3 - Results

Process

System Description

The lower BAC countermeasure is one of a class of alcohol-related crash counter-measures seek to prevent driving after drinking by deterring and incapacitating drinking- drivers. As is the case with most countermeasures of this class, the Traffic Law System is the major societal system that operates the lower BAC countermeasure, doing so through three of its major subsystems, Enforcement, Adjudication, and Sanctioning [See (Jones and Lacey, 2001) for a definition and discussion of these subsystems and their functions.] Drivers explicitly targeted by the lower BAC law are those who have been convicted and sanctioned for OUI or for violating the State’s .08 per se law.

The process starts with a suspension of the driver license for OUI (Figure 3-1). The suspension is administrative in nature and is imposed by the Bureau of Motor Vehicles (BMV). The minimum period of suspension is 90 days for a first offense, but increases to 18 months for a second offense within 10 years, and to four years for a third and sub-sequent offense within 10 years. These periods have not changed since the lower BAC law first became effective in 1988. The suspension is a “hard” suspension for persons with two or more offenses, in that no work or limited licenses are granted.

The offender is also required to attend an alcohol education program and, if indicated by the BMV, to participate in alcohol treatment and rehabilitation. The State’s Driver Education and Evaluation Program (DEEP) administers the program. After the period of suspension and after satisfactory completion of the program, the driver may apply for a conditional license. The license prohibits driving with any amount of alcohol in the blood, and the license will be granted if and only after the BMV has been notified of the driver’s completion of DEEP. The conditional license is annotated with the letter “Q” on its face as a restriction, and the reverse side of the license explicitly defines the restriction as “Q – Conditional License. May not operate after consuming intoxicating liquor.” In 1988, the conditional license prohibiting driving with any alcohol in the blood was in effect for two months, but in 1995 the effective period was increased to one year for a first offense, and to 10 years for a second and subsequent offense.

If the holder of a conditional license is lawfully stopped by a law enforcement officer (e.g., for speeding or an equipment violation), the officer notes the driver has a conditional license and observes for any signs of drinking. If any sign is observed, for example, an odor of alcohol, the officer may administer field sobriety tests such as NHTSA’s Standardized Field Sobriety Test (SFST). If the driver is found not to have a BAC of .08 or higher but to have consumed alcohol, the officer prepares a short report indicating the driver “operated or attempted to operate a motor vehicle while having any amount of alcohol in the blood.” At this point the driver is allowed to proceed, and the officer will submit the report to the BMV. The driver will later be notified of driver license action (a one-year suspension of the driver’s license) by the BMV

Figure 3-1: Processing Of a Drinking Driving Suspect in Maine

figure 3-1 click [d] for long description[d]

Another course of action is followed when an SFST test or other factors lead the officer to believe the driver is at or above Maine’s standard BAC limit of .08 and may have violated the provisions of the DUI law. Then, the driver is transported to a BAC testing facility and asked to submit to a BAC test. (The use of preliminary breath testers [PBT] is not permitted in Maine.) If the driver refuses, the officer reports the refusal to the BMV, and the conditional license is suspended for a period of two years for violating the State’s implied consent law. If the driver takes the BAC test and is found to have a BAC of .08 or more, the conditional license is suspended for the same periods as apply to an OUI conviction for a second or third offense, 18 months and four years, respectively. Finally, if the driver takes the BAC test and is found to have a BAC substantially less than .08 (but greater than .00) and is deemed not to be impaired, he or she is released, and the BMV is notified the driver has violated the non-drinking requirement of the conditional license. The driver is then notified his or her license has been suspended. Drivers who have violated the conditional license law, the OUI law, or the implied consent law will then begin the above process all over again.

Note that driver license suspensions can be appealed and then adjudicated by a BMV hearing officer. However, the suspension may remain in effect pending a hearing. Decisions rendered by a hearing officer also may be appealed through the court system. Such appeals are heard initially by the Superior Court.

Note also that procedures for detecting suspected violators of the lower BAC law differ somewhat from those followed for suspected violators of the OUI law. Because of the very low BACs involved, signs of gross impairment will not be present, and as indicated above, detection will usually occur after a stop for some other driver action. Such actions are observed most often during routine patrol. Evidence of drinking most often consists of such signs as the presence of containers of alcoholic beverages in the vehicle and the odor of alcohol in the vehicle. However, SFST tests are administered when a driver appears to be impaired, but PBTs are not permitted by law. Officers we inter-viewed said they seldom used the lower BAC law as a basis for license action, having found most convicted OUI offenders who had been drinking were at a high enough BAC to charge for OUI. If a later BAC tests indicated the driver’s BAC was too low to obtain an OUI conviction or there were other impediments to obtaining an OUI conviction, then the lower BAC law is used.


Level of General OUI Enforcement Activity

The number of OUI arrests in Maine has varied between 8,000 and 12,000 per year since 1978, with no noticeable trends (Figure 3-2). If anything, the number of arrests appeared to rise after the law changes of 1988 and 1995, only to decline again to their pre-law levels. By contrast, the number of OUI convictions tends to trend downward since 1984, while reflecting the peaks and valleys of the arrest figures (Figure 3-3).

Meanwhile, the mix of OUI convictions with respect to number of prior OUI convictions showed little change prior to 1995. The mix changed noticeably in 1995, when OUI convictions of drivers with priors became a larger component of all OUI convictions (Figure 3-4). Multiple offenders as a percent of all offenders went from an average of 28% in the 1988-1995 period to an average of 37% in the 1996-2000 period (Figure 3-5).

Figure 3-2: OUI Arrests in Maine, 1978 - 2000

[d]

Figure 3-3: OUI Convictions in Maine, 1978 - 2000

[d]

Figure 3-4: Distribution of OUI Convictions in Maine by Offense Number 1988 - 2000

[d]

Figure 3-5: Mean Percent of OUI Convictions with 95% Confidence Intervals for Multiple Offenders, Second Offenders, and Third Offenders by Time Period

[d]

For second offenders, these percentages are 23% and 26%, respectively, and for third offenders, 5% and 11%, respectively. Because of the small 95% confidence intervals surrounding these figures (indicated by vertical bars on the figure), all of these changes are highly significant statistically (p≈0.0001).

While these increases in the multiple-offender component of OUI convictions could indicate an increase in system emphasis on this group, they are more likely to be due primarily to an increase from 6 years to 10 years in the so-called “look-back period” for defining a prior offense. Thus, some offenders who would be classified as a first offender in the 1988-1995 period would be classified as a second or third offender in the 1996-2000 period, as there were more years in which the offense could have occurred.

Administrative per se license suspensions increased markedly in 1984 when the State’s first administrative per se law (.10) went into effect (Figure 3- 6). In 1984, about 4,300 licenses were suspended under this law compared to about 6,100 in 1990. These suspensions declined again, and then reached their 1984 level in 1994, where they remained through 2000.

Finally, refusals to take a BAC test averaged about 1,500 per year in 1978 through 1986 (Figure 3-7). Refusals then increased sharply to about 2,000 per year in 1990, and dropped quickly again to about 1,100 per year, where they remained through 2000. However, refusal rates showed a quite different pattern, starting out at about 18% to 19% in 1978, dipping to 14% in 1984, rising to 18% in 1987-1988, and then falling steadily to 10% in 1999 (Figure 3- 8). (Arrest data for 2000 were not available at this writing.) It is interesting that the rate for 1987 (18.4%) is almost identical to the mean of the rates in 41 States as noted in a NHTSA-sponsored study of implied consent refusal impact (Jones, Joksch, and Wiliszowski, 1991).


Conditional License Actions

In 1989, the first year after the passage of the first lower BAC law, the BMV suspended a total of 50 conditional licenses (Figure 3-9). The suspensions rose rapidly to about 250 in 1993 and remained there until 1997 before falling to about 200 in 1998. At this writing, only about half of the suspensions (100) are for non-refusal violations (Figure 3-10). Non-refusal conditional license suspension hearings are less than 2.5 % of all BMV hearings (Figure 3- 11) and numbered fewer than 60 per year over the past few years (Figure 3- 12).

Additional data were available for hearings for all conditional license suspensions scheduled for the year 2000, including those suspended for driving after drinking and those suspended for refusing to take a BAC test. A total of 202 such hearings were scheduled, of which 101 were continued until later, and 101 were disposed. Only 43 reached the actual fact-finding stage before a hearing officer, and of these, only four (9%) resulted in the suspension being rescinded. However, another 27 of the 101 cases (27%) were rescinded because the law enforcement officer did not appear at the hearing, so 31 of the total 101 disposed cases were rescinded. The preponderance of all the rescinded cases (87%) was due to the law enforcement officer failing to appear at the hearing.

Figure 3-6: Administrative Per Se Suspensions in Maine, 1984 - 2000

[d]

Figure 3-7: BAC Test Refusers in Maine, 1978 - 2000

[d]

Figure 3-8: BAC Test Refusal Rate in Maine, 1978 - 1999

[d]

Figure 3-9: Total Number of Conditional License Suspensions in Maine, 1988 - 2000

[d]

Figure 3-10: Number of Conditional License Suspensions in Maine by Type of Suspension, 1988 – 2000

[d]

Figure 3-11: Hearings for Non-Refusal Conditional License Suspensions as a Percent of All Hearings in Maine, 1989 - 2000

[d]

Figure 3-12: Number of Hearing Dispositions for Non-Refusal Conditional License Suspensions in Maine, 1989 - 2000

[d]

Operational Environment

The lower BAC law was one element of an ongoing series of legislation aimed at alcohol-impaired drivers, and must be assessed against this background. Typically, such laws lowered BAC limits, raised fines, made incarceration more likely and longer, and placed more emphasis on administrative sanctions. Such legislation was not unique to Maine, but occurred in most other States as well. As indicated above, Maine passed the first version of its lower BAC law in 1988, and then revised it in 1995 to explicitly pro-scribe any amount of alcohol as illegal for convicted OUI offenders. However, other traditional sanctions also became more severe as a result of the 1995 legislation, so that, strictly speaking, any effect the lower BAC law may have had post-1995 would have to be attributed to the totality of the legislative changes, plus any other changes in OUI enforcement processes that may have occurred.

However, our discussions with enforcement officials in Maine indicate their procedures for detecting, apprehending, and processing OUI offenders changed very little or not at all after 1988. Further, while our two comparison States also strengthened their impaired driving laws after 1988, neither State had a lower BAC law comparable in any way to Maine’s law. Overall, the main difference between the operational environments in Maine and the comparison States with respect to impaired driving by convicted OUI offenders was Maine had a lower BAC law and the comparison States did not. Thus, any differential impact between Maine and the two comparison States could credibly be ascribed to the addition of lower BAC component to Maine’s law. The impact analysis follows below.

Impact

All Fatal Crashes

First, we examined the ratio of percent convicted offenders in Maine to percent convicted offenders in the two comparison States (New Hampshire and Vermont). Our interrupted time series model incorporated interventions at January 1988 (BAC limit =.05) and January 1995 (BAC limit=.00), and modeled the logarithm of the ratio differenced by one semi-month period. The logarithmic transformation gave the better fit, and differencing was necessary to achieve stationarity. Moving average components at 2 and 4 lags and an autoregressive component at 1 lag were also included in the model.

No effect was found at the 1988 intervention point, but the 1995 intervention did show a 27% reduction in the ratio from of what it would have been post-1995 without the law (Figure 3-13). However, the effect, though fairly large, was nowhere near significance (p=.396) because of the large standard error of the data.

Figure 3-13: Ratio of Percentage of Convicted Offender Drivers in Fatal Crashes in Maine to the Percentage of Convicted Offender Drivers in Fatal Crashes in New Hampshire and Vermont, 1975 - 2000

[d]

We then examined the Maine series and the comparison series separately, again using logarithms of the percentages differenced by one period. Maine series incorporated two moving average components, one at 1 lag and the other at 2 lags. The comparison series incorporated three autoregressive lags, at 1, 2, and 9 lags.

Figure 3-14: Convicted Drivers in Fatal Crashes as a Percentage of All Drivers in Fatal Crashes in Maine, 1975 - 2000

[d]

Figure 3-15: Percent Convicted Drivers in Fatal Crashes in Comparison States, 1975-2000

[d]

We found a near-significant reduction (p=0.059) in convicted offenders as a percent-age of all drivers in fatal crashes. This percentage decrease amounted to 45% (from 12.9% to 7.1%) in the Maine series at the 1988 intervention point and continued on through the remainder of the series (Figure 3-14). Further, there was another 10% reduction (p=0.480) at the 1995 intervention point. By contrast, the series for the comparison group containing Vermont and New Hampshire increased slightly (11%) at the 1995 intervention point (Figure 3-15).

Thus, the time-series analyses strongly suggest a fairly large reduction in Maine’s percent convicted-offender drivers in fatal crashes after the lower BAC law introduction in 1988. There is also evidence this lower percentage continued through the year 2000, and may have become even lower after the 1995 reduction in BAC to .00. Precise estimates of these reductions and their statistical significance are difficult because of the small number of crashes in Maine and resultant large standard errors.


Alcohol-Related Fatal Crashes

We also analyzed the effect of the Maine law on alcohol-related fatal crashes, using convicted offenders as a percentage of fatal-crash involved drivers with a given BAC as the effectiveness measure. We used FARS data going back to 1982 for this analysis. This is the earliest year for which imputed values of driver BAC4 were available, allowing analyses for BACs of .01-.09 and .10+. Because of the very small Ns involved, a time series analysis such as was used for all fatal crashes was no longer feasible, and a Generalized Linear Model (GLM) analysis was used instead. Two independent variables were used in the analysis: Group (Maine and Comparison States) and Period (B, 1982-1987; A1, 1988-1995; and A2, 1996-2000). Both main effects and interaction effects of these two independent variables were computed.

Figure 3-16 shows the results for the .01-.09 measure. In period B (before the .05 limit for convicted drivers), 7% of the Maine drivers at .01-.09 were convicted offenders. This percentage increased to about 9% in period A1 (after the .05 limit but before the .00 limit), and then decreased again to about 7% in period A2 (after the .00 limit). For the comparison group, the percentage of convicted offenders at .01-.09 had an increasing trend starting at 10% and ending at 12%. The interaction effect between Group and Period was nowhere near statistical significance (p=0.756).

The results for Maine at the .10+ BAC level indicated essentially no change from B to A1, but a decrease of 18% to 13% from A1 to A2 (Figure 3-17). Again, in the comparison States, the percentage of convicted offenders trended upward, going from 15% in B to 16% in A1 to 17% in A2. The interaction effect between Group and Period significance began to approach statistical significance (p=0.168). For the last two periods, Maine’s reduction compared to the comparison group’s increase was closer to being significant (p=0.106), but was still not significant at the 0.05 level.

Figure 3-16: Convicted Drivers as a Percentage of All Fatal-Crash Involved Drivers in Maine with a BAC of .01 - .09 by Period

[d]

Figure 3-17: Convicted Drivers as a Percentage of All Fatal-Crash Involved Drivers in Maine with a BAC of .10+ by Period
[d]

The change in percent convicted offenders of all drivers per year within period A2 is also of interest. For .10+ drivers in Maine this percentage follows a decreasing trend amounting to 1.9% per year (p=0.0698), and for the comparison states this percentage follows an increasing trend amounting to 2.4% per year (p=0.0230) (Figure 3- 18).

Figure 3-18: Convicted Drivers as a Percentage of All Fatal Crash-Involved Drivers with a BAC of .10+ by Group and Year, 1996-2001

[d]

Recidivism

The general design of our recidivism analysis was described earlier in this report on page 8. The recidivism file provided by the Maine BMV contained a record for each conviction of every driver in each cohort. Members of each cohort were drivers who had at least one alcohol-related traffic violation in a given study year. Years of concern were 1993, 1994, 1995, 1996, and 1997. Also included were the conviction date, the violation date, the violation type, driver age, and driver sex. All of each driver’s recorded convictions were included, not just those occurring in their cohort year. For each driver, an index violation date was computed as the date of the driver’s first conviction in the driver’s cohort year. Prior alcohol-related traffic convictions were measured from this index date, as was the time before the first subsequent conviction. Four types of violations/convictions were included as alcohol-related: operating under the influence of liquor (OUI), administrative per se, refusal to take a BAC test, and driving with a BAC ³ .05 or .00 (depending on the year of the violation).

Table 3- 1: Covariate Values by Period

Covariate
Period
Before
During
After
Age
<35
59%
58%
57%
=35
41%
42%
43%
Sex
Female
17%
18%
18%
Male
83%
82%
82%
Priors
0
61%
61%
61%
1+
39%
39%
39%

 

The recidivism analysis examined the probability of another alcohol-related violation on or before a given time T after the index violation. The primary objective was to determine if this probability P(T) was lower for cohorts from the two-year period after the 1995 legislation lowering the BAC limit to .00 than it was for cohorts from the two-year period before the 1995 legislation. To do this, survival analysis models of time to the first violation after the index date were developed using cohort time period as the independent variable, and age, sex and number of prior alcohol-related convictions as covariates. The major emphasis was on OUI violations, which had more severe con-sequences than did the administrative violations. To avoid over-counting priors, analyses of violations that included administrative violations used a database from which duplicate violations occurring on the same date were removed. (A driver could accumulate more than one conviction for the same drinking-driving incident, one for administrative per se, refusal, or being a convicted offender driving with a lowered BAC; and one criminal for OUI.)

The driving records of 35,433 drivers of age 16 or more were included in the analysis, broken down by covariate value and period as shown in Table 3-1. There was little change in any of the covariates with respect to period – about 58% of the drivers were under the age of 35, 82% were male, and 39% had at least one alcohol-related prior offense. Also, the number of drivers in a cohort changed very little during the five years studied – there were about 7,300 each in 1993, 1996, and 1997; and about 6,800 each in 1994 and 1995.

The survival analysis used an accelerated failure time (AFT) model with censoring of observations having no violations after the index date. A Weibull distribution of the ran-dom disturbance factor e was selected as giving an excellent fit to the data. The mechanization of the model provided in the SAS LIFEREG procedure was used. Driver age was not a significant covariate (p=0.102), but driver sex (p<0.0001), priors (p<0.0001), and most important, period (p=0.0047), were significant. The model indicated that, con-trolling for driver sex and priors, the recidivism time of the “after” group at any given value of P was about 12% longer than that of the “before” group. This means, for example, that if at P=0.15, the recidivism time of the “before” group was 24 months, then the recidivism time of the “after” group would be 1.12´24=27 months. The effect of driver sex and priors was larger, with females having a 33% longer recidivism time than males, and drivers without priors having a 38% longer recidivism time than drivers with priors.

The LIFEREG procedure also permits one to compute the recidivism probability P at any given time for given values of covariates. We used another model without the non-significant covariate “age” for this. Figure 3-19 shows the recidivism of male drivers with one or more priors by period. It is seen the “after” group has a lower recidivism probability than the “before” group for all values of months after index violation.

Figure 3-19: Modeled Recidivism of Male Drivers with Priors by Period

[d]

The data from Figure 3-19 were used to get a closer picture of the recidivism of the two groups one year and two years after the index violation. The results are shown in Figure 3-20 below for males in the before and after group. The recidivism of the “after” group was about 7% lower than the recidivism of the “before” group, both for drivers with priors and drivers without priors.


Awareness

To determine the level of awareness of the lower BAC law, staff from the Maine Driver Education and Evaluation Program conducted a survey of offenders statewide who were required to participate in the Program. Both first offenders and repeat offenders were included. Respondents were asked if they knew about the law, when and how they first heard about it, and how many times they had been convicted of OUI.

Results are summarized in Table 3 2. Of the 222 persons who responded, 67% said that they were aware of the law. Repeat offenders were more likely to know about the law than first offenders (91% of repeat offenders knew about it versus 61% of first of-fenders), but awareness did not differ significantly with respect to sex or age (under 40 or 40+).

Figure 3-20: Recidivism of Male Drivers One Year and Two Years after the Index Violation, With and Without Priors

[d]

Table 3 2: Awareness Survey Statistics

Independent Variable
% Aware
p
Odds Ratio of Aware
p
Sex
Female
68.9
0.741
--
0.376
Male
66.3
1.40
Age
< 40
69.9
0.280
--
0.295
40+
62.7
0.70
Prior OUIs
Yes
90.9
0.0001
5.34
0.0003
No
61.4
--

 

We also conducted a logistic regression analysis of law awareness with the same variables to determine the effect of one variable while controlling for the other two. Aware-ness (“yes” or “no”) was the dependent variable, and again sex, age, and prior OUIs (“yes” or “no”) were the independent variables. The analysis indicated a significant odds ratio for prior OUIs (OR= 5.3, p=0.0003), but none for offender sex and age. Thus, of the three independent variables examined in the survey, awareness of the lower BAC law among repeat offenders was very high for younger and older offenders of both sexes. Apparently, the existence of the law was well publicized: 45% of those who knew about the law learned about it before their latest OUI conviction.

These findings lend support to the hypothesis (but, of course, do not prove) that the lower BAC law has contributed to the reduction in alcohol-related fatal crashes in Maine.


Summary and Conclusions

Our evaluation of Maine’s lower BAC law considered both the process of enforcing the law, and the law’s impact on fatal crashes in general and alcohol-related fatal crashes in particular. Awareness of the law among convicted OUI offenders was also estimated.

With respect to the OUI enforcement process in general, data provided by the Maine Bureau of Motor Vehicles indicate no large changes occurred in the level of activity after 1988. Arrests were fairly stable, averaging about 10,000 per year, and convictions trended down slightly, starting at 8,000 in 1988 and reaching 6,000 in 2000. Administrative per se actions against the driver license declined slightly to about 4,000 in the year 2000, but the BAC test refusal rate declined nearly 50%, from 18% to 10% from 1988 - 1998.

For enforcement of impaired driving laws aimed at convicted OUI offenders, data from Maine’s driver records file indicate convictions of previously convicted offenders increased slightly to about 37% of all OUI convictions, but this was probably due to an increase in the lookback period for determining the existence of prior OUI convictions. License suspensions flowing from violations of the lower BAC law reached 250 in 1993 and settled down to about 200 in the year 2000, about half of which were for refusing to take a BAC test. Such suspensions amounted to less than 5% of all OUI-related administrative suspensions, and less than half were disposed in an administrative hearing. Of the 101 disposed cases in 2000, 41 of the suspensions were rescinded, 87% of which were due to the law enforcement officer failing to appear

Our discussion with staff from enforcement agencies and Maine’s Bureau of Motor Vehicles indicated the lower BAC law did not place any significant burden on those agencies.

Our impact analysis used two neighboring States, Vermont and New Hampshire, to help account for factors other than the lower BAC law that may have influenced the occurrence of fatal crashes in Maine. Neither of these two comparison States had a lower BAC law, and both had an OUI enforcement environment similar to Maine’s. The impact analysis showed that in Maine, OUI-convicted drivers in fatal crashes as a percent of all drivers in fatal crashes decreased by 45% (from 12.9% to 7.1%) after 1988, and de-creased still more after 1995. At the same time, the percent increased slightly in the two comparison States.

With respect to alcohol-related fatal crashes, we found that for drivers at a BAC of .01-.09, convicted offenders as a percent of all such drivers in Maine stayed about the same over the 1988-2001 period, but increased gradually in the comparison States. But for drivers at .10+, the percent in Maine decreased in the 1988-2001 period; and the per-cent in the comparison States increased. And after 1995, convicted offenders as percent of fatal-crash involved drivers at .10+ decreased in Maine, and increased in the comparison States.

Our analysis of OUI recidivism in Maine revealed the law affected recidivism very little. For drivers with prior convictions who were convicted in 1993-1994, the two-year recidivism rate fell from 14.6%, to 13.6% for drivers with prior convictions who were convicted in 1996-1997. A similar decrease of 10.7% to 9.9% occurred for drivers with no prior convictions.

Finally, awareness of the lower BAC law in Maine was high in late 2002: 67% of the OUI offenders surveyed said they were aware of the law, and 45% knew about it before their most recent OUI conviction. Awareness was significantly higher among repeat of-fenders than among first offenders.

These measures of activity, impact, and awareness suggest strongly Maine’s lower BAC law had a positive effect on fatal crashes involving drivers with prior OUI offenses, and also on fatal crashes involving such drivers with a BAC of .10 or more. No effect was noted for fatal-crash-involved convicted offenders with a lower BAC. Further, the small number of enforcement actions against violators of the lower BAC law and the absence of any meaningful reduction in OUI recidivism, suggest the effect was a general deterrent effect rather than a specific deterrent effect.

Despite the strong indications of an effect, we cannot say unequivocally the lower BAC law alone was responsible because of the concurrent existence of other OUI countermeasures. For example, Maine’s implementation of its standard .08 per se law for the general public (and other repeat offender provisions) also occurred in 1988, and these could have been at least partially responsible for the reduction in percent convicted of-fenders in fatal crashes after 1988 noted above. However, it can be said with some confidence the law was an important element of Maine’s overall effort to decrease alcohol-impaired driving among convicted OUI offenders.


4 FARS contains so-called imputed values of driver BACs for cases where BAC was not measured. NHTSA’s new multiple imputation model was used here.

|
|
|