug treatment without testing or a control group (Toborg, et al. 1989; Vischer 1992). Subsequent replications of the DC program found that: (1) pre-arraignment drug testing did not contribute to predicting failure to appear independently of other available "
infonnation; and (2) testing of offenders during pretrial release did not result in lower rearrest rates and failure to appear rates (Kapsch and Sweeny 1990; Goldkamp, Jones, and Gottsfredson 1990; Gottfredson, Britt, and Goldkamp 1990). However, these replication programs experienced significant problems in program implementation which may have
7
weakened the validity of the evaluations (Goldkamp 1989; Visher 1992). These findings have led to conclusions by some ·e xperts that pretrial drug testing is not cost-effective (Belenko, Mara-Drita, and McElroy 1992). Evaluation of drug testing during probation and parole has also produced mixed I
i I
results. Evaluation of Drug Intensive Supervision Probation (ISP) in five sites found no
J reduction in officially-recorded recidivism, regardless of the drug testing schedules, system responses, and offender characteristics (Turner, Petersilia, and Deschenes 1991). However, the authors note that sanctions for testing failures were not consistently applied and found that technical violations increased due to improved .offender monitoring, which increased the pressure on overcrowded jail .a nd prison facilities. The implication is that unless drug testing programs plan resources for sanctioning failures consistently and swiftly, the programs are unlikely to deter continued drug use and criminal activity. Other studies suggest that urinalysis may be an effective tool for managing offenders when coupled with speedy and certain sanctions. The Drug Reduction of Probationers (DROP) program found that drugtested offenders respond to sanctioning. The DROP program in Oregon arrests and jails probationers for two days after each positive drug test. Offender drug use declined after penalties were applied: 53% tested positive again after their first sanctioning, 24% tested positive again after a second sanctioning, and 6% tested positive again after a third
,.
sanctioning. Overall, positive drug tests among probationers declined from 43% before program implementation to lO% .after program implementation (Kushner 1993). This finding is consistent with other studies which recommend combining legal sanctions with treatment (Hubbard, et al. 1989; Anglin and Hser· 1990).
8
Systemwide Drug ,Testing in Multnomah County The components of
amodel systemwide drug testing program are summarized by
Cavanagh (1993) as follows: Pretrial drug testing is an important component of a comprehensive systemwide drug testing program. In this component, the initial testing occurs upon entry into the criminal justice system, at booking and/or following an initial hearing. If the initial testing is conducted at booking, criteria for inclusion in the program are prespecified, and the results are made available to the court at the time of the initial hearing to serve as a basis for determining the conditions of release. The initial booking test results should be provided to the magistrate, judge, or other presiding official as well as to the arrestee's counsel and any prosecuting officials. Presiding officials, defense attorneys, and prosecutors should all have a clear understanding of how drug tests at booking, if used, enter into the decision to release and the setting of release' conditions. The drug test report could enter into this decision in one of two ways: as part of a decision-making algorithm or as one additional factor to be considered in a case management system. In the fust scenario of a decision-making algorithm, a simple set of rules, based on factors such as drug test results, community ties, and so on, constrain magistrates' decisions on whether an arrestee will be released and the conditions of release. In the case management approach, drug test results are one of a number of individual factors that magistrates are required to consider when deciding whether to release an arrestee and the conditions of release. If release decisions are based on case management, the administering agency should have an advocate at arraignments to explain the implications of drug test records to presiding magistrates. With or without drug tests at booking, a systemwide drug testing' program will include testing as an option for inclusion as a condition of pretrial release. The hearing magistrate or judge at an initial hearing or arraignment can order testing on the basis of reports from pretrial services or records on the arrestee's criminal history or selfreported drug use. The order of release should specify the frequency of drug testing and its circumstances (e.g., random versus fixed schedule) and may include required treatment. The pretrial testing component of a systemwide program can be handled by a court unit responsible for pretrial services or by a central unit responsible for systemwide pretrial and post-adjudication testing. The tests can be conducted by a lab under direct contract to the courts or by another agency (e.g., a TASC program or treatment provider) under contract. If the drug testing program is not administered at a central location, the police or program staff at the booking site should use predefined procedures to collect the specimen and relevant information to be forwarded to the
9
supervising unit or contractor. Post-prosecution Testing. Remanding convicted offenders to probation, sentencing them to prison or jail, or releasing prisoners on parole can and, in a comprehensive system, should include drug treatment and drug testing options. In addition, drug testing can be included at the point of diversion in those jurisdictions which use a prejudgement, post-prosecution option for monitoring offenders. Goals of post- , prosecution drug testing are somewhat different from those at pretrial, focusing on longer-tenn outcomes. At this point, drug testing is part of a sanctioning/remediation program intended to increase compliance with other court-mandated sentencing conditions, to deter drug use by linking to sanctions for use, to improve treatment compliance, and--the bottom line-to decrease criminal activity. As described in the pretrial component above, the use of drug testing under these circumstances requires clearly delineated policies and procedures for when to require testing, how testing should be conducted, and how records should be maintained. The success of the program is likely to depend heavily on consistent testing, monitoring, record-keeping, and sanctioning drug use--factors to be considered in evaluation. Post-prosecution testing can be provided by community corrections agencies or by a central unit. Eligible participants may be identified using objective guidelines based on level of need for supervision (using offender classification systems) and/or the type or severity of drug use (using assessments conducted by the program, court-affiliated treatment programs or agencies such as TASC). In addition, special groups of offenders (e.g., pregnant drug-involved offenders, felony drug offenders, offenders in '-../ bootcamps or other special programs) may be targeted for drug testing. Again, the testing may be provided under direct contract to the court agency or by a courtaffiliated treatment unit. The process evaluation noted several departures from the model in the Multnomah DTE program that may have attenuated
pr~gram
impact.
In the pretrial component of DTE,
these departures included: • "
•
Poor participation compliance with testing requirements in the pretrial portion of the DTE program. The majority of PRSP clients in DTE were unsuccessfully terminated from the DTE program, mostly for failure to appear for testing. Between January 1991 and October 1992, only 16.5% (86) of PRSP clients in the DTE program successfully terminated the DTE program. However, most of. these same clients (70%) successfully terminated the PRSP. Poor participation in, and utilization of, pretrial evaluations of the severity of drug problems. Of the 615 clients who entered the PRSP DTE program between January 1991 and October 1992, only 304 were evaluated due to
10
backlogs in scheduling as a result of staff illness, limited time on ·pretrial release due to expedited docketing of drug cases, and the failure of more than half of all clients to appear for scheduled evaluations.
•
Limited early entry into treatment as a', result of DTE participation. Very few pretrial clients ended up in treatment before the end of pretrial supervision.
Improved detection and appropriate, early treatment referrals were the primary reasons why the Multnomah Drug Testing program included the capacity for professional assessment of offenders referred to the program. Several reasons account for the relatively few client evaluations completed by the program: clients failed to appear for many appointments, the referrals for evaluation came primarily from the pretrial program and could not be scheduled before the end of the pretrial release period, and competition for treatment slots was intense. The lack of evaluation and 'referral services for pre-trial clients is unfortunate because the relatively small number of evaluations which were completed indicate substantial need for various kinds of interventions and treatments in this population. Among those who were evaluated, the results show that:
•
22% needed medical treatment
•
36% needed family or social counseling or intervention
•
27% needed treatment for alcohol dependence
•
52% needed treatment for drug addiction
."
42% needed assistance with legal issues
•
23% needed psychological counseling or treatment
•
34% needed employment counseling
In the probation and parole component of the DTE program, the departures involved primarily a shift in the way in which officers used the testing program (the evaluation design 11
was adapted to reflect the implementation practices). Significant aspects of the way postsentencing drug testing is used in Multnomah include; •
Case selection practices varied from branch to branch and officer to officer. The allocation of DTE slots to probation/parole officers varied among the branch offices. Most branch offices relied on a "first come, first served" method for providing available DTE slots and decisions were based on·officer perceptions of the need for testing.
•
Few clients were referred for evaluations. Probation and parole officers viewed treatment as scarce, expensive or difficult for clients to access, and thus often used testing alone without a treatinent referral.
•
Formal sanctions for testing failures were used infrequently. Although officers were careful to record all violations of the testing conditions in their records, formal sanctions such as revocation hearings or jail were rarely applied for testing violations due to lack .of jail space and pressure on the court system.
•
Informal sanctions and heightened client surveillance were used in response to testing failures. Informal sanctioning procedures were widely reported by officers when clients failed to appear for tests or ttsted positive for drugs. These informal sanctions could include a search of the client's home for drugs, drug paraphernalia, or other contraband. additional drug testing. additional reporting requirements, and other limits. In some cases, drug testing was instituted as a form of increased surveillance for other types of infractions.
The plans for the evaluation of systemwide drug testing in Multnomah. described in the following sections, were shaped by the findings of the process evaluation. The evaluation of testing for probation and parole clients focuses on the use of DTE in combination with the supervisory practices of corrections officers, examining the main effects of drug testing and the use of drug testing in combination with fonnal and infonnal sanctions for non,.
compliance. The evaluation of testing for pretrial clients examines differences in arrest rates between tested and untested clients on the grounds that testing might deter continued drug use and criminal activity. However, the effects of DTE on entry into treatment or appearance at court hearings was not examined because lack of sufficient exposure to the planned 12
intervention (assessment of drug problem severity, early treatment entry, and sanctioning for positive urinalysis results) would make interpretation ambiguous.
'w
/'
13
DESIGN AND METHODS
The research was designed to evaluate whether DTE reduced criminal activity and increased compliance with court orders among defendants or offenders released under. court supervision. Theoretically, DTE is expected to achieve these outcomes through: (1) the deterrent effects of sanctions for continued drug use; and (2) improved detection of offenders in need of treatment and more appropriate treatment referrals. Deterrent effects are achieved
if offenders reduce their drug use or abstain entirely from drug use to avoid penalties. Thus, the success of ~ drug program aimed at deterring drug use is expected to depend on the certainty, severity, and perhaps the celerity, of negative sanctions for testing positive or failing to appear for tests without an acceptable, .verifiable
excus~..
The success in improved
detection and treatment referrals for offenders with serious drug problems is expected to depend upon using the results of assessments of drug problem severity to get offenders into appropriate treatment programs. The impact analysis focuses on the deterrent effects of DTE. Because so few client drug assessments were completed, no analysis of their impact on treatment utilization could be conducted. The effects of the pretrial use of DTE and the probation and parole use of DTE were examined separately.
The evaluation of the pretrial phase of the DTE program was based on
an experimental design which compared a treatment group of 169 clients randomly selected from eligible pretrial clients to a control group of 83 clients who were not selected for DTE services. The evaluation of probation and parole use of DTE was based on a cross-sectional comparison of caseload violation rates for 53 corrections officers, based on outcomes on 504
14
clients under their supervision. The following sections describe the designs and procedures used for these evaluations.
The Evaluation Design The evaluation of the pretrial phase of the DTE program is based on an e'xperimental design. Eighty-five percent of the clients were randomly selected for testing from pretrial release clients referred to DTE. For this study, a sample of tested clients are compared to a control group comprised of all eligible pretrial clients not tested. The evaluation of DTE as used by probation and parole officers is based on a crosssectional comparison of 53 officers that varied in their use of DTE and their response to testing violations (failure to appear for testing 'or testing positive for drugs), based on outcomes for 504 clients under their supervision. This design was adopted because equivalent
.•
groups could not be constructed for quasi-experimental comparisons. Assignment of clients to DTE was made by the supervising probation or parole officer based either on: (1) conditions' set forth in the court-order authorizing the offenders release, or (2) officer perceptions of client need for drug use monitoring. TIris process for deciding who entered DTE was highly selective and often based on transitory events, rather than stable characteristics of the client, such as risk category and offense, which could be measured and used as control yariables. As a result, DTE clients may have been more likely to violate probation/parole than other clients, even after controlling for group differences, which would bias estimates of treatment impact downward. It was also not possible to coristruct appropriate pre-post comparisons of comparable clients supervised before and after DTE implementation. The county and state probation and parole offices were merged in July 1991, just prior to the study period, and
15
,
resulted in such significant changes in client handling that pre-post comparisons would not be valid. The selected design is based on the more plausible assumption that, on average across probation officers, groups of clients within risk categories will have the same likelihood of "failure." That is, that the propensity of high risk cases to reoffend or fail to complete probation requirements is the same for Officers A, B, C ... n. So that in the absence of DTE, the same failure rate is expected for all officers, controlling for differences in the risk level of their clients. This leads to a design that compares client outcomes controlling for risk category, officer use of DTE, and other officer characteristics expected to affect outcomes. This approach does, however, suffer from the same shortcoming that plagu.e s quasiexperimental comparisons, namely that variables not included in the analysis may explain differences in outcomes and these cannot be assumed away in the absence of random assignment.
The Survey of Probation and Parole Officers The evaluation design for the probation and parole use of DTE builds on the survey of probation and parole officers in Multnomah County conducted as part of 'this study. The decision to conduct this survey resulted from the descriptions of how officers used DTE as a case management tool provided in a limited number of in-person interviews during our initial site visit in January 1992. To collect more systematic information on how DTE was used by corrections officers, a mail survey was conducted in August and September 1992. The questionnaires were distributed to all officers by the branch chiefs.
16
,
!
I -.J/
,
Principle findings included: •
Officer discretion was a key determinant of who got tested. Over half the officers (57%) used DTE for clients suspected of current drug use; only 37% said they referred clients to DTE because it was required by their court or parole board orders.
•
Not all clients suspected of using drugs or needing treatment were . referred to DTE. Seventy percent of the officers said that 1/5 or more of their clients in the prior month were using drugs, and 64 percent thought that 1/5 or more of their past-month clients needed drug. treatment However, only 18 percent of the officers reported that 1/5 or more of their past month clients were in
DTE. •
Officers differed widely in the likelihood of referring a drug-using client to DTE. The ratio of clients in DTE to clients thought to use drugs showed considerable variation. For 1/3 of the probation officers, the difference between the percentage using drugs and the percentage in DTE was 11 % or less. For another third, the difference was 13% to 34%. Thus, for one third the difference was 34% or higher (up to 76%).
•
Officers' accounts of how they respond to DTE violations indicated variation in the use of formal and informal sanctions. The officers used · many forms of informal sanctions, including referrals for additional counseling or drug treatment, more intensive drug treatment, requiring additional drug tests, making home visit inspections, and adding extra requirements (calling in, keeping a log, maintaining a curfew). A significant portion used warnings or reprimands for missed tests, but relatively few did so if the client tested drug positive (13% reported warnings for I or 2 positive tests; 2% for more than 2 positive tests) (see Tables A-I and A-2 in Appendix A). One-third of the officers mentioned using DTE to increase supervision, and 21 % used DTE to manage the client
•
Officer sanctioning practices did not vary significantly by the proportion of clients in DTE"
•
,. Offtcen were concerned about the lack of effective sanctions for DTE violations. This problem and the problems of insufficient drug treatment slots and hoors of operation which interfered with client employment were cited as major problems with DTE by about half of the officers. The accuracy and timeliness of drug test results were widely endorsed.
The probation/parole officers in our survey provided a variety of informal sanctioning 17
procedures they might use to deal with clients who violate DTE restrictions. These included: providing warnings
~r
reprimands (44%), adding
addi~onal
probation or parole requirements
'-..,....-
(42%), requiring clients to obtain new or additional counseling and/or treatment (27 %), requiring clients to submit to more frequent drug testing (12%), and subjecting clients to home visits (5%). Overall, the results of the survey are consistent with the key assumptions for the impact evaluation: (1) there was variation in DTE utilization among officers, (2) there was variation in sanctioning of violations of testing conditions among officers, and (3) the variation in DTE use was independent of other supervisory practices and client risk level.
The Sample The Pretrial Sample. Multnomah County's pretrial DTE program serves PRSP clients with a prior or current felony drug charge or who admit to drug use. During the period from January 1991 thiough October 1992, the PRSP program supervised about 300 clients per year '-" who were eligible for pretrial DTE. Although 50 DTE-eligible clients were released to PRSP supervision each month, only about 35-40 pretrial DTE slots became available each month. PRSP case managers randomly selected over 80% of incoming PRSP clients for the DTE program based on the last digits of their identification number. The remaining eligible clients did not participate in the DTE program and are used as a control group for this study. The evaluation is .based on a sample of 252 pretrial cases: a treatment group of 169 DTE clients, randomly selected from those tested, and
a control group of the 83 eligible clients who were
not tested. Forty-eight cases orginally selected for the sample were excluded because the records on arrests could not be obtained or matched to DTE files.
18
The Probation and Parole Sample. The bulk: of post-sentence clients are in the probation and parole program. This program has slots for testing 415 clients. The number of client enrollments in a given year can exceed the total number of available slots since clients typically spend less than a year in the program. Between December 1990 and October 1992, 2449 clients entered DTE. During the period of this study clients spent an average of 107 days in the DTE program. Thus one slot could serve more than three clients per year. A two stage sampling procedure was used to select a sample of probation and parole clients. At stage 1, all probation officers who responded to the Officer Survey on DTE conducted in the Fall of 1992 and were employed as probation officers in Multnomah County in August of 1991 were included in the sample if their caseloads included medium or high risk-clients (or clients with special drug conditions) who might be referred to DTE. The final sample included 53 officers employed in Multnomah County comihunity corrections in August of 1991. Of 99 eligible officers in Multnomah County at the time of the survey in October 1992, 86 (87%) completed the survey. Of these, seven officers were eliminated because they did not supervise any clients with drug conditions or medium or high risk classifications; 21 were deleted because the officers did not identify themselves on the survey so we could not match their clients; and five were deleted because they were new to the department and did not have active clients in August through October 1991. At stage 2, a random sample of 802 clients supervised by officers from August to October 1991 was selected for the study. This time period was chosen to provide a sample of clients -after the DTE program was fully implemented and after the merging of the County and State corrections July 1, 1991. It provided a sufficient number of clients and an
19
--= .-----
r
I
18 month follow-up period during which to observe client outcomes. The sample was randomly selected from a computer printout prepared by the Oregon Offender Profile System. The printout listed all clients active during August, September and October 1991 by branch and probation officer with the risk classification, a code for ' special conditions associated with drugs, and a state offender identification number. High risk offenders were deliberately oversampled to increase the proportion of DTE-eligible clients in the sample and comprise 40% of the sample, with another 40% classified as medium risk. Selection was proportionate to caseload size, with a minimum of 13 clients per caseload.
Data Collection. Procedures Data for the analysis were collected for both the pretrial and post-adjudication samples , from three computer systems:' (1) the Portland Police Department database (PPDS); (2) the
'.
Oregon State Law Enforcement Data System (LEDS); and the DTE program fIles. Additional data for the sample of probation and parole clients were collected from the Oregon Justice Information Network System (OJINS) and the Oregon probation and parole tracking database (AS400). •
The Oregon State Law Enforcement Data System (LEDS) 'and the Portland Police Data System (PPDS). Records maintained by law enforcement agencies for the state of Oregon and for the city of Portland include arrest information (date and ch'arge). LEDS contains information on prior conviction (date and charge), dates of incarceration in' jail or prison, and dates of probation or parole revocation. ,-
•
,Data from the DTE Program Files. Records were provided for all DTE clients seen before December 1, 1992. Data includes the testing results (number, outcome) and evaluation results of clients in DTE as well as demographic descriptors, and dates ·of entry and exit.
•
The Oregon Justice Information Network System (OJINS). This system contains court records of court orders conditions and records on officially
20
...... ~ ...
-.-'
recorded violations of probation or parole conditions.
•
The AS-400 System. The Oregon probation and parole tracking database (AS400) contains current information on client status, the type and duration of supervision by each officer, score on the Oregon Risk Assessment instrument and risk classification category. '
The SIDS number or name and date of birth was entered into the network computer system used by Multnomah County corrections officers and selected screens containing requested information were printed. The printed screens on each client from multiple data systems were sent to The Urban Institute for coding and data entry. Data from the various systems were merged by individual SIDS number or by social secUrity number. Data on officer sanctioning and supervisory practices, attitudes towards DTE, and use of the DTE were merged by officer identification number. Plans to ·collect data from hard copy probation and parole records on compliance with ·W
conditions other than DTE (including attendance at drug treatment, other types of treatment, ,~
i~'
community service, restitution, etc) and sanctions imposed by the officer were abandoned because this information could not be retrieved reliably for a sufficient number of cases. Missing data from key records resulted in dropping a number of cases from the sample as follows: 287 cases were deleted due to missing OJINS files, and 11 were deleted because matches could be made on SIDS or social security numbers. There is no reason to believe data was correlated with DTE participation or recidivism. ,.
Although most clients (55%) were supervised by only one officer between August 1 and January 1993, 27% were supervised by two officers and 18% had more than two For the analysis, clients were assigned to the officer who supervised them for the period of time. 21
"
.
The Conceptual Model and its Operationalization The analysis of the impact of DTE examines the effects of DTE on client criminal activity, and, for the post-adjudication use .of DTE, the interaction between DTE and officer characteristics which affect the supervisory context within which DTE is used.
The analysis
is guided by a conceptual model (see Exhibit A) in which program benefits or outcomes are hypothesized to be a function of participation in drug .testing. Exhibit B illustrates the variables used to operationalize the model, with category labels or unit of measurement in the second column and data source in the third column. Program Outcomes. The outcomes of interest (shown in the block on the far right of Exhibit A) include arrest, technical violations of the conditions of court orders, and the number of violations of any kind (arrest, technical violations, and other infractions) recorded in the LEDS and OJINS systems. The outcomes are measured as n'llmber of violations per 30 days: (1) during DTE participation, and (2) between the start of DTE participation and January 1993. Violations per month (30 days) of time spent on probation or parole were used in an effort to control for differences in opportunity to commit violations in the absence of "'Vll.n~'~"'l'l'"
reliable data on days of imprisonment and thus the incapacitating effects of Because it was not possible to identify which violations involved a positive
test, DTE participants may have higher violation rates because the risk of detecting drug was increased by testing. For this reason, the analysis examines arrests, which should not affected by DTE participation, separately. Exposure to DTE. These outcomes are expected to be a function of the DTE program . supervision while on community release (shown in the blocks in the center of the page).
22
~
l
( Outcomes
Corrections Practices
Offender Characteristics: * age * sex * race
DTE * DTE
Current Supervision * probation/parole * offense * duration * risk classification * risk score
Prior criminal history/risk * prior drug arrests * prior other arrests * prior convictions
-----..1
participation * test results . * Drug problem assessment
Failure on Probation/Parole
* violations per month * rearrests per month
* technical violations . per month
Officer Characteristics * DTE, utllzatlon * case load * fonnel! sanctioning * Inforrtlal sanctioning * experience
/L---I
_ - - - - I
Exhibit A. Evaluation Framework
Exhibit B Data for Impact Analysis Variable
Measurement
Source
Client Characteristics Demographics Age Sex Race
18-25;26-34;35+ years old male/female white/African-American/other
LEDS LEDS LEDS
probation/parole . drug/violent crime/property crime/ other offense high/other days of probation/parole score on Oregon Risk Assessment Instrument at start of supervision
AS400 LEDS/pPDS
# prior drug arrests # prior arrests - nondrug # prior convictions
LEDS/PPDS LEDS/PPDS LEDS
yes/no · nwnber of tests any and by drug nwnber and % of total tests nwnber and % of total tests any 'and by drug yes/no cwnulative # days 0-3 times
DlE DlE OlE DlE OTE OlE OlE
average # clients in past month years 3 = very tough; 2 = moderately tough; 1 not very tough
Survey Survey Survey
1 = would use only informal sanctions
Survey
Current Supervision Type Offense Risk classification
Duration of supervision Risk score Criminal History/Risk Drug offenses Other offenses Convictions DTE Experience . Any testing Testing history Failure to appear for test Positive drug tests Any drug assessment . Duration in DlE # times in DlE Officer Characteristics Caseload size Experience in corrections Willingness to request revocation Informal sanctions 1) for more than 2 positive drug tests 2) for more than 2 missed J.
J.
=
AS400 AS400 AS400
MIS MIS MIS MIS MIS MIS MIS
o = would use formal sanctions
drug tests
Probation/Parole Outcomes Total violations Rearrests Technical violations Drug violation
# per month
per month # per month average # of violations for positive drug test. test no-show. or self-reported drug use #
24
LEDS/OJINS LEDS/OJINS LEDS/OJINS LEDS/OJINS
Participation in DTE is the key treatment variable, although data on the duration of DTE
J
participation, number of tests, test outcomes, and drug severity assessment results are also available from the DTE MIS and were examined. Officer Characteristics. The second block of intervening variables includes officer characteristics such as years of experience, caseload size during the prior month, use of formal and informal sanctions for testing infractions, and factors related to caseload management. These characteristics are used in the analysis of DTE during probation and parole because they may affect outcomes directly and interact with DTE to affect outcomes. Officer characteristics were based on information provided by officers in the survey conducted as part of this study in August-September 1992. Use of penalties for noncompliance with DTE requirements was measured by questions about how the officer would respond to four types of violation: failure to appear for testing 1 or 2 times, failure to appear for testing more than 2 times, testing positive 1 or 2 times, and testing positive more than 2 times. Officers were classified as "tough," "moderate," or "easy" in their use of revocation hearings for DTE violations as follows: tough officers were those who reported that they usually or sometimes recommended revocation for a DTE client who failed to show up for a drug test once or twice. Easy officers were those who said sometimes or never recommended ·revocation for more than 2 dirty tests. The remainder were classified as moderate. Based on ,,~~,
this definition, 41 % were classified as tough, 33% as moderate, and 26% as easy with respect
,'" to the use of formal sanctions. Those officers who used only informal penalties for DTE infractions were contrasted those who used a mix of formal and informal sanctions. Officers were asked what
25
requirements they would impose on DTE clients who test positive for drugs more than twice and classified as: (1) using formal sanctions if they
m~ntion
revocation hearings, jail, or
enhanced court-mandated conditions of release, alone or in combination with informal sanctions; or (2) using informal sanctions only if they mentioned increased surveillance, additional reporting requirements, additional testing, or increased drug treatment, but not any formal sanctions. 2 Officers were also grouped into these two categories based on their reports of what they would do if a client failed to appear for more than two scheduled tests; the 13 officers who said they would warn the client or do
n~thing
for more than 2 no-shows
were placed in a third category. Client Risk of ProbationlParole Violations. Client characteristics which may independently affect recidivism and compliance with court orders include the client's criminal '
.
" history, demographic characteristics, and current offense and status (as shown in the blocks on the left in Exhibit A). As a result, the analysis controls for age in years, sex, race (classified as white, African-American, and other), and current offense (drug offense, a Part 1 property . crime, a Part 1 violent crime, and other offenses3), as recorded in the LEOS system. The sis also controls for number of prior arrests on drug charges, number of prior arrests on charges, and number of prior convictions recorded in either LEOS or the PPOS. The 's score on the Oregon Risk Assessment Instrument, the risk classification category versU9·medium, low, or limited) at the start of supervision, and the number of days on
2
The question was phrased' hypothetically to avoid the problem that some officers might not have faced the violation pattern described in the question. Pan 1 property crimes include burglary. larceny. auto theft and arson. Part 1 violent crimes
include homicide. aggravated assault. rape. and robbery.
26
'-.../
probation or parole supervision were obtained from the AS400 system. The Data Analysis
The analysis of the pretrial component of the DTE program tests the hypothesis that the DTE participants had fewer rearrests following referral to the program than a control group of clients referred for DTE, but not served. SAS general linear model procedures are used to test this hypothesis. The analysis of the probation and parole component of the DTE program tests hypotheses that officers who make more extensive use of DTE have caseloads with lower rates of rearrest, technical violations and total violations, controlling for the risk of a sample caseload of clients under their supervision. In addition, the analysis examines the hypothesis that the willingness to impose sanc·tions for DTE violations reduces violations rates and interacts with use of DTE as a client management tool to achieve compliance with
...
conditions of release. These hypotheses are tested using hierarchical linear models (Bryk and Raudenbush 1992), specifying officer as subject. This procedure estimates the caseload outcomes for each officer based on the selected sample of clients in the officer's caseload and then estimates the effects for the sample of officers (the intercept) by specifying intercept as a random variable.
,.
27
THE EFFECTS OF DTE ON REARREST DURING PRETRIAL RELEASE
The OTE pretrial drug testing program serves clients in the Multnomah Pretrial Release Supervision Program (PRSP). The goal of the pretrial testing program was primarily to facilitate early diagnosis and intervention with clients using drugs. Pretrial testing was also intended to ensure that clients did not use drugs while ·on community release to reduce the likelihood of failure to appear for a scheduled hearing and to decrease the likelihood of .(~
continued criminal activity related to their drug use. Most OTE clients in PRSP were charged with drug felony offenses. The majority of these were male (79%) and most were white (58%). A comparison of the clients randomly selected for testing (n=168) to those not tested (n=83) shows no significant differences in the age, race, sex or type of charge between the treattnent and control group included in the sample. The two groups did not differ significantly in the number of prior arrests, the variable used to control for risk of continued criminal activity. Although 36% had been arrested less than three times before the current offense (34% of the treattnent group and 40% the control group), most had lengthier arrest records, with 18% reportiIlg 11 or more prior
The OTE testing records of those in the treattnent group show high rates of violations program requirements (Table 2). Only 14% of the clients appeared for all scheduled tests; % missed more than half of their schedUled tests. Of those who appeared for at least one (n=139), 60% tested positive on one or more tests. Over one-third (34%) tested positive more than three-quarters of their tests. The widespread violations of OTE conditions led to
28
Table 1. ....,J
Description of Sample of Pretrial Clients
Received DTE (n=168)
Did Not Receive DTE (n=83)
Total (n=251)
36% 36% 29%
33% 42% 28%
35% 38% 28%
Male Female
80% 20%
78% 22%
79% 21%
White Other
60% 31% 9%
54% 30% 16%
58% 31% 12%
Drug felonies Other felonies
88% 12%
88% 12%
88% 12%
34% 49% 17%
40% 41% 19%
86% 10% 3% 1%
87% 12% 0% 1%
86% 11% 2% 1%
54% 20% 10% 16%
55% 23%
55% 21% 10% 15%
Age Groups 18-25 26-34 35 and older Sex
Race
African-American Offense
Number of Prior Arrests 0-2 3-10 11 or more ' Number of Rearrests During DTE 0 1 2 3 or more
3 or more
10% 12%
J'
29
·W
36% 46% 18%
termination from the testing program but did not result in an unsuccessful completion of PRSP for most clients.
In general, DTE participants faced few penalties from the courts for
failure to appear or positive tests. The impact analysis examines the effects of DTE on any arrests: (1) during DTE participation, and (2) between referral to DTE and January 1993 (3 to 23 months depending on date of referral to DTE). Although 45% of the clients were rearrested, 25% more than once, after starting DTE, there was no difference between the treatment and control groups in the percentage arrested nor in the frequency of arrest, prior to controlling for other variables and the time available for rearrest (Table 1). The multivariate analyses shown in Table 3 fails to detect a significant difference in ~
.
;. : the proportion of the treatment and control group arrested during DTB participation, after controlling for prior arrests, age, race, sex, and offense type,· either during DTE participation (Modell) or following entry into DTE (Model 2). Modell in Table 4 examines the number arrests per 30 days between entry into DTE and January 1993 as a function of DTE controlling for other variables. Again, there were no significant differences ')8n~p.f~n
the treatment and .control groups.
The high rates of violations of DTE requirements and the lack of legal consequences violations may have rendered the program ineffective. Other analyses indicated, however, in this sample of tested clients drug problem severity was not related to rearrests. The pretrial clients were referred for a standardized assessment of the severity of a variety of nal problems. Evaluations were completed for 69% of the treatment group sample. The .ty of the clients (59%) had drug problem severity scores of six or higher, indicating
30
Table 2.
DTE Violations for Tested Pretrial Clients Percentage of Scheduled Tests Missed (n= 168) None 1-25% 26-50% 51-75% 76% or more
14% 16% 24% 18% 28%
Percentage of Tests Which Were Positive None 1-25% 26-50% 51-75% 76% or more
~"
40% 7% 12% 7% 34%
considerable or extreme problems and the need for treatment. Severe alcohol problems were less prevalent, -with 28% of those evaluated scoring six or higher 'on the alcohol problem severity scale. Model 2 in Table 4 shows that there was no significant difference in rearrests per month associated with differences in the drug problem severity score of those members of the treatment group who were ever given the OTE evaluation. The same was true for alcohol problem severity. One alternative explanation for the lack of impact on rearrest rates is that pro~lems
as measured by the OTE assessment procedures are not a significant
determinant of rearrests. However, this conclusion is inconsistent with the basic assumption I-
drug testing programs and with findings in the literature reviewed earlier, namely that activity is higher among drug involved clients. An alternative explanation is that the drug evaluation procedures need to be examined to confirm that they adequately drug problem severity when used in the context of a pretrial services program.
31
Clients may distort their answers to screening instruments used in this context and additional validation of the procedures is probably inciicated.
Table 3.
Logistic Regression Odds Ratios of the Effects of DTE Participation on Rearrests of Clients on Pretrial Release Modell Any Arrest During DTE Intercept
Model 2 Any Arrest After DTE Start
.26
NumberofPriorArres~
1.14
Drug Felony Charge
1.11
1.90
•••
1.12
•••
1.39
Race (other omitted)
'.
White
2.67
Black
1.30
.82
.39
.71
Male
1.20
Age Group (35+ omitted) .30
26-34
.72
.57
.78
.89
242
242
DTE Participation df '''':
'Y"
I:
•
18-25
•
P < .05 •• p < .01 ···p<.(xn
/ '
32
.38
••
lble 4. 'gression Coefficients of the Effects of DTE Participation and Drug Problem Severity on Rearrests of Offenders Pretrial Release Model 1 Number of Arrests After Start of DTE
:ept
.01
Jer of Prior Arrests
.01
Felony Charge
Model 2 Number of Arrests After Start of DTE
-.03
•••
.01
-.01
-.03
lite
-.02
-.01
ck
-.01
-.01
.03
.02
•••
'o ther omitted)
IUp
(35+ omitte
5
.05
.00
icipation
••
.06
.00
.01
-lem Severity
.01
~)
33
241
107
.13
.29
•
arrests on other or parole.
'F DTE ON CORRECTION OFFICER CASELOAD OUTCOMES
is described in
with the number
. uestion in the analysis of the impact probation and parole component of 's who made more extensive use of DTE had clients with fewer .
for cocaine, .:njt'u·.. ·o
and only a
olations than officers who used DTE less frequently, controlling for :lients in their caseload. A second, related question is whether DTE 'e r case management practices, specifically the use of formal and lfractions of testing requirements.
The research hypotheses tested
make more extensive use of DTE have lower rates of new arrests ts than officers less willing to use formal sanctions, controlling for
willing to use formal sanctions such as revocation for violations, f new arrests among their clients than officers less willing to use :ontrolling for client risk. lake more extensive use of DTE and are more willing to impose r violations have lower rates of new arrests among their clients than :olling for client risk. ike more extensive use of DTE have lower rates of technical eir clients than officers less willing to use formal sanctions, t risk. illing to use formal sanctions such as revocation for violations. :echnical violations among their clients than officers less willing to controlling for client risk. I
te more extensive use of DTE and are more willing to impose violations have lower rates of technical violations among their
hypothesis. officers who make more extensive use of DTE have rates of new ir clients as high or higher than officers who make less use of DTE. ent risk.
34
history of these clients who averaged .76 prior arrests on drug charges, 4.78 arrests charges, and 1.7 convictions prior to the offense for which they were on probation .or The DTE testing history of the 87 program participants in the sample is Table 6. On average just over 7 drug tests per DTE client were completed, with the of completed tests ranging from zero to 32. Most tests included the screen for cocaine, marijuana, and/or amphetamines. Less than half the tests included opiate screening and few screened for barbiturates or benzodiazapine. Many DTE participants violated DTE conditions and faced the risk of sanctions. Almost half (47%) of the DTE participants in the sample tested positive at least once and 20% tested positive more than twice. A relatively large portion of the offenders in the sample (54%) had at least one officially recorded violation of their probation or parole conditiOJ'ls. The total number of violations between August 1991 and January 1993 was 518: 184 new arrests, 138 technical violations of conditions of release, and 196 other violations (e.g.; failure to benefit from probation and other non-specific reasons). Thus, offenders in the sample averaged 1.03 violations, .37 new arrests, .27 technical violations, and .39 other violations in 17 months. The two dependent variables examined were new arrests and technical violations. New arrests represent new officially detected criminal activity. Because arrests generally result from police activity, not correction officer reports, the assumption was made that the J'
identification of new criminal activity is largely independent of the level of supervision. In contrast, technical violations are largely the result of correction officer detection of violations of conditions of probation or parole and may well increase as a result of the improved monitoring through drug testing. Thus DTE may increase the number of violations known to
37
&
Table 5. Description of Sample of Clients (N=504)
Number
Percent of Sample
127 191 126 60
25% 38% 25% 12%
Male Female
421 83
84% 17%
White
352 127 25
70% 25% 5%
Current Offense Drug-related Violent crime Property Other
154 166 105 79
31% 33% 21% 16%
Type of Supervision Probation Parole
378 126
75% 25%
Initial Risk Classification Limited Low Medium High
131 19 205 149
26% 4% 41% 40%
Any DTE testing
87
17%
Any DTE Assessment
-7
1%
Age of client 18-27 26-34 35-44 45-68 Gender
Race Black
Hispanic/Other
Table 6.
,-
Client Criminal History
Mean Number of Prior Drug Arrests
.76
Number of Prior Non~g Arrests
4.78
Number of Prior Convictions
l.70
38
tl
the officer and recorded in the corrections management information system. We expected to reject the first three null hypotheses using a one-tailed test and find that use of DTE reduces officially detected criminal ac·ti.vity among clients (arrests), particularly when used by officers more willing to impose formal sanctions for violations. Because DTE is likely to increase the detection of technical violations of probation and parole, fmdings to support hypotheses four through six would require that reductions in technical violations by clients would be of a magnitude to offset improved detection of violations, reducing the chance of fmding a significant reduction due to DTE. Table 7 presents the results of the effects of extent of DTE use and willingness to .
.
request a revocation hearing for DTE violations on the new arrest rate for officers. The HLM
.•
models were run first without controlling for client risk (Modell) and then controlling for client risk (Model 2). The analysis used hierarchicallinear models
(~M)
using maximum
likelihood estimation in which intercept for each officer was defmed as a random effect and all other variables as fixed effects. The results show that DTE use and willing to request revocation hearings did not significantly reduce the average number of new arrests per month, nor was the interaction between these two variables in either model. Similarly, officers· who mentioned only informal sanctions for more than two positive drug tests (Table 8) did not have significantly fewer arrests in their caseloads than officelSo /.
who used formal sanctions. The introduction of a control variable for client risk (Model 2)
'".
did not affect the results. As expected, caseloads with more high risk offenders had higher rearrest rates. Moreover, there were significant differences in the rearrest rates among officers (the intercept), controlling for differences in client risk.
39
Greater use of OTE, officer willingness to request a revocation hearing (Table 9).
a
preference for infonnal sanctions for more than two positive drug tests did not significantly reduce the number of technical violations (Table 10) . . The interactions between OTE and sanctioning practices were not significant. As with rearrests, the rates of technical violations were significantly higher among caseloads with more high risk offenders than other caseloads, and varied significantly among officers after controlling for client risk. Total violations also were not significantly related to the level of OTE use, officer willingness to request a revocation hearing (Table 11), a preference for informal sanctions for more than two' positive drug tests (Table 12), or interactions between OTE use and sanctioning practices. Ih addition to client risk, prior arrests and convictions of clients iJ1dependently predicted the total number of violations and
/.
40
Table 7. Arrests Predicted by Officer Use of DTE. Client Characteristi~s. and Officer Willingness to Request a Revocation Hearing
MOdel I
Model 2
Estimates
Estimates
Fixed Effects Within Officer Variables
Intercept
.43·
Age
.77** -.01
Prior drug arrests
.01
Prior other arrests
.00
Prior convictions
-.03
Risk classification
.44*··
Effects of Between-Officer Variables
DTE use
.06
-.13
Willingness to request revocation
-.02
-.05
DTE use· Willingness to request revocation
-.m ·
p <.OS •• P < .01 ... P < .001
.•
/.
41
.03
Table 8. ............"
Arrests Predicted by Officer Use of DTE, c:lient Characteristics, (lnd Officer Informal Sanctioning of More than Two Positive Drug Tests
Modell
Model 2
Estimates
Estimates
.29
.63*
Fixed Effects Within Officer Variables
Intercept Age
-.01*
Prior drug arrests
.01
Prior other arrests
.00
Prior convictions
-.03
Risk classification
.44***
Effects of Between-Officer Variables
DTE use
-.01
-.10
Infonnal sanctioning
.13
.04
-.08 ..
.04
DTE Use
* Infonnal sanctioning
• p < .OS •• P < .01 ••• P < .001
,.
\.
42
Table 9. Technical Violations Predicted by Officer Use of DTE. Client C;haracteristics. and Officer Willingness to Request a Revocation Hearing ,
Modell
Model 2
Estimates
Estimates
.25
.34*
Fixed Effects Within Officer Variables
Intercept Age
-.00
Prior drug arrests
-.01
Prior other arrests
-.01
Prior convictions
-.01
Risk classification
.27***
Effects of Between-Officer Variables
DTE use
.12
Willingness to request revocation
.02
DTE use * Willingness to request revocation
-.08
• p < .05 •• P < .01 ... P < .001
"
43
.10 .02
."
-.06
Table 10.
Technical Violations Predicted by Officer Use of DTE. Client Characteristics. and Officer Informal Sanctiol'-../ of More than Two Positive Drug Tests Modell
Model 2
Estimates
Estimates
Fixed Effects Within Officer Variables
Intercept
.33"
.48 ......
Age
-.00
Prior drug arrests
-.01
Prior other anests
-.01
Prior convictions
-.01
Risk classification
-.28*"'*
Effects of Between-Officer Variables
•
DTE use
-.21
-.25
Infonnal sanctioning
-.05
-.11
DTE use ... Infonnal sanctioning
.20
.27
p <.OS
•• p < .01 ••• P < .001
"
44
Table 11. TotaL VioLations Predicted by Officer Use of DTE, Client Chara.cteristics, and Officer Willingness to Request Revocation
Modell
Model 2
Estimates
Estimates
Fixed Effects Within Officer Variables
1.22***
Intercept
1.30*** -.01*
Age
Prior drug arrests
.04
Prior other arrests
.04**
Prior convictions
-In*
Risk classification
1.07***
Effects of Between-Officer Variables
•
DTE use
-.05
-.35
Willingness to request revocation
-.04
-.06
DTE use * Willingness to request revocation
-.10
.09
p < .05
•• P < .01 ••• P < .001
/.
45
! I
Table 12.
Total Violations Predicted by Officer Use of DTE . Client Characteristics. and Officer Informal Sanction'o/ More than Two Positive Drug Tests Model 1
Model 2
Estimates
Estimates
Fixed Effects Within Officer Variables
Intercept
1.17***
Age
1.06** -.01
Prior drug arrests
.03
Prior other arrests
.04***
Prior convictions
-.07*
Risk classification
.94***
Effects of Between-Officer Variables DTE use
-.39
-.44
Informal sanctioning
-.02
-.16
*
.16
.36
DTE use •
Informal sanctioning
p <.OS
•• p < .01
••• P < .001
J'
46
IMPLICATIONS
The results of this evaluation indicate that the PTE program of drug testing did not significantly reduce arrests among offenders or increase compliance with conditions of court orders. No differences in any arrests during DTE participation or after the start of DTE, nor in the average number of arrests per month after the start of DTE were found between the participants in the Multnomah County PRSP program and the randomly selected control group of eligible defendants who were not tested. Probation and parole officers who made more extensive use of DTE did not have caseloads with lower rates of arrest, technical violations or total violations than officers who made less use of DTE, controlling for client risk. These results may reflect the lack of sanctioning of positive tests and failure to appear for scheduled tests. Rates of violations were high. Among the pretrial sample, only 14% appeared for all scheduled tests; 46% missed more than half of their scheduled tests. Of those who appeared for at least one test (n=139), 60% tested positive on one or more tests. In the probation and parole sample, 54% had at least one officially recorded violation of their
probation or parole conditions. This group of 504 clients recorded 518 violations: 184 new arrests, 138 technical violations of conditions of release, and 196 other violations (e.g.; failure to benefit from probation and other non-specific reasons) across the 17-month study period. Many,.probation and parole officers, aware of the lack of options for formal sanctions, relied on informal sanctions for DTE violations. These infonnal sanctions included home visits, additional testing requirements, additional reporting requirements, and increased treatment requirements. However, officer willingness to request a revocation hearing and
47
relative toughness in applying infonnal sanctions to OTE violations did not significantly affect the arrest and technical violation rates of their
c~seloads.
Moreover, the interactiOrr"
between OTE utilization and use of sanctions was not significant, providing no support for the thesis that sanctioning preferences of officers combined with use of OTE increased positive outcomes. As in earlier evaluations of drug testing programs, the results are weakened by problems in implementation. These included the lack of sanctions for OTE violations, the limited use of evaluations of client drug problems, limited access to treatment, and the apparent lack of coordination between the pretrial and post-adjudication phases of the program. These findings lead us to echo the c·ail of others for swift and sure sanctions in drug testing programs and expanded access to treatment for offenders with drug problems (Visher 1990). The limitations of this study design must also be considered in interpreting these----./ results. This evaluation shares with others the problem that the outcome variables, detected criminal activity and violations of conditions of release, are imperfect measures of the underlying prevalence of criminal activity and violations in the population of offenders. Because drug testing increases surveillance of participants, DTE may increase the detection of infractions, thereby masking any reductions that might occur. Indeed, the RAND evaluation of intensive supervision probation/parole for drug offenders indicated increased violations and recidivism for offenders subject to additional scrutiny while on release (Turner, Petersilia, and Deschenes 1993). Weaknesses in the data included missing OnNS records on 36% of the orignially selected probation and parole clients, lack of information on which technical
48
violations resulted from positive test results, and lack of infonnation on days spent in jail after start of DTE. These problems allow for the possi~ility that significant impact of testing at probation and parole would be detected with more comprehensive data. Moreover, the cross-sectional design of the probation and parole analysis falls well short of an experimental design. Although efforts were made to control for client and officer characteristics which predict caseload outcomes, it is possible that variables omitted from the model played a significant role in affecting violation rates and might have masked positive effects of DTE. Although the pretrial analysis was based on an experimental comparison of randomly assigned groups, . many pretrial clients failed to appear for their scheduled tests and evaluations, and those that did appear skipped many tests. As a result, exposure to treatment was weak arid it may be that more intensively applied drug testiftg might have had the desired effects. Similarly, the DTE pretrial evaluations of client drug problems did not appear to predict arrest rates, suggesting the need to examine further the validity of the procedures used for identifying treatment needs. On balance, the fmdings echo those of earlier studies and suggest that testing alone is not effective in reducing criminal activity and improving compliance with ·court orders. In the absence of consistent sanctioning for testing violations and access to treatment for clients with severe drug problems, the potential benefits of a systemwide comprehensive drug testing program cannot be said to have been fully tested.
49
REFERENCES .
.
American Correctional AssoCiation. 1981. Drug Abuse Testing: Successful Models For Treatment And Control In Correctional Programs. College Park, MD: American Correctional Association. . Anglin, M. Douglas and Yih-Ing Hser. 1990. "Treatment of Drug Abuse." In Drugs and Crime, edited by M. Tonry and lQ. Wilson (393-460). Chicago: Chicago Press. Ball, J.e., L. Rosen, J.A. Flueck, and D. N. Nurco. 1981. "The Criminality of Heroin Addicts When Addicted and When Off Opiates." In The Drugs-Crime Conneciion, edited by l A. Inciardi. Beverly Hills, CA: Sage Press. Ball, lC., Corty, E., Petroski, S.P., Bond, H., and Tomrnasello, A. 1986. "Medical Services Provided ~ 2394 Patients at Methadone Programs in Three States." Journal of Substance Abuse Treatment, 3: 203-209. Belenko, Steven, Iona Mara-Drita, and Jerome E. McElroy. 1992. "Pre-Arraignment Drug Tests in the Pretrial Release Decision: Predicting Defendant Failure to Appear." NYC .. Criminal Justice Agency Brief Report Series. (December): 1-15. New York: NDR!. oW
Blue Ribbon Commission on Inmate Population Management. 1990. Final report. SacramentC' CA: State of California, Governor's Office. '--" BOTEC Analysis Corporation. 1987. "A Drug Enforcement Program for Santa Cruz County." Report prepared for the Office of the District Attorney and the Santa Cruz Criminal Justice Coordinating Council, Cambridge MA: BOTEC Analysis Corp. BOTEC Analysis Corporation. 1990. "Program Evaluation: Santa Cruz Regional Street Drug Reduction Program." Report prepared for the Office of the District Attorney and the Santa Cruz CririrlnaI Justice Coordinating Council, Cambridge MA: BOTEC Analysis Corp. Brounstein, P., H. Hatty, D. Altshuler, and L. H. Blair. 1989. Patterns of Substance Use and Delinquency Among Inner City Adolescents. Washington, D.C.: The Urban Institute Press:
\.
Bryk, Anthony S. and Stephen W. Raudenbush. 1992. Hierarchical Linear Models: Applications and Data Analysis Methods. Newbury Park, CA: Sage. Carver, 1. A. 1986. "Drugs and Crime: Controlling Use and Reducing Risk Through Testing." NlJ Reports. (September): 3-8.
50
Cavanagh, Da~id. 1993. "A Process Evaluation of the Multnomah County Drug Testing and EvaluatIon Program." Report prepared br the National Institute of Justice. (December), Cambridge MA: BOTEC Analysis Corp. Chaiken, Marcia. 1989. "Prison Programs for Drug-inyolved Offenders." NIJ Research in Action. (October). Washington, DC: U.S. Department of Justice, National Institute of Justice. Chaiken, Marcia. 1986. "Crime Rates and Substance Abuse Among Types of Offenders." In Crime Rates Among Drug-abusing Offenders, edited by B. D. Johnson and E. Wish. New York: Narcotic and Drug Research, Inc . .. Chaiken, J.M., and Chaiken, M.R. 1983. "Crime Rates and the Active Offender." In Crime and Public Policy, edited by 1. Q. Wilson. New Brunswick, NJ: Transaction Books. Chaiken, Jan and Marcia Chaiken. 1982. Varieties of Criminal Behavior. Santa Monica, CA: RAND. Chaiken, Marcia and Bruce D. Johnson .. 1988, "Characteristics of Different Types of Druginvolved Offenders." National Institute of Justice Research Issues and Practices. Washington, D.C.: National Institute of Justice, U.S. Department of Justice.
. Collins, J. J., R. L. Hubbard, and J. V. Rachal. 1985. "Expensive Drug Use and illegal Income: A Test of Explanatory Hypotheses." Criminology 23 (4): 743-764. Dembo, Richard, et al., 1990. The Relationship Between Cocaine Use, Drug Sales and Other Delinquency Among a Cohort of High Risk Youths Over Time. Washington, D.C.: NU and NInA. Falkin, G. P., H. K. Wexler, and D. S. Lipton. 1990. "Drug Treatment in State Prisons." Treating Drug Problems. Vol. 2. edited by D. R. Gernstein and H. 1. Harwood. . Washington, DC: Government Printing Office. Goldkamp, John S. 1989. ·"Pretrial Drug-testing in Six Jurisdictions -- Measuring the Impact of Drug Testing at the Pretrial Stage." The Pretrial Reporter. 13 (1, February). Goldkamp, John·S., Peter R. Jones, and Michael R. Gottfredson. 1990. Pretrial Drug Testing in Milwaukee and Prince George's County. Philadelphia: Crime and Justice Research Institute. Goldstein, Paul. 1985. "The Drug-Violence Nexus: A Tri-Partite Conceptual Framework." Journal of Drug Issues. 15: 493-506.
51
Gottfredson, Michael, Chaster Britt, and John Goldkamp. 1990. Evaluation of Arizona -.....-' Pretrial Services Drug Testing Programs. Tucson: University of Arizona, Department of Management and Policy. Gropper, B. A. 1985. "Probing the Links Between Drugs and Crime." NlJ Research in Brief Washington, DC: U.S. Department of Justice, National Institute of Justice. . Hubbard, Robert L., Mary Ellen Marsden, J. Valley Rachal, Henrick Harwood, Elizabeth Cavanaugh, and Harold M. Ginzburg. 1989. Drug Abuse Treatment: A National Study of Effectiveness. Chapel Hill, NC: The University of North Carolina Press. Inciardi, James A. 1979. "Heroin Use and Street ·Crime." Crime and Delinquency. 25: 335346. Innes, C.A. 1986. Profile of State Prison Inmates: [Special ReponJ. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. Innes, C.A. 1988. "Drug Use and Crime." State Prison Inmate Survey, 1986. [Special ReponJ. Washington, DC: U.S. Department of Justice, Bureau of Iustice Statistics. Johnson, Bruce D., P. J. Goldstein, E. Preble, J. Schmeidler, D. S. Lipton, B. Sprunt, and T. Miller. 1985. Taking Care of Business: The Economics .of Crime by Heroin Users. Lexington, MA: Lexington Books. Kapsch, Stefan and Louis Sweeny. 1990. Multnomah County DMDA program Evaluation Final Report. Portland, OR: Reed College Public Policy Workshop. Kushner, Jeffrey. 1993. "Salient and Consistent Sanctions: Oregon's Key to Reducing Drug Use." Treatment Improvement Exchange Communique. Center for Substance Abuse Treatment (Spring). McGlothlin, W. H., M. D. Anglin, and B. D. Wilson. 1977. An Evaluation of the California Civil Addict Program. Washington, D.C.: Government Printing Office. National Institute on Drug Abuse. 1988. Annual Data: 1987: Data from the Drug Abuse Warning Network (DAWN). Rockville, MD: NIDA. National Institute of Justice. 1990. Drugs and Crime 1989: Drug Use Forecasting Annual Repon. Washington, D.C.: U.S. Department of Justice. Nurco, D. N., T. E. Hanlan, and T. W. Kinlock. 1990. Offenders, Drugs, Crime, and Treatment. Draft. Washington, D.C.: Bureau of Justice Assistance.
52
Rolph, John E. and Jan M. Chaiken. 1987. Identifying High-Rate Serious Criminals from Official Records. Santa Monica, CA: RAND. ,
,
Smith, Douglas A., and Christina Polsenberg. 1992. "Specifying the Relationship Between Arrestee Drug Test Results and Recidivism." The Journal of Criminal Law and Criminology. 83 (2): 364-377 Speckart, George and M. Douglas Anglin. 1986a. "Narcotics and Crime: A Causal modeling Approach." Journal of Quantitative Criminology 2: 3-28. Speckart, George and M. Douglas Anglin. 1986b. "Narcotics Use and Crime: An Overview of Recent Research Advances." Contemporary Drug Problems 13: 741-769. Toborg, Mary A., John P. Bellassai, Anthony M. J. Yezer, and Robert B. Trost. 1989. Assessment of Pretrial Urine Testing in the District of Columbia--Summary Report, Final Report of NIJ Grant No. 83-IJ-CX;.K046, Washington, D.C.: National Institute of Justice. , ' Toborg, Mary A.f John P. Bellassai, and Anthony M. J. Yezer. 1986. "A Summary of Interim Findings." The ,Washington, D.C. Urine,Testing Program for Arrestees and Defendants Awaiting Trial. Washington, D.C.: Pretrial Services Agency. (June). Turner, Susan, Joan Petersilia, and Elizabeth P. Deschenes. 1992. "Evaluating Intensive Supervision ProbationlParole (lSP) for Drug Offenders." Crime and Delinquency. 38 (4, October): 539-556. Turner, Susan, Joan Petersilia, and Elizabeth P. Deschenes. 1991. "The Implementation and Effectiveness of Drug Testing in Community Supervision: Results of an Experimental Evaluation." Paper submitted to Doris L. MacKenzie and Craig Uchida (eds.), Drugs and the Criminal Justice System: Evaluating Public Policy Initiatives. Newbury Park, CA: Sage Publications. Visher, Christy. 1992. "Pretrial Drug Testing." National Institute of Justice Research In Brief. (September): '1-8, Washington, DC: U.S. Department of Justice, National Institute of Justice. Visher, Christy A 1990. "Incorporating Drug Treatment in Criminal Sanctions." NIJ Reports. (June): 2-7, Washington, DC: U.S. Department of Justice, National Institute of Justice. Wexler, H.K., Lipton, D.S., and Johnson, B.D. 1988. A Criminal Justice System Strategy for Treating Cocaine-Heroin Abusing Offenders in Custody (Document No. NCJ 108560). Washington, DC: U.S. Department of Justice, National Institute of Justice.
53
Wish, Eric D., Mary A. Toborg, and John P. Bellassai. 1987. "Identifying Drug Users and Monitoring Them During Conditional Release." NlJ Briefing Paper. Washington, D ~ U.S . Department of Justice, National Institute of Justice.
,.
54
·. APPENDIX A RESULTS OF THE SURVEY OF MULrnOMAH COUNTY CORRECITONS OFFICERS
,.
55
Table A-I.
Officer DTE Use by Sanctioning for Positive Tests or Failure to Appear for Test Higher (n=15)
Medium (n=35)
Lower (n=33)
Total (n=83)
Tough
47%
43%
38%
42%
Moderate
20%
23%
47%
32%
Easy
33%
34%
16%
27%
=
1 Tough officers usually or sometimes recommend revocation for a DTE client who failed to show up for a drug test once or twice; easy officers = sometimes or never recommended revocation for more than 2 dirty tests; moderate others.
=
1
J'
56
----------........... Table A-2.
Actions Probation and Parole Officers Would Take For One or Two No Shows
Percent of Officers Who Said: 41%
Formal Sanctions Recommends Revocation Formal Sanctions
32%
Informal Sanctions New Counseling or Treatment More Intensive Treatment More Tests Home Visit Extra Requirements Other
26% 7% 12% 4% 44% 4%
No Sanctions Waming/Reprimand Nothing
47% 4%
For More Than Two No Shows Formal Sanctions Recommends Revocation Formal Sanctions
77% 59%
Informal Sanctions New Counseling or Treatment More Intensive Treatment More Tests Home Visit Extra Requirements Other
27%
10% 11%
5% 38% 10%
No Sanctions '" Waming/Reprimand Nothing
31% 1%
57
Table A-2. Actions Probation and Parole Officers Would Take Table A-2, Cont.
For 1-2 Positive Tests Formal Sanctions Recommends Revocation Other Fonnal Sanctions
88% 59%
Infonnal Sanctions New Counseling or Treatment More Intensive Treatment More Tests Home Visit Extra Requirements Other
47% 23% 4% 4% 30% 11%
No Sanctions . . Waming/Reprimand Nothing
13% 1%
For More Than Two Positive Tests Formal Sanctions Recommends Revocation Other Fonnal Sanctions
99% 78%
Infonnal Sanctions New Counseling or Treatment More Intensive Treatment More Tests Home Visit Extra ~equirements Other
42% 33% 5% 2% 19% 7%
No Sanctions Waming/Reprimand Nothing
2%
1%
'".
58
'--""
