Banner-Evaluation of Minnesota's High-BAC Law

3. Evaluation of Minnesota's High-BAC Law

Rates of Recidivism

Survival Analysis Models

Survival analysis further examined the association between the alcohol test result (test refusal, BAC .17-.19, BAC < .17, BAC > .20) and the rates of recidivism among first-time and repeat offenders. Separate models were developed for first-time and for repeat offenders arrested in each of the years 1998 and 1999. In addition to the alcohol test results, predictors included offenders’ gender and age.

In the presence of other predictors, the alcohol test result among first-time offenders arrested in 1998 was significantly associated with the rate of recidivism during the 12 months following arrest (Table 12). In comparison to high-BAC offenders, the likelihood of recidivism was higher among persons who refused the alcohol test (HR = 1.27, p < .01) and persons with BACs .17-.19 (HR=1.23, p < .05), but was not significantly different from persons with BACs less than .17 (HR = 0.89, p < .14). Male offenders were more likely to recidivate than female offenders (HR=1.28, p < .001). Compared to drivers ages 55 and older, the likelihood of recidivism also was higher among drivers younger than 21 years old (HR=1.95, p < .001) and drivers 21-24 years of age (HR=1.89, p < .001).

Table 12

Multivariate Survival Analysis Model of First Subsequent Offense
during 12 Months Following Arrest First-time DWI Offenders Arrested in 1998, Minnesota

 

Parameter Estimate

Standard Error

Hazard Ratio (HR) (Confidence Interval)

p-value

Gender

       

  Female

   

1.00

 

  Male

0.25

0.07

1.28 (1.12-1.47)

<.001

Age (yr)

       

  55+

   

1.00

 

  < 21

0.67

0.16

1.95 (1.44-2.64)

<.001

  21-24

0.63

0.15

1.89 (1.41-2.53)

<.001

  25-34

0.26

0.15

1.29 (0.97-1.72)

0.08

  35-54

0.11

0.15

1.12 (0.84-1.49)

0.44

Test Result

       

> .20

   

1.00

 

Refused Test

0.24

0.10

1.27 (1.05-1.53)

0.01

.17-.19

0.20

0.09

1.23 (1.04-1.45)

0.02

< .17

-0.11

0.08

0.89 (0.77-1.04)

0.14

Figure 1 graphically depicts the results from Table 12 for the cumulative adjusted likelihood of having a subsequent DWI offense by the alcohol test result for first-time offenders arrested in 1998. Throughout the first year following their arrest, the likelihood of re-arrest among high-BAC first-time offenders was not significantly different than the likelihood for lower-BAC offenders, but was significantly less than the likelihood for offenders who refused the alcohol test and offenders with BACs .17-.19.

Figure 1 – Adjusted Cumulative Hazard Function:

Cumulative Likelihood of Subsequent DWI Offense over First 12 Months by Alcohol Test Result
First-time DWI Offenders Arrested in 1998, Minnesota

click "d" for long descriptiond

Several other survival models for first-time offenders were constructed. A one-year survival analysis model for first-time offenders arrested in 1999 produced the same pattern of association between the alcohol test result and the rate of recidivism as the model for first-time offenders arrested in 1998, but the results were not statistically significant. Models also examined recidivism among first-time offenders for the first two years following arrest. Although these models also produced similar patterns for the effects of alcohol test results on recidivism as were produced in the models looking at the first year after arrest, the effects were not significant.

In one-year survival analysis models developed for repeat offenders for 1998 and for 1999, alcohol test result was not a significant predictor.