Go to Table of Contents Acrobat Format


Final Report

 

Evaluation of An Individualized Sanctioning

Program For DWI Offenders

 

Ralph K. Jones

John H. Lacey

 

December 1998

Prepared for:

U.S. Department of Transportation

National Highway Traffic Safety Administration

Washington, DC 20590

Contract Number DTNH22- 94-C-05124

Mid-America Research Institute, Inc. of New England

Winchester, Massachusetts


Back to Table of Contents

 

  1. Report No 

  2. Government Accession No. 

  3. Recipient's Catalog No.

  4. Title and Subtitle

Evaluation of an Individualized Sanctioning Program for DWI Offenders

  5. Report Date
  December 1998

  6. Performing Organization Code

  7. Author(s)

  Jones, R. K.; and Lacey, J. H.

   8. Performing Organization Report No.

 

  9. Performing Organization Name and Address

  Mid-America Research Institute
  611 Main Street
  Winchester, MA 01890
  10. Work Unit No. (TRAIS)

 


  11. Contract or Grant No.
DTNH22-94-C-05124

12. Sponsoring Agency Name and Address

  National Highway Traffic Safety Administration
  Office of Research and Traffic Records
  400 7th Street, S.W.
  Washington, DC 20590

 

  13. Type of Report and Period Covered

Final Report

 

  14. Sponsoring Agency Code

  15. Supplementary Notes

Amy Berning was the Contracting Officer’s Technical Representative (COTR) for this project.

  16. Abstract

Evaluates two different approaches to sentencing DWI offenders in Georgia. One approach (called the "Todd Program" after the judge who developed it) imposed individually-tailored combinations of traditional and alternative sanctions, and the other approach generally imposed only the minimum sanctions required by state law. The measure of effectiveness was the statewide DWI recidivism of the offenders studied. The evaluation found that this program was more effective by a wide margin than the sentencing program that imposed the minimum sanctions. It was concluded that the Todd Program approach to DWI sentencing could be used in other jurisdictions.  

  17. Key Words

DWI, impaired drivers, sanctions, sentencing, recidivism, court procedures, program evaluation

  18. Distribution Statement

This report is available from the National Technical Information Service, Springfield, Virginia 22161

  (703) 487-4650

   19. Security Classif. (of this report)

   20. Security Classif. (of this page)

   21. No. of Pages

   22. Price

Form DOT F 1700.7 (8-72) Reproduction of completed page authorized

 


Back to Table of Contents

ACKNOWLEDGMENTS

 

 

     The authors are grateful to the many individuals who helped with this project. We would especially like to thank Judge William F. Todd, Jr. for his willingness to share information about his program and to subject it to an objective evaluation.
     Penny Thompson and Susan King of his staff developed and maintained the sentencing data base Judge Todd uses and were extremely helpful in providing us with those data and in helping us to interpret them. Peggy Reigner was most willing to provide computerized data bases from the Court’s central computer system.
     Cheryl Fisher Custer, District Attorney, John E. Manget, Probation Officer and SFC Cliff Miller of the Georgia State Patrol were most helpful in developing a description of the DWI control system in Rockdale County.
     At the state level, Lt. Sammy L. Hill and Deborah Durham guided us through the process of obtaining access to driver records and matching them to our data base and Thomas M. Dellinger coordinated searching the statewide driver history file and providing us with an analyzable data base which could be used to perform the recidivism analyses.
     We thank all who assisted in the planning, implementation, interview process and analyses of data in connection with this study.

 


Back to Top

TABLE OF CONTENTS

ACKNOWLEDGMENTSi
TABLE OF CONTENTSiii
TECHNICAL DOCUMENTATION PAGE
FIGURESiv
TABLESiv
EXECUTIVE SUMMARYv
1 - INTRODUCTION 1
      BACKGROUND 1
      PROJECT SCOPE AND APPROACH 2
      ORGANIZATION OF THE REPORTS 3
2 - PROGRAM DESCRIPTION AND ENVIRONMENT 5
      DESCRIPTION 5
      ENVIRONMENT 6
             Population and Socio-Economic Factors 6
             DWI Enforcement 6
             Adjudication of DWI Cases 7
3 - PROGRAM EVALUATION 9
      APPROACH 9
       PROGRAM IMPACT 9
            Experimental Design 9
             Data Sources 10
             Analysis Techniques 11
              Results 12
       PROGRAM ACTIVITY AND SENTENCING POLICY 14  
              Program Activity 14
              Sentencing Policy 14
       SUMMARY AND CONCLUSIONS 17
4 - CONCLUSIONS 21
REFERENCES 23

 


Back to Table of Contents


Back to Table of Contents

FIGURES

 

Figure 3-1: Modeled Recidivism of Test Group Drivers and Comparison Group Drivers With No Prior DWIs 13
Figure 3-2: Effect of Number of Prior DWIs on the One-Year Recidivism Rate of the Test Group and the Comparison Group 13

 

 

TABLES

Table 3-1: Variables In Judge Todd’s Data Base 10
Table 3-2: Selected Test Group Characteristics, n=86912
Table 3-3: Percentage of Test Group Receiving Various Sentence Components (n=869)14
Table 3-4: Effect of Offender Attributes on Jail and House Arrest Sentences15
Table 3-5: Effect of Offender Attributes on Alcoholics Anonymous, Daily Breath Test, and Electronic Monitoring Sentences16
Table 3-6: Effect of Offender Attributes on the Severity of Two Sentence Packages Used by Judge Todd17
Table 3-7: Offender Characteristics That Were Significantly Related to One or More Sentence Components or Sentencing Packages18

 


Back to Table of Contents

EXECUTIVE SUMMARY

 

Legal sanctions such as jail, fines, suspension or revocation of the driver’s license are the ultimate response of traffic law systems to alcohol-impaired driving. In recent years, more attention has been paid to so-called alternative sanctions, including referral of drivers to treatment and education, community service in lieu of or in addition to jail, electronic monitoring, intensive supervision probation, impoundment or forfeiture of vehicles or license plates, victim restitution, visits to a hospital emergency room that treats traffic crash victims, ignition interlocks, and using license plates that identify the vehicle owner as a driving-while-intoxicated (DWI) offender, among other sanctions.
     This report documents the results of an evaluation of a sanctioning program that used a combination of traditional and alternative sanctions packaged so as to meet the needs of individual offenders, be they repeat offenders or first offenders. We refer to this program as the Todd Program because it was conceived and developed by Judge William F. Todd, Jr. of the State Court of Rockdale County, Georgia. The program has been operated since its inception by Judge Todd and his staff.
     The evaluation sought to determine the effect of the Todd Program on the future drinking-driving behavior of program participants. The measure used for future drinking-driving behavior was the time from a subject’s index DWI offense to another DWI offense, that is, DWI recidivism. We compared the DWI recidivism of the Todd Program participants with that of offenders in a nearby, similar jurisdiction that did not employ a tailored sanctioning approach, but generally imposed the minimum sanctions required by Georgia law.
     We found that the Todd Program was more effective by a wide margin than was the sentencing program that imposed the minimum sanctions -- offenders sentenced in the Todd court had a recidivism rate that was about one-half that of offenders sentenced in the comparison court.
     A wide range of sanction choices was available to the Todd Program. Those that required abstinence and incorporated procedures to ensure abstinence were especially popular. A common factor among sanctions not involving jail was the close contact kept between the judge and the offender during the probationary period. The need for close and extended contact is consistent with the findings of recent NHTSA-sponsored research (Jones, Wiliszowski, and Lacey, 1997; Wiliszowski et al., 1996; DeYoung, 1997) which suggests that such contact with sanctioning agencies is very important to preventing future occurrences of DWI.
     The Todd Program used a number of factors in determining which sanctions to incorporate into an offender’s sentencing package. Factors that appear to have been weighted the highest in selecting sanctions were number of prior DWI convictions, refusal to submit to a breath alcohol test, BAC (if tested), offender age and sex, circumstances surrounding the DWI incident, the offender’s demeanor and appearance in court, and the judge’s personal knowledge of the individual being sentenced.
     We note that the concept of individualized sanctions is not new, nor is the practice of the concept unique to Judge Todd’s court. Criminal justice theorists and judicial educators alike have long espoused the use of sanctions tailored to the needs of individual offenders. However, our experience suggests that the serious practice of individualized sanctioning is less widespread than many supporters of the concept would like to see. In a recent re-examination of the DWI enforcement system (Jones, Lacey, and Wiliszowski, 1998), failure to impose appropriate sanctions was identified as a significant failure of DWI enforcement systems. Major factors contributing directly to this failure were lack of information available to judges, lack of sanctioning resources, lack of uniformity in sentencing, and insufficient attention to sentencing (pg. 63), and volume of cases.
     We have found no evidence in the research literature of the effectiveness (or lack of effectiveness) of individualized sentencing on the future drinking-driving behavior of DWI offenders. This study represents a first step toward filling this gap.
     We concluded that the Todd Program approach to DWI sentencing could be used in other jurisdictions with enough judicial interest in the drinking-driving problem to develop and operate such a program. Jurisdictions lacking sufficient sanctioning resources might be able to acquire them, as has Judge Todd, through "self-sufficiency" programs such as assessing fees on DWI offenders for the services rendered.

Back to Table of Contents

1 - INTRODUCTION

BACKGROUND
     For a number of years the National Highway Traffic Safety Administration (NHTSA) has been examining the effectiveness of sanctions imposed on drivers convicted of driving while intoxicated (DWI) by judicial and administrative agencies of state and local government. Traditionally, DWI sanctions have been in the form of a fine, incarceration, or a suspension (or revocation) of the driver’s license. In recent years, more attention has been paid to so-called alternative sanctions for drunk driving. Most commonly, these have involved referral of drivers to treatment and education, and such referrals have now become "legitimized" by statutes in many states. Other alternative or non-traditional sanctions that have been tried for DWI (and also legitimized in some instances) include community service in lieu of or in addition to jail, victim restitution, visits to a hospital emergency room that treats traffic crash victims, ignition interlocks, and using license plates that identify the vehicle owner as a DWI offender.
     In recent years, NHTSA has been conducting a program of research aimed at identifying alternative sanctioning programs for DWI and assessing their effective-ness in reducing the incidence of DWI. Much of this effort has focused on repeat offenders. A recent study (Jones, Lacey, and Wiliszowski, 1997) found that two programs, one involving intensive supervision of probation and the other employing electronic monitoring, were highly effective in reducing recidivism among repeat DWI offenders. DeYoung (1997) found that impounding vehicles driven by suspended / revoked and unlicensed drivers in California had a positive impact on several measures of recidivism of the affected drivers. Another study (Wiliszow-ski et al., 1996) examined why repeat DWI offenders continue to drink and drive. To reduce recidivism, Wiliszow-ski et al. recommended consideration of sanctions involving personalized assessments and reassessments, individualized treatment regimens, intensive supervision probation, and treatment during confinement for incarcerated offenders.
     This report documents the results of an evaluation of a sanctioning program that used a combination of traditional and alternative sanctions packaged so as to meet the needs of individual offenders, be they repeat offenders or first offenders. Thus, a repeat offender with many priors and clear evidence of a drinking problem might receive a sentence requiring a relatively long jail term, a heavy fine, participation in a treatment program, daily breath-alcohol testing, and close supervision by a probation officer. Another offender with no priors and no evidence of a drinking problem might receive a much lighter and less restrictive sentence consisting of two days in jail, a light fine, and attendance in a DWI school.
     We refer to this program as the Todd Program because it was conceived and developed by Judge William F. Todd, Jr. of the State Court of Rockdale County, Georgia. This court is not limited to traffic offenses, but has jurisdiction over other classes of misdemeanors as well. (All DWI offenses not involving a fatality are classified as misdemeanors.) The program has been operated since its inception by Judge Todd and his staff.
     We note that the concept of individualized sanctions is not new, nor is the practice of the concept unique to Judge Todd’s court. Criminal justice theorists and judicial educators alike have long espoused the use of sanctions tailored to the needs of individual offenders. However, our experience suggests that the serious practice of individualized sanctioning is less widespread than many supporters of the concept would like to see. In a recent re-examination of the DWI enforcement system (Jones, Lacey, and Wiliszowski, 1998), failure to impose appropriate sanctions was identified as a significant failure of DWI enforcement systems. Major factors contributing directly to this failure were lack of information available to judges, lack of sanctioning resources, lack of uniformity in sentencing, and insufficient attention to sentencing (pg. 63), and volume of cases.
     We have found no evidence in the research literature of the effectiveness (or lack of effectiveness) of individualized sentencing on the future drinking-driving behavior of DWI offenders. This study represents a first step toward filling this gap.

Back to Table of Contents

PROJECT SCOPE AND APPROACH
     This research project involved the description and evaluation of the Todd Program. We also described briefly the environment in which the program operated, including an overview of DWI enforcement and adjudication procedures in Rockdale County. The descriptive material was developed through interviews with Judge Todd and his staff, police officers, prosecutors, jail staff, probation officers, and treatment providers.
     The evaluation sought to determine the effect of the Todd Program on the future drinking-driving behavior of program participants. The measure used for future drinking-driving behavior was the time from a subject’s index DWI offense to another DWI offense, that is, DWI recidivism. We compared the DWI recidivism of the Todd Program participants with that of offenders in a nearby, similar jurisdiction that did not employ a tailored sanctioning approach, but as a rule imposed only the minimum sanctions required by Georgia law. We also obtained data on gross measures of the performance of the Todd Program, and examined how sentence packages were determined for various types of offenders.
     Quantitative data for the evaluation came from two sources, driver records files from the- Georgia Department of Public Safety and a detailed data base maintained by Judge Todd and his staff. Judge Todd’s data base contained descriptive data and sentencing data for each adult DWI offender sentenced in his court.

Back to Table of Contents

ORGANIZATION OF THE REPORT
     The substance of the report is contained in the following three chapters. Chapter 2 contains the description of the Todd Program and its operating environment. The evaluation and its results are presented in Chapter 3, and the project’s conclusions and recommendations are given in Chapter 4.

 


Back to Table of Contents

2 - PROGRAM DESCRIPTION AND ENVIRONMENT

 

 

Back to Table of Contents

DESCRIPTION
     Judge Todd’s program was implemented in 1992. His sentencing approach is characterized by the use of a wide variety of punitive, rehabilitative, and treatment sanctions offered in packages that are carefully tailored to each offender.
     The Todd Program differs in two ways from other sanctioning programs that include several components. First, the range of sentencing options is much wider than that available in most jurisdictions (especially jurisdictions of moderate size such as Rockdale County). In addition to the traditional sanctions of fines and jail time, the judge may include house arrest (with or without electronic monitoring), intensive supervision probation, frequent breath-alcohol testing, work release, and participation in a wide variety of treatment programs that include Alcoholics Anonymous (AA) and in-patient treatment for up to four months. Pictures of convicted DWI offenders are published in the newspaper along with a description of their sentences. All DWI offenders must attend a victim’s impact panel, and all must serve some jail time.
     A second feature of the Todd approach is that the judge does his own present-ence investigation using a data base that he developed and maintains with the support of his staff. The tailoring of sanctions is based on information elicited by the judge from the offender during the sentencing process, and on criminal records and driver records available to the judge during sentencing. At this writing, the judge’s data base contains records on some 1,800 offenders. As a part of our evaluation, we queried Judge Todd and examined his data base to learn more about factors that he takes into consideration when arriving at a sentence package for an offender. The results of this analysis are discussed in Chapter 3 beginning on page 14.
Most of Judge Todd’s non-traditional sanctions are imposed as a condition of probation. Todd relies on a private probation company (paid by offender fees) to supervise probation to ensure that the required conditions are met. The probation agency reports probation violations to Judge Todd, and he takes action on their reports. Serious, non-technical violations of probation conditions (such as failure of a breath-alcohol test) result in incarceration. Offenders who are required to submit to periodic breath-alcohol tests may do so though the use of electronic monitoring devices if they are willing to pay the required electronic monitoring fee of $7.50 per day. Otherwise, no fee is assessed, but the offenders may be required to report to the probation office, the work release center, or the Sheriff’s office for testing, in some instances as often as twice daily.
     All of this is accomplished in a setting in which few offenders (estimated at less than five percent) elect to go to trial. In the vast majority of cases, the sentences are the result of a plea bargain.

Back to Table of Contents

ENVIRONMENT
Population and Socio-Economic Factors
The Todd Program is operated in Rockdale County, Georgia. Rockdale, the county with the smallest area in Georgia, is southeast of Atlanta and encompasses small urban, suburban and rural areas. The county seat is in Conyers, Georgia. According to the Bureau of the Census, the population of Rockdale County has been increasing steadily from 36,600 in 1980 to a 1995 population of 64,500. In 1990, roughly 64% of the 54,100 individuals residing in Rockdale County were between the ages of 18 and 64, 28% were under age 18, and 8% were 65 or older. Ninety percent of the population in 1990 were white, 8% were black, 1% Asian or Pacific Islander, and 1% were other races. Per capita personal income for the County in 1993 was $19,267. The 1994 unemployment rate was 3.7%, lower than the Georgia state unemployment rate of 5.2% for that year.

Back to Table of Contents

     DWI Enforcement
      DWI laws adjudicated in Todd’s court are enforced by Conyers City Police, the Rockdale County Sheriff’s Department, and the Georgia State Patrol. DWI "road checks" are conducted every holiday period, and "concentrated patrols" are carried out once a month. Conyers City Police have no DWI task force, but a "special operations" group is planned which will include DWI enforcement. Judge Todd urges police officers to obtain as much information as possible from citizens reporting DWI suspects, encourages the citizens reporting the incident to follow the DWI suspect (if possible to do so safely), and encourages the police to obtain the reporter’s name and address.
     Many patrol cars are equipped with video cameras, and officers in those cars videotape DWI suspects beginning immediately upon suspicion that the driver is DWI. After stopping a DWI suspect, the officer asks the driver to take a Standard-ized Field Sobriety Test (SFST). Passive breath-alcohol sensors are not used in determining whether the driver should be arrested for DWI. Arrests are made at the scene before transporting the suspect.
     The officer stays with the vehicle until it is impounded or released to another person; the procedure depends on the location of the stop. A vehicle release form is filled out. Impaired passengers are driven home or released to a sober person.
     The suspect is transported to the jail for a breath test or to the hospital for a blood test. If certified, the arresting officer will administer the breath test; otherwise another officer who ois certified will conduct the test. A mandatory 20 minute waiting period from time of initial personal contact is required before an evidentiary breath test is taken. After the test, individuals are released to the jailer. If the suspect’s blood alcohol concentration (BAC) is much lower than the legal limit (0.10%) and the officer sees impairment inconsistent with the BAC reading, the officer will request a blood or urine test for drugs other than alcohol. Individuals are held until their BAC is 0.05 or less. Adults who have a BAC of 0.08 or more are held for 24 hours. Juveniles (suspects age 16 or less) are released to their parents or to a juvenile probation officer.

Back to Table of Contents

     Adjudication of DWI Cases
     DWI tickets are brought to the prosecuting clerk’s office. DWI defendants do not get a chance to come to court on just the ticket; prosecutors have to file the case first. This allows prosecutors time to prepare the case before the speedy trial clock starts. The laws regarding speedy trials require a case to be tried within the remainder of the current court session or by the end of the next court session. Each court session is three months long, so a case going to trial will be completed within a maximum of 180 days. A Criminal Case Management System has been in place since 1987. This system tracks each case through the entire criminal justice system including jail.
     Rockdale County Court is classified as a "state court" with jurisdiction over misdemeanors. All DWIs not involving a fatality are classified as misdemeanors. DWI cases are limited to those in which the person charged is an adult. One judge (presently, Judge Todd) serves the State Court of Rockdale County.
     The prosecution receives the ticket and breath alcohol testing (BAT) tape, runs a criminal history and driver history, and then starts filling out the forms required for case filing and prosecution. Driver history data appear to go back to 1976. Priors are recorded and classified, for example, first in five years and second in life. Investigators compile a witness list, make sure all the necessary information is in the case folder, and handle any investigative work needed. They also prepare a summary of the facts of the case and the charge. The file includes the police officer’s incident report (IR), any refusal information, car impound form and bond sheet. The formal document filed is called an "accusation," analogous to a "criminal complaint." The Uniform Traffic Citation (UTC) can also be ratified and filed as an accusation. The package is then sent to the responsible prosecutor. Prosecutors meet once a month to discuss the accusations that are then batched and sent to the court clerk or else released (not prosecuted).
     Refusals can be used as evidence in a DWI trial. When building a case, the prosecutors believe it best to rely on the officer’s observation of the suspect’s behavior and not entirely on a BAT. They believe that officers should observe every SFST as if there were not going to be a BAT, and they should document observa-tions. This is because, if the information on SFST results is not available in the paperwork submitted for discovery, it may not be admitted into evidence. Also, a recorded narrative of the suspect’s behavior should accompany the videotape, since the camera is fixed in place and cannot be manipulated to record subtle actions by the suspect.
?
Back to Table of Contents

3 - PROGRAM EVALUATION

APPROACH
     The evaluation concentrated on the effect of the Todd Program on the future drinking-driving behavior of program participants, that is, program impact. The measure used for future drinking-driving behavior was the time from a subject’s index DWI offense (as defined in Georgia Statutes Annotated) to another DWI offense.
     Other areas of concern were the level and nature of activity involved in the program. This aspect of an evaluation is often referred to as a process evaluation. Our approach was to obtain data on gross measures of program activity, and we also queried program staff on the particulars of their activities, especially those related to the determination of sentence packages for offenders.
     Thus, our evaluation had two components, program impact and program operation. Each of these components is described below.
? ? Back to Table of Contents
PROGRAM IMPACT
Experimental Design
The major research question addressed was:
What is the DWI recidivism of offenders participating in the individu-alized sanctions program and how does it compare with the DWI recidivism of offenders given traditional sentences--?
     DWI recidivism is defined here as the probability of the occurrence of a subsequent DWI on or before time T. The statewide recidivism of the offenders receiving the individualized sanction (the test group) was compared to that of a comparison group composed of DWI offenders in a nearby county that used a more traditional approach to sentencing. The County State Court judge in the comparison county us-ed a sentencing policy based on the minimum sentences set forth in pertinent Georgia statutes, and did not attempt to tailor his sanctions to individual offenders as did Judge Todd.
     In the comparison county, misdemeanor DWI cases other than those requiring a jury trial are heard in Probate Court. Although there is no absolute standard sentence for any case, the customary sentence for convicted DWI offenders follows the general pattern indicated below:
  First Offense - $465 fine and surcharge, one day in jail, 40-50 hours of community service and write a 500 word essay on "Why I Shouldn’t Drink and Drive."
  Second Offense - $840 fine and surcharge, 2 days in jail and 80-100 hours of community service.
  Third Offense - $2,490 fine and surcharge, 10-30 days in jail and 200-250 hours of community service.
     Although these sanctions are not absolutely the legislatively mandated minimum sanctions for DWI, they approximate them closely. Throughout this report we have for convenience referred to the above pattern of sentencing as the mandatory minimum sanctions.
     The comparison county is west of Rockdale County. The two counties are quite similar in many respects, most important to this study being their approach to and performance in enforcing DWI laws. Both had very similar DWI arrest rates during the study period (8.5 per 1,000 population in Rockdale County versus 8.9 per 1,000 popula-tion for the comparison county).
     Since subjects were not randomly assigned to the two groups, analytic adjustments of the data were made to account for possible differences between the test group and the comparison group. Covariates available for adjusting the modeled recidivism were offender age at the time of the index violation, offender sex, and number of prior DWI offenses at the time of the index violation.
Back to Table of Contents
    Data Sources
     Dates of alcohol-related offenses for the test group and the comparison group were obtained from the- Georgia Department of Public Safety (DPS). The data contained variables indicating the dates of each subject’s DWI offenses. Subject date of birth and sex were also contained in the DPS data, but were needed only for the com-parison group.
     Judge Todd maintains a detailed data base containing descriptive data and sentencing data for each DWI offender sentenced in his court. Pertinent variables from that file are listed in Table 3-1. Test group subjects were defined as all subjects in the Todd file whose violation occurred after December 31, 1992 and whose driving record could be matched in the DPS file. Based on Todd’s database, we estimate that an average of about 360 individuals are adjudicated, found guilty and sentenced for DWI each year in Rockdale County.
     No such file on DWI offenders exists in the comparison county. Instead, comparison group subjects were selected from the DPS records of all persons who had been convicted of an alcohol-related traffic offense in the comparison county court. Only those offenders who had a violation after December 31, 1992 were chosen for the comparison group. This was to ensure that both the test group and the comparison group were driving in roughly the same statutory and law-enforcement environment, and were living in a comparable socio-economic environment.
Back to Table of Contents
     Analysis Techniques
     The primary technique used for the impact analysis was sur-vival curve analysis. This technique allows the study of complex time patterns of recidivism, for example, a recidivism rate that is initially high, but lower later.
     The formal factor reflecting the evaluation design was a variable indicating whether the subject belonged to the test group or the comparison group. As indicated above, -prior alcohol-related driving offenses, age, and sex were used to control for differences in composition of the test and comparison groups.
     In the survival analysis, we used the time from the index offense to the first "failure" (that is, another DWI offense) as the dependent variable. The index offense for each test group subject was taken directly from the Todd file (and was also matched exactly by an offense in the subject’s DPS record). The index date for each comparison group subject was defined as that subject’s first DWI offense occurring after 1992. Offenders in both counties were track-ed from their index offense to their first subsequent alcohol-related traffic offense or until December 31, 1997, whichever occurred first.
     The time-varying recidivism (that is, probability of a failure) as a function of group (test or comparison) was of primary interest in the recidivism analysis. The analysis was designed to indicate wheth-er there was any difference in recidivism with respect to group and if so, the direction and amount of the difference, and also the probability p that the difference was due to chance alone. We were also interested in whether various subgroups (for example, young offenders and old offenders, male offenders and female offenders) had differing recidivism times with respect to group. The statis-ti-cal techniques contained in the SAS� PHREG procedure were used in the analysis.
Back to Table of Contents
     Results
     A total of 869 of the test group subjects were listed in the DPS driver records file. Some of their characteristics are shown in Table 3-2. The comparison group contained 637 subjects and differed significantly from the test group with respect to sex and number of priors, indicating that these group differences would have to be taken into account in the analyses. Compared to the comparison group, the test group had a larger percentage of females and a smaller percentage of drivers with one or more prior DWIs.
     The recidivism analysis indicates a signifi-cantly lower recidivism rate for the test group than that of the comparison group (p<0.0001)1. The covariates age and sex had no significant effect on recidivism, but number of priors did have a significant effect (p=0.0024). The effect of group (test or comparison) is illus-trated in Figure 3-1 which shows the percent of drivers with no prior DWIs recidivat-ing at various times after the index conviction. After one year, about 6.0% of the Todd group had committed another DWI offense, compared to 11.1% of the comparison group. After four years, these percentages increased to 13.8% and 24.7%, respectively. A similar effect was noted for drivers with various numbers of prior offenses (Figure 3-2). At any given time after conviction, the recidivism rate of each group increased about 8% per prior, and the recidi-vism rate of the Todd group was only about 54% that of the comparison group.

Back to Table of Contents

 

 

 

?

Back to Table of Contents

  PROGRAM ACTIVITY AND SENTENCING POLICY

     Program Activity
     Descriptions of the sanctioning process and the various components of the sentenceing packages (for example, house arrest with electronic monitoring) were presented in Chap-ter 2. Table 3-3 shows how often various sentence components were imposed on the 869 members of the test group. All offenders received some jail, with about 62% receiving 10 days or fewer. Jail terms exceeding 30 days were relatively rare, occurring for only about 18% of the offenders. Nearly one-half (44%) of the offenders were required to attend AA meetings. About 20% were placed on house arrest, with about half of these being subjected to electronic monitoring. (All of the 11% of the offenders given electronic monitoring were also given house arrest.) Periodic breath alcohol tests were also quite common, being required in 48% of the cases.

Back to Table of Contents

    Sentencing Policy
     Our examination of Judge Todd’s sentencing policy attempted to determine which if any offender characteristics could be associated with given sentences. In the first series of analyses, we studied each major sentence component separately as a function of offender characteristics. In the second series of analyses, we looked for combinations of sentence components that were associated with certain offender characteristics.
     Sentence components. Two sentence components, jail time and house arrest time, had continuous data in units of days. We used the SAS� GLM procedure to analyze these components as dependent variables. Independent variables were the following: age (< 35 or 35+), sex (male or female), race (white or non-white), marital status (married or not married), children (yes or no), employed (yes or no), prior DWIs (yes or no), breath test refusal (yes or no), and BAC (>0.15 or <0.15). The break point for age was chosen as 35, the mean age of the test group. The results are shown in Table 3-4.
     Age, sex, marital status, priors, refused, and BAC were statistically significant (the column labeled "p") predictors at the 0.05 level of number days of jail imposed on offenders, with priors having the largest effect. Older offenders got 14 more days of jail than younger offenders; male offenders got 11 more days than females; married offenders got 11 fewer days than unmarried offenders; offenders with priors got 59 more days than offenders without priors; test refusers got 13 more days than non-refusers; higher BAC offenders got 12 more days than lower BAC offenders.
     By contrast, only the variables sex and priors had a significant effect on number of days of house arrest, with male offenders getting about 12 fewer days of house arrest than female offenders, and offenders with priors getting about 56 more days of house arrest than offenders without priors.
     Some sentence components were presented only as categorical variables in the Todd data base and could not be analyzed properly by the GLM procedure. These components were: Alcoholics Anonymous (yes or no), daily breath tests (yes or no), and electronic monitoring (yes or no). We transformed these categorical variables into 0-1 variables and used logistic regression (the SAS� LOGISTIC procedure) as the analysis technique. With the exception of BAC, the dependent variables were the same as those used in the GLM analysis. For BAC, we used a 0-1 variable, with a value of 1 indicating a BAC > 0.15 and a value of zero indicating a BAC < 0.15.
     The results of this analysis are shown in Table 3-5 below. In this table, the results are presented in terms of the "odds ratio" and the associated p-level of the odds ratio. The term "odds" is defined as the probability that a given sentence was imposed divided by the probability that the sentence was not imposed. The odds ratio is the ratio of the odds that an offender with a given characteristic received a given sentence component to the odds that an offender without that characteristic received that sentence component. For example, the odds that an offender with priors was sentenced to electronic monitoring are 12.58 times the odds that an offender without priors was sentenced to electronic monitoring. An odds ratio greater than one for a sentence component indicates that the presence of a characteristic (such as having one or more priors) increases the chance of an offender’s receiving that sentence component.
     From the table, it is seen that several offender characteristics had a statistically significant effect (usually taken as 0.05) on the three sentence components. Priors had a strong effect on all components with odds ratios in the 13 to 35 range. The variable sex had a significant effect for Alcoholics Anonymous and breath test, and a marginally significant effect on electronic monitoring. Males were more likely to receive AA and periodic breath tests, but less likely to receive electronic monitoring. Refusing a breath test for the index offense had a significant effect only for receiving the breath test component, with refusers being more likely to be required to undergo breath testing as a condition of probation. Note that the table shows only main effects and not interaction effects, e.g., the odds ratios for males vs. priors and the odds ratios for females vs. priors.
     Sentencing Packages. The analysis of -sentencing packages used factor analysis techniques to see whet-her Todd’s data would produce any factors composed of heavily loaded linear combinations of sentence variables. The sentence variables analyzed were jail time (>30 days or <30 days), work release (yes or no), Alcoholics Anonymous (yes or no), periodic breath tests (yes or no), electronic monitoring (yes or no), and house arrest (yes or no). The statistical techniques contained in the SAS� FACTOR procedure were used in the factor analysis.
     Two factors emerged from the factor analysis. One factor had high loadings on jail, Alcoholics Anonymous, work release, -and breath test. We labeled this factor "Traditional." The second factor had high loadings on house arrest and electronic monitoring and was labeled "Restrictive." A recidivism analysis was conducted using these two factors as independent variables. The analysis showed that the factor Traditional was significantly related to recidivism (p=0.002), and the factor Restrictive was non-significantly related to recidivism (p=0.291). Higher values of Traditional were associated with increased recidivism. This suggests that the offenders given the Traditional package were the harder core offenders and thus correctly classified by Judge Todd. Similarly, it suggests that the persons given the Restrictive package were not the harder core offenders and were also correctly classified.
     We also sought to identify which if any offender characteristics were associated with Traditional and Restrictive. For this analysis we used the GLM procedure with Traditional and Restrictive as dependent variables and the offender characteristics as independent variables. The results are shown in Table 3-6.
     Only the variables sex and priors had a significant effect on Restrictive. Thus, females were more likely to receive a Restrictive package than were males, and persons with priors were more likely to receive the Restrictive package than were persons without priors. Age, sex, refuse, priors, and BAC had a significant effect on Traditional. This means that persons over 35, males, BAT refusers, persons with priors, and persons with a BAC>0.15 were more likely to receive a Traditional package than were persons 35 or less, females, persons without priors, and persons with a BAC < 0.15, respectively.

Back to Table of Contents

  SUMMARY AND CONCLUSION
     The statewide recidivism of two groups of convicted Georgia DWI offenders was studied. One group (the test group) was convicted and sentenced in a court in which the sentencing judge devoted considerable time and effort to crafting sentence packages tailored to the needs of individual offenders. The components of the sentence packages were selected from a wide range of traditional and alternative sanctions. The second group (the comparison group) was convicted and sentenced in another court in a nearby county where the policy was to impose generally only the minimum sentences mandated by state law.
A recidivism analysis using survival analysis techniques showed that the test group had a recidivism rate that was roughly one-half that of the comparison group.

These rates were adjusted for any differences in the two groups with respect to offender age, sex, and number of prior DWI offenses. The difference in recidivism between the two groups was highly significant statistically. This result gives strong support to the hypothesis that tailored sentences constructed by an informed judge are more effective in terms of reducing DWI recidivism than are statutorily mandated minimum sentences. This is important from a judicial policy standpoint, since many jurisdictions (especially those with limited sanctioning resources) impose only minimum sentences.

     We also examined the question of how the sentence packages were determined by the judge in the test-group court, that is, which sentence packages composed of which components were selected for which types of offenders. We used data from a data base kept by the judge in this study. The data base contained information on the sentence given each offender and information on the characteristics of that offender. The results of our analysis are summarized in Table 3-7 which lists the offender characteristics that had a statistically significant relationship (p�0.05) to at least one sentence component. The table also lists the offender characteristics that had a statistically significant relationship to at least one of two sentencing packages that emerged from the analysis.
     The analysis showed that having a prior DWI offense was significantly associated with all components and all packages. Compared to offenders without priors, offenders with priors:
  • got more days in jail and more days of house arrest,
  • were more likely to have to participate in AA,
  • were more likely to have to submit to periodic breath-alcohol tests, and
  • were more likely to have to undergo electronic monitoring.
     In general, offenders with priors:
  • received more severe Restrictive sentence packages, and / or
  • received more severe Traditional sentence packages.
     Other characteristics of the offense that were significantly related to the sentence were refusal to submit to a breath-alcohol test and having a BAC of 0.15 or higher. Both characteristics were associated with more days in jail, a higher likelihood of having to submit to periodic breath tests, and more severe Traditional sentence packages. Having a BAC of 0.15 or higher was also associated with a greater likelihood of having to participate in AA.
     Three offender demographic characteristics were significantly related to at least one of the sentence components or sentence packages. These were an age of 35 or more years, being of male sex, and being married. Offenders 35+ years of age and male offenders were given more days in jail, more likely to have to participate in AA, more likely to have to submit to periodic breath alcohol tests, and a more severe Traditional sentence package. Males were less likely to have received house arrest or a Restrictive sentence package than were females. Finally, married offenders were given fewer days in jail than were unmarried offenders. There were also three demographic characteristics in Todd’s data base that were not significantly related to any sentence component or sentence package. These characteristics were race, number of children, and employment status.
     In conversations with the authors, Judge Todd confirmed in general these analytically-determined relationships, but also indicated other important, more subjective, factors that did not appear explicitly in his data base. Some of these related to the circumstances surrounding the DWI incident, for example, the drinking location and the appearance of the offender in court. Offenders with a high BAC but who performed well on the roadside sobriety test were viewed as possible problem drinkers and were considered for sentences involving AA and close monitoring of their drinking. Other factors involved personal knowledge of the individual being sentenced and the offender’s demeanor and appearance in court. The sentences given offenders whom the judge had sentenced before for DWI were also taken into consideration. In general, Judge Todd favors sentences requiring abstinence, especially for those offenders indicating a drinking problem.
     It is apparent that Judge Todd performs many of the functions that are performed by a probation department in some larger jurisdictions. He conducts his own pre-sentence investigations using data he has compiled and analyzed. He requires close supervision and reporting of offender compliance with his sentences and acts quickly on failures to comply with sentence conditions. For example, offenders who fail a breath alcohol test go immediately to jail.
     This close contact with offenders during their probationary period is remindful of DWI Court programs in which offenders report back periodically to the sentencing judge for a review of their progress (or lack of progress) in completing the conditions of their "contract" with the court. Both the DWI Court program2 and the Todd Program involve judicial interaction with offenders for a period beyond traditional adjudication and sanctioning. Recent research on DWI sanctions strongly suggests that extended contact with sanctioning agencies is a very important to reducing the incidence of DWI among DWI offenders (Jones, Wiliszowski, and Lacey, 1997; Wiliszowski, et al., 1996).
     Resource requirements for the court component of the Todd Program do not appear to be any greater than those for a court imposing minimum sentences. However, sanctioning resources and a follow-up capability are necessary. So-called "self-sufficiency" programs such as assessing fees on DWI offenders are one way of financing these resources and capabilities. In fact, Judge Todd has used this approach quite successfully in his program.

     Thus, it appears that the Todd Program approach could be used in many jurisdictions. The major requirement for successful adoption of this approach would seem to be enough judicial interest in the drinking-driving problem to put such a program in place.


Back to Table of Contents

4 - CONCLUSIONS

     We evaluated two quite different approaches to sentencing DWI offenders in Georgia. One approach (called the "Todd Program" after the judge who developed it) imposed individually tailored sanctions, and the other approach imposed generally only the minimum sanctions required by state law. Our measure of effectiveness was the statewide DWI recidivism of the offenders studied. We conclude that the Todd Program was more effective by a wide margin than was the sentencing program that imposed the minimum sanctions.
     A wide range of sanction choices was available to the Todd Program. Those that required abstinence and incorporated procedures to ensure abstinence were especially popular. A common factor among sanctions not involving jail was the close contact kept between the judge and the offender during the probationary period. The need for close and extended contact is consistent with the findings of recent research which suggests that such contact with sanctioning agencies is very important to preventing future occurrences of DWI.
     The Todd Program used a number of factors in determining which sanctions to incorporate into an offender’s sentence package. Factors that appear to have been weighted the highest were number of prior DWI convictions, refusal to submit to a breath alcohol test, BAC (if tested), offender age and sex, circumstances surrounding the DWI incident, the offender’s demeanor and appearance in court, and the judge’s personal knowledge of the individual being sentenced.
     We conclude that the Judge Todd’s individualized sanctioning approach to DWI sentencing could be used in other jurisdictions with enough judicial interest in the drinking-driving problem to develop and operate such a program. And it is likely that other judges currently operating similar sanctioning programs can have confidence in their appropriateness. Jurisdictions lacking sufficient sanctioning resources might be able to acquire them, as has Judge Todd, through "self-sufficiency" programs such as assessing fees on DWI offenders.

Back to Table of Contents

REFERENCES

Jones, R.K.; Wiliszowski, C.H.; and Lacey, J.H. (1998). Problems and solutions in DWI enforcement systems. Report No. DOT HS 808 666. Washington, DC: National Highway Traffic Safety Administration.
Jones, R.K.; Wiliszowski, C.H.; and Lacey, J.H. (1997). Evaluation of alternative programs for repeat DWI offenders. Report No. DOT HS 808 483. Washington, DC: National Highway Traffic Safety Administration.
DeYoung, D.J. (1997). An evaluation of the specific deterrence effect of vehicle impoundment on suspended, revoked and unlicensed drivers in California. Report No. DOT HS 808 727. Washington, DC: National Highway Traffic Safety Administration..
Wiliszowski, C.H.; Murphy, P.V.; Jones, R.K.; and Lacey, J.H. (1996). Determine reasons for repeat drinking and driving. Report No. DOT HS 808 726. Washington, DC: National Highway Traffic Safety Administration.