Abstract
Past studies have demonstrated that police officers fail to detect a substantial proportion of alcohol-impaired drivers during traffic enforcement and that the use of passive alcohol sensors (PAS) could increase the driving-under-the-influence (DUI) arrest rate. Does the use of a PAS in routine traffic enforcement by officers without specialized DUI training increase the detection and arrest rate of alcohol-impaired drivers? In Anne Arundel County, Maryland, the Police Department provided the PAS devices to 24 randomly selected officers, divided equally between two squads of 12 officers each (one squad with the PAS and one squad without). After both squads made approximately 500 traffic stops each, the squads switched roles with regard to using the PAS, and the pattern was repeated. Overall, there were no significant differences in the DUI arrest rate between the officers with the PAS and the officers without the PAS, although there was evidence that the PAS helped some officers increase their DUI arrests. In summary, the PAS is probably best used at sobriety checkpoints rather than during routine stops.
Drivers in the United States admit to driving within 2 hours of drinking alcohol close to one billion times annually [Royal, 2003]. Clearly, deterrence to impaired driving in the United States is constrained, as only 1 driving-under-the-influence (DUI) or driving-while-intoxicated (DWI) arrest is made for every 88 reported episodes of driving over the illegal blood alcohol concentration (BAC) limit [Zador, Krawchuk, and Moore, 2000]. Deterrence, as described by Ross [1984], is a function of the perceived probability of apprehension, the severity of the resulting sanction, and the swiftness with which the penalty is administered. There is substantial evidence that the most important of those factors is the probability of apprehension [Ross and Voas 1989; Ross, 1992]. Thus, raising the perceived probability of apprehension is the most essential element of an effective DUI enforcement program.
The passive alcohol sensor (PAS) is a device developed to assist police in identifying drinking drivers. The PAS draws in mixed expired and environmental air from in front of the subject’s face and passes it into a fuel cell sensor that can detect very small amounts of alcohol. The sensor, which is built into the standard police flashlight, displays the estimated BAC of a driver using a nine LED light bar indicator running from .02 (1 bar) to .12 BAC and higher (9 bars). Evidence from police use of the PAS with drivers stopped at random roadside surveys indicates that using a threshold of 6 out of the 9 light bars can identify 74% of drivers with BACs of .08 or higher with a 99% accuracy (i.e., specificity) [Voas, Romano, and Peck, 2006]. Because the PAS measures alcohol in mixed expired air, it is not as accurate as the evidential devices used to determine the BAC of arrested drivers. The results of these evidential testers are used as evidence in court proceedings. Unlike the evidential devices, however, which are considered a “search” under the Fourth Amendment to the U.S. Constitution and cannot be administered without reason to believe the driver is impaired, the PAS is not considered a “search.” The PAS is considered by many in the enforcement arena as “an extension of the officer’s nose”; consequently, the use of a PAS at any point in the investigation is believed to be constitutional [Fields and Henricko, 1986; Manak, 1986].
The PAS has been used most successfully by officers at sobriety checkpoints. A number of studies [e.g., Jones and Lund, 1986; Ferguson, Wells, and Lund, 1995; Ferguson, 1995; Wells, Greene, Foss, et al., 1997] have demonstrated that officers at checkpoints fail to detect about half of the over-the-limit drivers they interview and that the use of passive sensors at checkpoints increases the detection of these impaired drivers by approximately 50%. Research conducted by the Insurance Institute for Highway Safety (IIHS) indicates that the PAS units have a smaller effect when used by police on special DUI patrol operations [Lund and Jones, 1987] and on routine patrol [Kiger, Lestina, and Lund, 1993] (an increase in detection of about 8 to 10%).
Technically, any indication of drinking or impairment is a basis for requesting the driver to get out of the car for a field sobriety test. Following the field sobriety test, officers typically request a preliminary breath test (PBT) because experienced drinkers may perform adequately on the field sobriety test and still have high BACs. The use of the PAS has been shown to detect drinking drivers and result in the conduct of field sobriety tests for drivers with lower or borderline BACs than would be the case for officers without such devices. The PAS has demonstrated a strong correlation with evidential results [Foss, Voas, and Beirness, 1993; Fiorentino, 1997]. A recent field study showed that the PAS readings correlate very highly with the result of PBTs (a Spearman correlation of .79), but they do vary by different subgroups of drivers (crash-involved vs. stopped on the roadside) [Voas et al., 2006].
Police officers frequently use the odor of an alcoholic beverage on the breath of drivers for decisions on proceeding further for sobriety testing. In an experiment using 20 experienced police officers, however, alcoholic beverage odors were not detected 40% of the time when subjects with BACs exceeding .08 had eaten some food [Moskowitz, Burns, and Ferguson, 1999]. It would appear that the use of the PAS by police would increase the detection of alcohol-impaired drivers during brief traffic stops. PAS devices have the ability to detect low levels of alcohol that can help with drivers younger than 21 where zero-tolerance laws (.02 BAC) apply.
An enforcement strategy that has the potential to attract media and public attention is the use of technology by police that helps them detect impaired drivers when they are stopped for a traffic violation [Voas and Layfield, 1983]. Passive alcohol sensors are devices that have been designed to assist police officers to better detect and identify alcohol-impaired drivers at traffic stops, sobriety checkpoints, and crash scenes. The PAS has been shown under laboratory conditions to identify 95% of drivers with BACs=.10 when it is used within 5 inches of the subject’s mouth [Cammisa, Ferguson, and Wells, 1996].
With this as background, the Maryland Highway Safety Office decided to purchase and issue PAS devices to traffic patrol officers in Anne Arundel County, Maryland. The objective was to determine if the PAS devices increased the detection and citation rate of alcohol-impaired drivers during routine traffic enforcement.
METHODS
The Anne Arundel County, Maryland, Police Department provided 12 officers (randomly selected) in one patrol squad (Squad A) with PAS devices to see if they facilitate detection of impaired drivers. The manufacturer of the sensors (PAS Systems International) provided the PAS training on how to use and properly maintain the devices. A comparison squad (Squad B) of 12 officers (randomly selected) were not given the PAS devices during the first round of traffic stops. Each squad of officers conducted routine traffic stops over several months, and each squad submitted approximately 500 Traffic Stop data forms. At the end of that period (Round 1), the squads were switched: Squad B officers were then trained by PAS Systems International and equipped with the PAS devices, and Squad A officers enforced the traffic laws without the PAS. Officers from both squads went about routine traffic enforcement for several more months, and another 500 Traffic Stop data forms were received from each Squad (Round 2). Round 1 enforcement activities were conducted from June 2004 to July 2005, and Round 2 enforcement activities were conducted from October 2005 to February 2006 in Anne Arundel County, Maryland. By switching the use of the PAS devices from Squad A to Squad B, factors such as officer motivation, training, and certain individual differences could be controlled to some extent in the analyses. The actual usefulness of the PAS in detecting impaired drivers should be more obvious. Table 1 in the next section illustrates the design and shows the number of traffic stops made by each squad in each round.
Table 1.
Number of Traffic Stops
| Round 1 | Round 2 | Total | ||
|---|---|---|---|---|
| Squad A (with PAS) | 508 | Squad A (w/o PAS) | 538 | 1046 |
| Squad B (w/o PAS) | 494 | Squad B (with PAS) | 579 | 1073 |
| 1002 | 1117 | 2119 | ||
Information on each traffic stop made by each participating officer in both squads was captured on a data form and sent to the authors for processing and analyses. For the squad using the PAS, usage at the stop, the PAS reading, a PBT result (if taken), and whether an arrest was made for DUI were recorded. In addition, the reason for the traffic stop and demographics of the driver were recorded. Any arrests, citations, warnings, and violations were recorded, along with the results of any evidential tests. The same data were recorded for the squad not using the PAS, except the data concerning the PAS. The data were processed for 2,119 traffic stops and distributions and various tables were generated and examined. Chi-square tests (two-tail) were used to determine if there were any significant differences in the proportions or rates of interest using the Statistical Package for Social Sciences software [SPSS, 1997]. The main effect tested was the DUI arrest rates for officers using the PAS compared to officers not using the PAS.
At the conclusion of the data-collection period, a questionnaire was issued to each officer eliciting information about their general experience with using the PAS and whether they would continue using it. This was done to gain insight into officers’ reactions to DWI enforcement and the use of passive sensors in this context.
RESULTS
During the traffic stops in both rounds when the PAS was used, 7% resulted in a DUI arrest. During the stops (in both rounds) when the PAS was not used, the DUI arrest rate was 5%. To a great extent, this reflects the fact that the PAS was used almost twice as often during night stops (93%) than during day stops (51%). Many officers equipped with the PAS were reluctant to use it during daytime stops. Because the PAS was built into a flashlight, officers felt more comfortable using the device at night. In the 810 traffic stops where the PAS was used, 70% were nighttime (6 p.m. – 6 a.m.) stops. In the remaining 1,309 traffic stops where the PAS was not used, 38% were nighttime stops. The DUI arrest rate for night stops when the PAS was used was 10% (54 arrests during 566 stops), and when it was not used, it was also 10% (51 arrests during 490 stops) (see Table 2). It should be noted that the definition used for nighttime (6 p.m. – 6 a.m.) was an attempt to keep equal hours for daytime (6 a.m. – 6 p.m.) and nighttime. Some of the “nighttime” stops may not have been made in darkness, especially during the summer months.
Table 2.
DUI Arrest Rates by PAS Use (Nighttime only)
| Arrests for DUI at night stops (both rounds) | |
|---|---|
| PAS used | 54/566 (10%) |
| PAS not used | 51/490 (10%) |
| No Difference | |
The basic objective of this study was to determine if the use of the PAS by officers during routine traffic stops detected more impaired drivers and increased the DUI arrest rate. Table 3 shows those results. In Round 1, Squad A (with the PAS) had a significantly higher DUI arrest rate than Squad B (without the PAS): 9% vs. 3% (p<.01). In Round 2, Squad A (now without the PAS) also had a significantly higher DUI arrest rate than Squad B (now with the PAS): 7% versus 3% (p<.01). Overall, when both rounds are combined, there were no significant differences in the DUI arrest rate between the two squads: with the PAS vs. without the PAS. There were also no significant differences in the DUI arrest rates during night stops, either with or without the PAS.
Table 3.
Does the use of the PAS increase the DUI arrest rate?
| Round 1 | Round 2 | |||
|---|---|---|---|---|
| Squad A (with PAS) | Squad B | Squad A | Squad B (with PAS) | |
| DUI Arrest | ||||
| Yes | 48 (9%) | 17 (3%) | 37 (7%) | 17 (3%) |
| No | 460 (91%) | 477 (97%) | 501 (93%) | 562 (97%) |
| Total | 508 | 494 | 538 | 579 |
| No Difference (combining both rounds) | ||||
There was one very productive DUI arrest officer in Squad A who accounted for more than half of the Squad A total DUI arrests. When that officer is removed from the analyses, there were no differences in arrest rates for the officers with the PAS versus the officers without the PAS in any round.
However, the PAS seemed to help some officers under some circumstances. In Table 4, the question as to whether the PAS helped officers who had no DUI arrests in Round 2 without the PAS but had arrests with the PAS in Round 1 is addressed. During 76 stops in Round 2 without the PAS, these officers had no DUI arrests. In 19 stops in Round 1 with the PAS, these officers made 2 DUI arrests (11%). This difference was statistically significant (p<.05). There were no significant differences in the proportion of stops that occurred at night in both rounds for these officers, so this was not a factor in this finding. When isolating night stops, however, their DUI arrest rate in 14 night stops with the PAS was 14% compared to no DUI arrests (0%) during 34 night stops without the PAS.
Table 4.
Did the PAS help two officers who had no DUI arrests in Round 2 without the PAS but had arrests with the PAS in Round 1?
| Round 1 (with PAS) | Round 2 (without PAS) | ||
|---|---|---|---|
| Stops | DUI arrests | Stops | DUI arrests |
| 19 | 2 (11%) | 76 | 0 (0%) |
| Yes (p<.05) | |||
The use of the PAS also appeared to help that one very productive DUI arrest officer. In Round 1 with the PAS, that officer made 32 stops that resulted in 31 DUI arrests (97%). In Round 2, that officer made 50 stops without the PAS that resulted in 21 DUI arrests (42%). This difference was statistically significant (p<.01) (Table 5). However, 94% of his stops when using the PAS in Round 1 were at night compared to 68% of his stops in Round 2. Therefore, when isolating night stops only, this officer made 30 DUI arrests out of 30 night stops with the PAS (100%) compared to 20 DUI arrests during 34 night stops without the PAS (59%). This difference was also statistically significant (p<.01).
Table 5.
Did the PAS help one very productive DUI arrest officer?
| Round 1 (with PAS) | Round 2 (without PAS) | ||
|---|---|---|---|
| Stops | DUI arrests | Stops | DUI arrests |
| 32 | 31 (97%) | 50 | 21 (42%) |
| Yes (p<.01) | |||
| Night stops | DUI arrests | Night stops | DUI arrests |
| 30 | 30 (100%) | 34 | 20 (59%) |
| Yes, (p<.01) | |||
Another important question was whether the use of the PAS helped officers detect impaired drivers during daytime stops when they may not suspect a DUI. Table 6 indicates that the PAS apparently did not help. The DUI arrest rate during daytime stops was 1% with and without the PAS.
Table 6.
Did the PAS help officers detect DUI drivers during daytime stops?
| With PAS | Without PAS | ||
|---|---|---|---|
| Daytime Stops | DUI arrests (rate) | Daytime Stops | DUI arrests (rate) |
| 237 | 3 (1%) | 801 | 10 (1%) |
| No difference | |||
Finally, did the PAS help officers who had no DUI arrests when not using the PAS (both rounds combined)? Table 7 indicates the PAS did help these officers somewhat. Without the PAS (both rounds combined), these officers made no DUI arrests in 445 stops (by definition of the subgroup), but they made 10 DUI arrests (2%) in 402 stops when using the PAS (both rounds combined). However, 48% of their stops with the PAS were at night compared to 32% at night without the PAS, which was significant. So, isolating night stops, these officers made no DUI arrests (0%) in 141 night stops without the PAS compared to 10 DUI arrests (5%) during 194 night stops with the PAS. This difference was statistically significant (p<.01).
Table 7.
Did the PAS help officers who had no DUI arrests when not using the PAS (both rounds combined)?
| Without PAS (both rounds) | With PAS (both rounds) | ||
|---|---|---|---|
| Stops | DUI arrests | Stops | DUI arrests |
| 445 | 0 (0%) | 402 | 10 (2%) |
| Yes (p<.01) | |||
| Night stops | DUI arrests | Night stops | DUI arrests |
| 141 | 0 (0%) | 194 | 10 (2%) |
| Yes (p<.01) | |||
The data collected on these 2,119 traffic stops also indicated that there were certain reasons for the traffic stops that increased the potential for impaired driving. Table 8 indicates that, of the 138 stops made for “unsafe maneuvers” (i.e., unsafe lane change, failure to drive right of center, and negligent driving) when officers used the PAS, 22% resulted in a DUI arrest compared to 2% of those stopped for the speeding and 6% for all other reasons combined. This was also statistically significant (p<.01). When examining all 2,119 traffic stops (with and without the PAS), the “unsafe maneuver” traffic stops resulted in a 21% DUI arrest rate, speeding resulted in only a 2% DUI arrest rate, and all other reasons resulted in a 5% DUI arrest rate. These differences were also statistically significant (p<.01).
Table 8.
Are there certain reasons for traffic stops that indicate the potential for arrests for DUI when the PAS is used?
| Reason for Stop | DUI arrest | ||
|---|---|---|---|
| Yes | No | Total | |
| * Unsafe maneuver (unsafe lane change/failure to drive right of center line/negligent driving) | 30 (22%) | 108 (78%) | 138 (100%) |
| Speeding | 9 (2%) | 488 (98%) | 497 (100%) |
| All others (e.g. following too close, failure to yield, etc.) | 26 (6%) | 426 (94%) | 452 (100%) |
Yes (p<.01)
When the PAS was used during 810 stops, 34 drivers were given a PBT, all with positive BAC results (Table 9). In 32 of those instances (94%), the PAS detected alcohol in the drivers. In only 2 instances (6%) did the PAS fail to detect these 34 drinking drivers confirmed by the PBT result. Even when the PBT result was less than .08 BAC (the illegal limit in every state), the PAS detected alcohol in 8 of 9 instances (89%).
Table 9.
How reliable was the PAS in detecting drivers with positive PBT results?
| PAS Result | PBT Result | ||
|---|---|---|---|
| .02–.079 | .08.–.159 | .16+ | |
| Positive Reading | 8 (89%) | 13 (93%) | 11 (100%) |
| Negative Reading | 1 (11%) | 1 (7%) | 0 (0%) |
| Totals | 9 (100%) | 14 (100%) | 11 (100%) |
| Reliable (even at low BACs) | |||
The questionnaire that was administered to the 24 participating officers after the study resulted in some interesting opinions. When asked to freely describe their experience when using the PAS, 9 officers said it was useful, 7 officers said it was impractical because it had to be held too close to the driver, 6 said it was not practical to use at all stops, and 2 said it served as a good flashlight only. When asked if it helped them detect impaired drivers, 13 said “no,” 9 said “yes,” and 2 said “sometimes.” Fifteen officers reported no technical problems using the PAS on traffic stops, whereas 4 reported battery problems and 5 reported other issues such as the rubber switch coming loose and the charging method being a problem.
Only 3 of the 24 officers responding to the survey reported continuing use of the PAS (after the study). The others reported turning in the PAS after the study with no current access to the device. When asked if the PAS helped the officers make more DUI arrests, only 5 said “yes.” Not quite half (10) of the officers reported that drivers were aware or were told that the officer was using a PAS at the stops. If drivers are aware that police are using the PAS during traffic stops, it could have a general deterrent effect on impaired driving [Voas and Layfield, 1983].
DISCUSSION
Did the PAS increase the detection of alcohol-impaired drivers at traffic stops in Anne Arundel County, Maryland? There is some evidence that it did. The PAS showed a positive reading overall in 11% of the nighttime stops. In Round 1, the PAS was positive during 20% of the nighttime stops. The PAS detected 94% of drivers with positive PBT results giving an indication of its reliability. The DUI arrest rate in night stops with the PAS was 10%, and during night stops without the PAS, it was also 10%. Therefore, combining both rounds of traffic stops, the PAS did not increase the DUI arrest rates of Anne Arundel County police officers. There is some evidence, however, that the PAS did help officers who typically do not make DUI arrests. Will the officers continue to use the PAS during routine traffic stops? Probably not and certainly not during the day.
These results are not surprising. Past research has shown that the PAS is most effective at detecting impaired drivers at checkpoints (increases detection rate by about 50%), but not during routine traffic stops (increases detection rate by about 10%). Past studies have also shown that police officers do not like using the PAS at traffic stops [Leaf and Preusser, 1996]. The PAS technology has been in use for more than 20 years, yet the penetration rate into the police community is very low. Substantial improvements in the devices have been made in response to police complaints. Yet only about 4,000 PAS units have been sold to police departments in the past 5 years, and only 8,000 in the past 10 years. Currently, an optimistic estimate of use would be about 2,000 of 500,000 traffic law enforcement officers around the country using the PAS (about a 0.4% penetration rate).
The use of the PAS is probably most beneficial at sobriety checkpoints when police typically only have a few seconds to determine whether a driver is impaired by alcohol. Although moderate social drinkers usually exhibit behavioral signs of intoxication at BACs exceeding .08, many alcoholics and problem drinkers show no outward signs such as slurring their words or fumbling with their driver’s licenses. Thus, it is important for police to use the PAS at checkpoint operations. With current Congressional support and funding for highly publicized national sobriety checkpoint blitzes at least two times a year, the use of the PAS at these operations could serve as a general deterrent to impaired driving [Fell, Lacey, and Voas, 2004] in addition to increasing the detection and arrest of impaired drivers at these checkpoints. The PAS may also be beneficial in the enforcement of zero tolerance laws (no alcohol) for drivers younger than 21.
CONCLUSIONS
The potential effect of passive sensors on detecting, arresting, and deterring alcohol-impaired drivers has not yet been fully realized. Without further research, education on their potential effectiveness, and improvements in the devices so they can be more easily used in the field, support for their use will probably not materialize. It would seem that police could use every reliable tool available to help them enforce the laws. The time they can spend on impaired-driving enforcement is dwindling as competing issues emerge (e.g., homeland security issues). The PAS has the potential to help police officers detect alcohol-impaired drivers, but more acceptance and motivation for their use is needed.
ACKNOWLEDGMENTS
This study was funded by the Maryland Highway Safety Office in Hanover, Maryland. Special thanks go to the Chief of that Office, Vernon F. Betkey, Jr., for his guidance and insight concerning the study design. Lisa Aguila-Lemaster, the Impaired Driving Prevention Coordinator from the Office, provided project oversight and monitoring.
REFERENCES
- Cammisa M, Ferguson S, Wells J. Laboratory evaluation of PAS III sensor with new pump design. Arlington, VA: Insurance Institute for Highway Safety (IIHS); 1996. [Google Scholar]
- Fell JC, Lacey JH, Voas RB. Sobriety checkpoints: Evidence of effectiveness is strong, but use is limited. Traffic Inj Prev. 2004;5(3):220–227. doi: 10.1080/15389580490465247. [DOI] [PubMed] [Google Scholar]
- Ferguson SA. Transportation Research Board 74th Annual Meeting. Washington, DC. Arlington, VA: Insurance Institute for Highway Safety; 1995. Use of passive sensors for alcohol-impaired driving enforcement. [Google Scholar]
- Ferguson SA, Wells JK, Lund AK. The role of passive alcohol sensors in detecting alcohol-impaired drivers at sobriety checkpoints. Alcohol Drugs Driving. 1995;11:23–30. [Google Scholar]
- Fields M, Henricko AR. Passive alcohol sensors—constitutional implications. The Prosecutor. 1986;20(1):45–50. [Google Scholar]
- Fiorentino D. A laboratory study of passive alcohol sensors. In: Mercier-Guyon C, editor. Alcohol, drugs and traffic safety: Proceedings of the 14th International Conference on Alcohol, Drugs, and Traffic Safety. Annency, France: CERMT Centre d’Etudes et de Recherches en Médecine du Trafic.; 1997. pp. 539–545. [Google Scholar]
- Foss RD, Voas RB, Beirness DJ. Using a passive alcohol sensor to detect legally intoxicated drivers. Am J Public Health. 1993;83(4):556–560. doi: 10.2105/ajph.83.4.556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jones IS, Lund AK. Detection of impaired drivers with a passive alcohol sensor. J Police Sc Admin. 1986;14:153–160. [Google Scholar]
- Kiger S, Lestina D, Lund A. Passive alcohol sensors in law enforcement screening for alcohol-impaired drivers. Alcohol Drugs Driving. 1993;9:7–18. [Google Scholar]
- Leaf WA, Preusser DF. Effectiveness of passive sensors (Contract No. DTNH22-93-C-05022)Final Report. Washington, DC: National Highway Traffic Safety Administration; 1996. [Google Scholar]
- Lund AF, Jones IS. Detection of impaired drivers with a passive alcohol sensor. In: Noordzij PC, Roszbach R, editors. Alcohol, drugs and traffic safety ’T86. New York: Excerpta Medica; 1987. pp. 379–382. [Google Scholar]
- Manak JP. Constitutional aspects of the use of passive alcohol screening devices as law enforcement tools for DWI enforcement. The Prosecutor. 1986;19(3) [Google Scholar]
- Moskowitz H, Burns M, Ferguson S. Police officers’ detection of breath odors from alcohol ingestion. Accid Anal Prev. 1999;31(3):175–180. doi: 10.1016/s0001-4575(98)00060-8. [DOI] [PubMed] [Google Scholar]
- Ross HL. Deterring the drinking driver: Legal policy and social control. 2. Lexington, MA: Lexington Books; 1984. [Google Scholar]
- Ross HL. Are DWI sanctions effective? Alcohol Drugs Driving. 1992;8(1):61–69. [Google Scholar]
- Ross HL, Voas RB. The new Philadelphia story: The effects of severe penalties for drunk driving. Washington, DC: AAA Foundation for Traffic Safety; 1989. [Google Scholar]
- Royal D. National Survey of Drinking and Driving Attitudes and Behavior: 2001, Volume I: Summary Report (DOT HS 809 549) Washington, DC: National Highway Traffic Safety Administration, U.S. Department of Transportation; 2003. [Google Scholar]
- SPSS Inc. SPSS Missing Value Analysis 7.5. Chicago, IL: MaryAnn Hill/SPSS Inc.; 1997. [Google Scholar]
- Voas RB, Layfield WA. Creating general deterrence: Can passive sensing help? The Police Chief. 1983;50:56–61. [Google Scholar]
- Voas RB, Romano E, Peck R. Validity of the passive sensor for identifying high BAC drivers at the crash scene. J Stud Alcohol. 2006;67(5):714–721. doi: 10.15288/jsa.2006.67.714. [DOI] [PubMed] [Google Scholar]
- Wells JK, Greene MA, Foss RD, et al. Drinking drivers missed at sobriety checkpoints. J Stud Alcohol. 1997;58(5):513–517. doi: 10.15288/jsa.1997.58.513. [DOI] [PubMed] [Google Scholar]
- Zador P, Krawchuck S, Moore B. Drinking and driving trips, stops by police, and arrests: Analyses of the 1995 national survey of drinking and driving attitudes and behavior (DOT HS 809 184) Washington, DC: National Highway Traffic Safety Administration; 2000. Dec, [Google Scholar]