Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving 


Analysis of Drinking Versus Arrest Locations for DWI Offenders Arrested in Bernalillo County, N.M.

Sandra C. Lapham, M.D., M.P.H.; Betty J. Skipper, Ph.D.; I-Yiin Chang, M.S.; Kerry Barton, M.A.; and Roderick Kennedy, J.D.

The Lovelace Institutes, Institute for Health and Population Research, 1650 University NE, Suite 302, Albuquerque, NM 87102, USA


The purpose of the study was to determine whether convicted DWI offenders arrested following a crash differed in sociodemographic variables or drinking locations from those arrested for other reasons. The study population included 3,650 clients who completed a structured personal interview during the period October 1, 1989 through April 30, 1994 following referral by Bernalillo County, NM, Metropolitan Court. Twelve percent of these offenders were involved in crashes. The independent sociodemographic and arrest variables included: age; ethnicity/race; gender; employment; previous DWI arrest; BAC grouo; vehicle age; city quadrant and times of arrest; drinking location; quadrant of residence; and estimated number of miles driven. About half of the clients (51%) reported having been drinking in a bar. Drinking locations for this subset were clustered in several discrete areas of the city, and arrest locations clustered in approximately the same areas. Among those who drove >0 miles, miles driven ranged from 0 to 19.2, with a mean of 2.6 miles (median = 1.6). Logistic regression was used to determine odds ratios for driving impaired >0 miles, and for crash involvement. Analysis of variance revealed that among those who drove >0 miles, those with BACs of >=.20%, those arrested from 6 am to 12 midnight and persons drinking at bars and restaurants drove fewer miles (persons arrested in the northwest quadrant of Albuquerque and persons who resided outside the city drove more miles). The only variable that was statistically associated with having driven 0 miles was being Native American. Crash involvement was associated with higher BACs, driving vehicles <10 years old, client felt intoxicated, driving between 6 am and 12 midnight, and being arrested in the southeast and southwest quadrants of the city.


Continued study of DWI offender populations unfailingly points to an expanding view of its heterogeneous nature. Yet, studies of drinking locations and drunkenness have revealed some patterns of concern to traffic safety researchers. The incidence of drinking in cars has proved to have an alarmingly high correlation with drunk driving, particularly among younger drivers (Snow, 1988; Snow & Landrum, 1986). One recent study found that significantly more persons involved in traffic crashes had been drinking in unlicensed locations than at bars or restaurants (Land & Stockwell, 1990). This may be attributable to increased police presence at licensed establishments, possibly preventing crashes by arresting intoxicated persons before they drive very far. In the present study we examine sociodemographic factors and miles driven prior to arrest in a population of mixed race and gender, to provide a different perspective to this discussion.

The objectives of the present study were: to explore the relationship between sociodemographic and arrest factors; to estimate the distance driven between the drinking episode and arrest; and to determine whether convicted DWI offenders arrested following a crash differed in sociodemographic variables, drinking locations, and arrest variables from those arrested for other reasons. The study population included persons convicted of a first offense DWI in Bernalillo County, NM, Metropolitan Court, who received screening services at the Lovelace Comprehensive Screening Program (LCSP).


Description of the Screening Program

Individuals convicted of a DWI offense were referred to the LCSP for evaluation by the Metropolitan Court system (Bernalillo County, N.M.). This program provides assessment for the presence and severity of alcohol- and drug-related disorders. The screening process consists of a series of standardized assessments and a structured personal interview and clinical evaluation of the client by a Master's degree level counselor (Lapham et al., 1995). The structured counselor interview obtains information from DWI offenders regarding where they were arrested, where they were drinking prior to their arrests (the street intersections and the quadrant of the city), and the circumstances of the arrest. These data are entered by counselors onto a computerized data file. The study population included all clients who completed a personal interview during the period October 1, 1989 through April 30, 1994.

Statistical Analyses

We used logistic regression to determine odds ratios for driving impaired >0 miles and for crash involvement (the dependent variables were: whether the offender drove >0 miles; and whether the arrest followed a crash). The independent sociodemographic and arrest variables included: age group (<31, 31+); ethnicity/race (non-Hispanic white, Hispanic, Mexican, Native American, other); gender; whether the client was employed at the time of screening; whether the client had a previous DWI arrest; BAC group (<.15, .15-.19, .20+); vehicle type [truck, car, motorcycle); age of vehicle (year at screening minus year of vehicle (5 or fewer, 6 - 9, 10+]; city quadrant of arrest, as reported by the client at the time of the screening interview; drinking location (bar or restaurant, home, private party, other, which includes vehicle, stadium, park, movie, drive-up window, or work); area of residence; whether the arrest involved a crash; perceived intoxication; time of day of arrest (6:01 am to 5 pm; 5:01 pm to 12 midnight, 12:01 am to 6 am); and whether the number of miles driven between the drinking and arrest locations was 0 or >0. In addition, an estimated minimum number of miles driven for each client was calculated using the shortest road distance between the location of drinking street intersections and the location of the arrest street intersections..


Population Demographics

The population consisted of 3,651 persons (59% of the 5,809 persons who completed the personal interview); 2,158 (42%) were dropped due to missing or invalid data. Almost all cases dropped from the investigation were clients who could not recall their drinking or arrest location. About 20% of the study population was female and the majority (almost 60%) were age 30 or under at the time of the LCSP interview (Table 1). Almost 80% were employed.

Table 1
Characteristics of Offenders, N = 3,650

<30 2132 59%
31+ 1493 41%
Male 2871 79%
Female 779 21%
Non-Hispanic White 1389 38%
Hispanic 1544 42%
Mexican 350 10%
Native American 270 7%
Other 93 3%
Age of Vehicle (yrs)
0-5 933 26%
6-9 825 24%
10+ 1795 50%
Year of Referral to Screening
1989-1990 436 12%
1991 1278 35%
1992 1021 28%
1993-1994 913 25%
Previous DWI Arrest
Yes 57 16%
No 3083 84%
Drinking Location
Bar or Restaurant 1858 51%
Home 436 12%
Private Party 317 9%
Other 1039 28%
BAC Group
<.15 1214 35%
.15-.19 1276 37%
.20+ 317 9%
Not Available 641 19%
Quadrant of Arrest
NW 550 15%
SW 470 13%
NE 2014 55%
SE 611 17%
Time of Arrest
00:01 to 06:00 1765 49%
06:01 to 17:00 441 12%
17:01 to 24:00 1423 39%
Miles Driven After Drinking
0 808 22%
>0 2843 78%
Perceived Intoxicated
Yes 2262 62%
No 1387 38%

Vehicle Characteristics and Circumstances of the Arrest

The majority of the clients (64%) reported driving a passenger car, van, or sports utility vehicle; 27% a truck; and <1% a motorcycle. The type of vehicle was unknown for 274 persons (8%). About half the vehicles were ten years old or older, while about one quarter were under 6 years old.

About half of the client population (51%) reported having been drinking in a bar prior to their arrests. An additional 9% reported drinking at a private party; 12% at home; and 28% at other locations. Sixty-one percent of the population reported being stopped by the police for violations such as running a red light, speeding, weaving, or failure to maintain lane boundaries prior to their DWI arrests. About one quarter of the population reported that they were arrested following a moving violation, such as not stopping at a stop sign or failing to signal a turn. Twelve percent were arrested because of crash involvement: 9% in a non-injury crash and 3% in a crash with injuries. About 3% could not recall the circumstances of the arrest or the data were missing for this variable.

About one quarter of the population reported drinking and being arrested at the same intersections. Logistic regression analysis comparing those with 0 and >0 miles driven indicated that the only statistically significant difference, after controlling for other variables in the model, was that Native Americans and persons in the "other race" category were less likely to have driven >0 miles. The minimum miles driven among the 75% of offenders who drove >0 miles ranged from 0.1 to 19.2, with a mean of 3.3 miles (median = 2.4).

Those involved in crashes were less likely to be of Hispanic or Mexican ethnicity, more likely to have higher BACs, and more likely to be driving vehicles under ten years old than persons who were not involved in a crash (Table 2). Persons involved in crashes also were more likely to report to the evaluating counselor that they felt intoxicated at the time of the arrest. Crash involvement was higher among those arrested between midnight and 6:00 pm, compared to other times of day. There also were statistically different crash rates among arrest location quadrants. Finally, persons screened within the last three years of the study period were more likely to be involved in crashes, compared with those screened in earlier years.


Our analysis demonstrated that a substantial percentage of clients reported drinking and being arrested at the same locations. Native Americans and persons of other races were overrepresented in the group of clients who reported driving 0 miles, after controlling for other factors. Since the statistical model involved 30 variables, 1.5 of the statistically significant comparisons would have been expected to occur by chance alone. Therefore, this association may have been due to chance or may represent a real difference in arrest patterns.

Persons who drove >0 miles before being arrested drove an average of about 3 miles; this driving distance was shorter for those with BACs of .20 mg/dl or higher, compared with that of persons with lower BACs. It is reasonable to expect that persons with such high BACs would have drawn attention to themselves by erratic driving. These persons also were more likely to be involved in a crash, thus cutting short their driving distances.

Table 2
Odds Ratios and 95% Confidence Intervals for Crash Involvement

<30 1.20 (0.96, 1.51)
31+ 1.00
Male 1.12 (0.85, 1.48)
Female 1.00
Non-Hispanic White 1.00
Hispanic 0.75 (0.58, 0.97)*
Mexican 0.61 (0.40, 0.94)*
Native American 0.73 (0.46, 1.18)
Other 1.34 (0.71, 2.54)
Age of Vehicle (yrs)
0-5 1.00
6-9 0.87 (0.64, 1.17)
10+ 0.77 (0.59, 1.00)*
Year of Referral to Screening
1989-1990 1.00
1991 1.20 (0.80, 1.79)
1992 1.60 (1.07, 2.41)*
1993-1994 1.64 (1.09, 2.47)*
Previous DWI Arrest
Yes 0.80 (0.59, 1.10)
No 1.00
Drinking Location
Bar or Restaurant 0.72 (0.51, 1.01)
Home 1.00
Private Party 0.96 (0.67, 1.36)
Other 0.84 (0.53, 1.33)
BAC Group
<.15 1.00
.15-.19 1.37 (1.02, 1.83)*
.20+ 2.23 (1.59, 3.12)**
Not Available 2.62 (1.91, 3.59)
Quadrant of Arrest
NW 1.22 (0.89, 1.68)
SW 1.51 (1.06, 2.16)*
NE 1.00
SE 0.69 (0.49, 0.97)*
Time of Arrest
00:01-06:00 1.00
06:01-17:00 0.54 (0.40, 0.73)**
17:01-24:00 0.41 (0.30, 0.56)**
Miles Driven After Drinking
0 1.0
>0 1.19 (0.93, 1.53)
Perceived Intoxicated
Yes 1.30 (1.03, 1.64)*
No 1.00
*p<.05; **p<.001

We found that Hispanics, Mexicans, and persons driving older vehicles were less likely to be involved in crashes. It appears that police are looking more closely at Hispanics' driving patterns and less closely at persons driving newer cars, unless there was a crash involvement. The higher BACs among crash-involved drunk drivers and those driving in the early morning hours has been described previously (Benjamin, 1980; Lang and Stockwell, 1991). The higher crash involvement rates among those referred to screening in the later years of the study can probably be explained by decreased enforcement efforts during these years.


Benjamin, F.B. Alcohol, Drugs, and Traffic Safety. Where do we go from here? Charles C. Thomas, Springfield, ILL 1980.

Lang, E., and Stockwell, T. Drinking locations of drink-drivers: a comparative analysis of accident and nonaccident cases. Accid. Anal. and Prev. 1991:23(6); 573-584.

Lapham, S.C., Skipper, B.J., Owen, J.P., Kleyboecker, K., Teaf, D., Thompson, B., and Simpson, G.L. Alcohol abuse screening instruments: Normative test data collected from a first DWI offender screening program. J. Studies on Alcohol (In Press, 1994).

Lapham, S.C. and Barton, K. Lovelace Comprehensive Screening Program. 1991 Annual Report. Lovelace Medical Foundation, 1650 University, N.E., Suite 302, Albuquerque, N.M. 87102, May 1992.

Olden S. and Gerstein, D.R. Alcohol in America: Taking Action to Prevent Abuse. National Academy Press, Washington, D.C., 1985.

Snow, R.W. Sociodemographic characteristics and drinking locations of convicted drunken drivers. Addictive Behaviors 1988:13(1); 119-122.

Snow, R.W., and Landrum, J.W. Drinking locations and frequency of drunkenness among Mississippi DUI offenders. Am.J.Drug and Alcohol Abuse 1986:12(4); 389-402.