House  |   Senate  |   Joint Departments and Commissions  |   Bill Search and Status  |   Statutes, Laws, and Rules 

DWI Recidivism Rate Tables

Introduction

It is well known that DWI offenders recidivate at high rates. However, more specific information about that recidivism is generally lacking. Thus, knowledge about DWI recidivism is more often based on folklore than empirical reality.

This study provides detailed information about the recidivism behavior of Minnesota's DWI offenders.  This information can be used to answer a range of policy questions related to the control of DWI offenders, such as:

This information can also be used to help evaluate specific programs for dealing with DWI offenders, by permitting one to calculate an estimate of the "deterrence effectiveness" of the program, or by comparing the recidivism rates of certain offenders to the recidivism rates of DWI offenders statewide.

An example of this kind of program evaluation is included along with the recidivism rate tables.  This illustration evaluates the innovative "staggered sentencing program" devised by Minnesota Tenth District Judge Dehn for use with repeat DWI offenders.  (Based on these recidivism rate tables and sentencing data provided by the courts, we estimate that staggered sentencing, as used within the Tenth Judicial District, reduces DWI recidivism by approximately 50 percent.)

Methodology & Using the Data

These tables were produced using SPSS Survival Analysis (Life Table Method). The analysis involves a four-year follow-up of all 22,533 Minnesotans known to have violated implied consent (IC) and/or DWI law in Minnesota during 1995. (The study excludes non-Minnesotans who offended in Minnesota, as well as Minnesotans whose triggering DWI violation occurred out of state.)

The underlying data source is the Minnesota driver's license data base, which is maintained by the Minnesota Department of Public Safety (DPS/DVS).

The data were extracted from the data base in May 1999, and thus contain three years (1,095 days) of follow-up history for every offender included, thereby permitting the aggregate calculation of recidivism trends over a 36-month period (in 15-day increments). Given the stability of those trends, a decision was made to project the aggregate recidivism rate trend out another year to 48 months in all by simply applying the rate change from each 15-day period from the third year to the corresponding month of the fourth year).

The data extraction was performed for the ongoing Simon Recidivism Study (Steve Simon, University of Minnesota Law School, principal investigator, Dan Storkamp, Minnesota Department of Corrections, and Jim Cleary, Minnesota House of Representatives.)  The Simon data base contains 27,326 records (offenders).

The subfile underlying these rate tables contains 22,533 observations, after excluding all non-Minnesotans, and Minnesotans whose triggering 1995 violation occurred outside the state.

The Life Tables reflect the DWI recidivism history separately for DWI offense levels one through six, and seventh or more. In using these recidivism rate tables---e.g., in the evaluation of Judge Dehn's staggered sentencing program in this example---the following definitions and operational procedures must be observed:

Three data fields are shown. (Other data columns are hidden from view for simplification, but can be viewed using the format/columns/unhide command.)

1) Interval Start Time: 15-day time intervals were used for the life analysis. The interval start time is shown in the first data column, where: 0 = days 0 to 14; 15 = days 15-29; 30 = days 30-44; etc.

2) Cumulative Proportion Surviving at End of Period: 

The number in this column indicates the proportion of 1995 IC/DWI violators who had not reoffended by the end of the corresponding time period (i.e., who have not been re-arrested for an implied consent or DWI violation).

3) Cumulative Proportion Recidivating by End of Period:

The number in this column indicates the proportion of 1995 IC/DWI violators who reoffended by the end of the corresponding time period (by again violating IC/DWI law). [Column 3 = 1.0 minus Column 2]

For example, for first-time offenders, scroll down to the fourth row of data. The 45 in the first column refers to the time period 45 to 59 days following the offense date (the exact date of course is unique to the offender; this interval is thus calculated relative to each offender's exact offense date). In the corresponding row of the second column, the figure 0.9862 indicates the proportion of first-time offenders who had not yet recidivated by being charged with another IC/DWI violation within 59 days of their initial 1995 offense.

In the corresponding row of the third column, the figure 0.0138 indicates the proportion of first-time offenders who recidivated within 59 days of their first offense (where column 3 is calculated as 1.0 minus column 2).

Columns "a to d" present an example of a program evaluation using these DWI recidivism rate tables in an evaluation of an innovative DWI sentencing program.  The analyst must follow the following steps:

  1. Sort the program participants (the offenders) by offense level  (for this step, do not count any recidivism while on the program - thus, a person who had two violations at the time of sentencing but who has since recidivated while on the program, is nevertheless coded as a second-time offender).
     

  2. For each participant, including those who have failed the program by committing a new IC/DWI violation, determine the number of days from the person's triggering incident date to the date at which data follow-up was terminated. For a person who has failed by reoffending with a subsequent impaired driving violation, the termination date would be the date of the person's arrest for that new violation.
     

  3. Column (a)
    For each participant - whether successful or not - tabulate the person in column (a) in the row corresponding to both his or her DWI offense level and days time spent successfully on the program.  If the person's actual time is greater than the maximum (1,095 days, or three years), then code that person in the highest category (interval start time = 1,080 days); (note that in this manner survival beyond three-years’ time is not credited; this limitation is an artifact of the available follow-up data for the study, which was limited to limited to three years’ total);

    (In this study, a four-year follow-up period was used, with the fourth year of recidivism data being projected by trending out from the third year of the data. Thus, the rate tables extend out to 1,460 days, with the final category beginning at 1,455 days, where any observed survival beyond four years not being fully credited.
     

  4. Column (b)
    This column is calculated automatically, as a function of column (3) x column (a).
     

  5. Column (c)
    If the person has recidivated while on the program, then tabulate that person in column (c) also.
     

  6. Column (d)
    For purposes of data checking, code each participant's name in column (d), as well;  by user option, this column may be hidden from view when printing the spreadsheet results.

Data Checking:

After entering the data, double check all entries and totals for accuracy.  Particularly check that the cases were entered into the proper row of the table, corresponding to the correct offense level and time period for the person.

Tabulation:

Following data preparation, entry, and checking, columns (a) to (c) are subtotalled within each offense level and totalled across all offense levels, and are compared as illustrated. See especially the summary table.

Interpreting the Results:

Results may be interpreted in the aggregate to acquire an overall measure of the program results versus the recidivism behavior of all similar Minnesota DWI offenders generally, or within any specific offense level to determine how well the program is working with that specific level of offender.

For the example herein, the grand totals shown at the end of the table indicate that recidivism is approximately one-half lower under the Staggered Sentencing Program (as used by Judge James Dehn in the Tenth Judicial District), as compared to Minnesota DWI offenders statewide (i.e. 6 failures, compared to an expected 9.3, based on the recidivism rate tables).

[Note that these results are based on the first 61 offenders who have been given staggered sentences by Judge Dehn, as monitored for the time period ending July 15, 2002.  These results are subject to change in either direction as more time passes and as more offenders are assigned to the program.]

From the Simon Recidivism Study
University of Minnesota Law School; Research Department, Minnesota House of Representatives; Sentencing Guidelines Commission
Data analysis: September 18, 2002

Return to Criminal Justice
Return to House Research home