Propensity for and correlates of alcohol sales to underage youth.

Abstract

The current study assessed the propensity for alcohol sales to youth in the late 1990s, following increased efforts to reduce youth access to alcohol. Male and female pseudo-underage buyers (i.e., age [greater than or equal to] 21 but judged to appear < 21) attempted to purchase alcohol without age identification at 741 alcohol establishments. One to five purchase attempts were made at each establishment, with 1,065 and 658 attempts at on-premise and off-premise establishments, respectively. The overall sales rate was 26%. Among establishments where more than one purchase attempt was made, 74% sold alcohol to pseudo-underage buyers at least once. The results of this study are encouraging, however, further work is needed to decrease the propensity of illegal alcohol sales to youth.

INTRODUCTION

Alcohol use among adolescents and young adults is high and is related to a wide array of public health problems, including traffic crashes, sexual assaults, suicides, falls, and drownings (Borowsky, Ireland, & Resnick, 2001; Hingson, Heeren, Zakocs, Kopstein, & Wechsler, 2002; Howard, Qiu, & Boekeloo, 2003; National Highway Traffic Safety Administration, 2004). To create sustained reductions in alcohol use and related problems among young people, changes in the environment that promotes alcohol use are needed, rather than only instructing individuals how to resist pressures to drink (Edwards et al., 1994). One part of the environment that contributes to alcohol use among young people is the availability of alcohol (Wagenaar & Perry, 1994). In a recent U.S. nationally representative survey, 83% of 10th graders and 67% of 8th graders reported that alcohol is very easy or fairly easy to get (Johnston, O'Malley, Bachman, & Schulenberg, 2004). Although most underage youth report that individuals 21-years-old and older are the most common sources of alcohol, 18- to 20-year-olds indicate that licensed alcohol establishments are the second most frequently used sources of alcohol (Jones-Webb et al., 1997; Wagenaar et al., 1996).

During the early to mid-1990s, eight studies confirmed that young people could easily purchase alcohol from on-premise (e.g., bars, restaurants) and off-premise (e.g., liquor stores, convenience stores, grocery stores) establishments (Forster, Murray, Wolfson, & Wagenaar, 1995; Forster et al., 1994; Lewis et al., 1996; O'Leary, Gorman, & Speer, 1994; Preusser & Williams, 1992; Preusser, Williams, & Weinstein, 1994; Schofield, Weeks, & Sanson-Fisher, 1994; Vaucher, Rehm, Benvenuti, & Muller, 1995). Half of the studies assessed the likelihood of illegal alcohol sales to underage youth by conducting purchase attempts with pseudo-underage buyers (persons 21-years-old or older judged to appear 18- or 19-years-old) and half conducted purchase attempts with actual underage buyers. Most studies sampled a small number of outlets in one or two jurisdictions; however, Forster et al. (1994; 1995) conducted purchase attempts in over 20 communities. Overall purchase rates across all the studies ranged from 45% to 88%.

In addition to assessing purchase rates, Forster et al. (1994; 1995) along with Wolfson et al. (1996a; 1996b) assessed whether specific server, establishment or buyer characteristics were associated with the likelihood of an illegal alcohol sale to pseudo-underage buyers. Servers who appeared younger than age 31 were more likely than older severs to sell alcohol to pseudo-underage buyers in on-premise outlets (Forster et al., 1995; Wolfson et al., 1996a). In off-premise establishments, male clerks were more likely than female clerks to sell alcohol to pseudo-underage buyers (Forster et al., 1995); however, female servers were more likely to sell to pseudo-underage buyers in on-premise establishments (Wolfson et al., 1996a). On-premise establishments with signs posted warning against serving minors were more likely to sell to the buyers than on-premise establishments without such signage; off-premise establishments affiliated with a chain of stores were more likely to sell than off-premise establishments without such affiliation (Wolfson et al., 1996a). In addition, liquor stores (as opposed to bars or convenience/grocery stores), and restaurants (as opposed to clubs/bars) were more likely to sell alcohol to pseudo-underage buyers (Forster et al., 1995; Wolfson et al., 1996a; 1996b).

During the early- to mid-1990s when most of these studies were conducted, laws prohibiting alcohol sales to minors were not well enforced (Wagenaar & Wolfson, 1994). However, the social environment surrounding underage drinking began to change later in the 1990s. Funding to address the issue of underage drinking was distributed to sites across the United States by the Robert Wood Johnson Foundation and the Office for Juvenile Justice and Delinquency Prevention in the Department of Justice (Horgan, Skwara, & Stricker, 2001). National funds allotted to combat underage access to alcohol amount to over tens of millions of dollars per year (Horgan, Skwara, & Stricker, 2001).

Only one U.S. study assessed propensity for illegal alcohol sales to minors since efforts to increase enforcement of underage drinking laws began. Freisthler, Gruenewald, Treno, and Lee (2003) used pseudo-underage buyers to assess likelihood of sales to underage youth at a census of off-premise establishments in two Sacramento, CA neighborhoods (50 and 28 establishments, respectively) and an additional 68 randomly selected establishments throughout the city (total establishments = 146). Buyers made two purchase attempts at each establishment. As with many of the previous studies, Freisthler et al. (2003) used only female buyers. The overall sales rate was 39%. Sales were more likely in neighborhoods with higher percentages of Hispanic residents and in areas with higher population densities, but no association was found between clerk characteristics and purchase attempt outcomes.

The current study has two objectives. First, we provide an estimate of propensity for illegal alcohol sales to underage youth during the late 1990s for both on- and off-premise establishments. Second, we build upon the extant research literature and examine relationships between a variety of buyer, seller and establishment characteristics and the propensity to sell to youth, using data from multiple purchase attempts at a large sample of 741 retail alcohol establishments.

METHODS

Our data come from the Complying with the Minimum Drinking Age (CMDA) Project, a five-year intervention trial using a multicommunity time-series design. The goal of CMDA was to evaluate effects of two interventions on propensity for alcohol sales to underage youth in four geographic areas, including: 1) two large metropolitan cities (seven neighborhoods in one city and 10 neighborhoods in the other city), and 2) two suburban areas (the first area composed of 10 incorporated cities and the second composed of eight incorporated cities). Community populations ranged from 2,364 to 382,618 (median = 22,435) and the percent of each community classified as white ranged from 53.5% to 96.6% (median = 90.4%; U.S. Census Bureau, 2003). Data used here consist of repeated pseudo-underage purchase attempts at 741 licensed alcohol establishments conducted at baseline from September 1997 through December 1998.

Alcohol Establishments

We obtained fists of all "intoxicating liquor" establishments in each city (including name, address, phone number, and liquor license type) from the state Liquor Control Division. We obtained fists of all low-alcohol (3.2%) beer establishments from each community. We used a census of all licensed alcohol establishments in the suburban study communities (i.e., no sampling of establishments) and approximately half of all licensed establishments in the metropolitan cities, including 455 on-premise establishments, 286 off-premise establishments; of these establishments, 389 had full liquor licenses and 352 had beer/wine or 3.2 beer-only licenses. The number of establishments per community ranged from 1 to 200 (mean = 37).

Alcohol Purchase Attempts

Thirty-two male and female individuals who were at least 21 years of age, but looked 17- to 20-years-old, attempted to purchase alcohol without showing age identification. We recruited potential buyers through advertisements in student and local newspapers. So as not to alert alcohol establishments of our research project, the advertisements did not refer directly to making alcohol purchase attempts at licensed establishments. We assessed apparent age of buyers using panels of eight to ten individuals not directly associated with the project, including individuals with experience in the hospitality industry. Each prospective buyer appeared before the panel and talked for a brief period while panel members recorded an estimated age. Mean estimated age for the buyers was 18.8 for buyers visiting onpremise establishments and 18.9 for buyers visiting off-premise establishments, with a perceived age range from 17 to 20. Mean actual age of the buyers was 22.6 years with a range from 21.1 to 24.8. All buyers participated in a training session to learn the study protocol. Buyers dressed casually and wore little or no make-up, jewelry, facial hair or hats. A similar protocol has been used in previous studies (Forster et al., 1994; 1995; Wolfson et al., 1996a; 1996b).

A same-sex team, including a pseudo-underage buyer and an accompanying observer who also appeared to be underage, conducted each purchase attempt. For off-premise establishments, buyers entered the establishment alone, carrying only money for the purchase attempt. Age identification was left in the car. Buyers attempted to purchase a 6-pack or quart of beer (Budweiser, Coors or Miller). We instructed buyers to go to the shortest line in the establishment. If asked how old they were, buyers answered honestly. If asked for identification, buyers stated they did not have identification on them. If the sale was refused or the seller requested buyers to get their identification, buyers exited the establishment. The observer remained in the car out of view of the seller. For on-premise establishments, the protocol was similar, although observers accompanied buyers into the establishment. Buyers ordered a Budweiser, Coors or Miller beer, whichever was cheapest, and observers ordered a soda. Buyers gave similar responses to requests for identification as noted for off-premise establishments. If served, observers drank their soda but buyers did not consume the beer. Both buyers and observers completed data collection forms in the car immediately after each purchase attempt.

Buyers conducted pseudo-underage purchase attempts approximately every other week. From September 1997 through April 1998, purchase attempts were conducted in a random sample of 60 to 80 establishments on each purchase attempt weekend, with establishments re-eligible for random selection each purchase attempt weekend. In May of 1998, we altered the selection protocol to ensure that purchase attempts were regularly conducted at each individual establishment. We constructed ten cohorts, each composed of a random sample of 80 to 100 establishments. Buyers visited each establishment approximately every 20 weeks. Within each cohort, buyers visited establishments in a specified order, with a random starting point in the pattern chosen during each purchase attempt weekend to vary time of purchase attempts at a given establishment. We randomly rotated buyers so that they did not make purchase attempts at the same establishments during consecutive purchase attempt weekends. Buyers made purchase attempts on Friday and Saturday evenings between 3:00 p.m. and 10:30 p.m. The number of purchase attempts per establishment ranged from one to five, with a total of 1,065 purchase attempts at on-premise establishments and 658 at off-premise establishments.

Variables

The dependent variable was purchase attempt outcome, recorded by buyers as: 1) purchased with no questions, 2) purchased after stating age, 3) purchased after stating age identification not with buyer, 4) purchased after explaining age identification in car, or 5) refused sale; this variable was later dichotomized for analyses as sale versus no sale. Independent variables included characteristics of servers, establishments, purchase attempts, and buyers. Buyers recorded server characteristics, including seller gender, and perceived age of seller (less than 21, between 21 and 30, and 31 or older, and later dichotomized into less than 31 versus 31 or older).

We obtained two establishment characteristics from licensing information: 1) type of establishment (on-premise, off-premise); and 2) license type (liquor, wine/strong beer, wine/3.2 beer or 3.2 beer only). Establishment characteristics reported by buyers and observers included: type of business (recorded initially as one of 21 categories and later divided into four categories for analyses--bar or restaurant-with-bar, restaurant, athletic venue, or other), type of neighborhood (recorded as five categories and later dichotomized as primarily residential versus not); proximity to other alcohol retail establishments (recorded as four categories and later dichotomized into another alcohol establishment next door versus not); maintenance of establishment (recorded as three categories and later dichotomized into good versus not), presence of signs warning against minor entrance and minor sales (present versus not), and tobacco sales (yes, no, don't know).

In addition, for off-premise establishments, buyers indicated the number of cash registers present (recorded as a continuous variable and later dichotomized as one versus more than one), and busyness of the establishment (no one in line versus one or more persons in line). In on-premise establishments, observers assessed the busyness of the establishment (no seating and many standing, no seating and few standing, or seating available), gender of customers (no customers, mostly male, mostly female, or even mix), and customers' ages (first recorded as three categories and later dichotomized as mostly 35 or under versus mostly older than 35 or even mix). We collapsed some variables to fewer categories to improve their reliability (Erickson, Toomey, & Wagenaar, 2003).

Purchase attempt characteristics included day of purchase attempt (Friday, Saturday), time of purchase attempt (recorded as actual time but dichotomized to before 6:00 pm versus 6:00 pm or after), and timeline point of purchase (weekend of purchase attempt across the 16-month data collection period). Buyer characteristics included gender and age (measured as a continuous variable).

Analyses

We first analyzed independent variables individually to determine whether they were related to the purchase attempt outcome (these analyses will be referred to as bivariate analyses from this point forward). We stratified analyses by on-premise and off-premise establishments because previous studies have found differences between these types of establishments (Forster et al., 1994; 1995; Wolfson et al., 1996a; 1996b). In the multivariate analyses, we included independent variables that were statistically significant at p [less than or equal to] 0.2 in initial bivariate analyses. We used backwards-stepwise regression analyses to identify the most parsimonious final on-premise and off-premise multivariate models. We retained all variables statistically significant at p [less than or equal to] 0.05 in the final models.

We conducted bivariate and multivariate analyses using SAS PROC MIXED (SAS Institute, 1999). Although PROC MIXED was designed for normally distributed outcome variables, simulation studies have demonstrated that this program is sufficiently robust for analyses using dichotomous outcome variables, provided samples sizes are sufficiently large (Murray, 1998; Hannah & Murray, 1996). We confirmed PROC MIXED bivariate results using GLIMMIX, a macro specifically designed for dichotomous outcomes; we found no substantive differences. Because of the complexity of the multivariate models, however, the GLIMMIX program did not converge.

We accounted for repeated purchase attempts at the same establishment in all analyses since purchase attempt results in the same establishment were likely to be correlated. Because we did not make purchase attempts at the same establishments at regularly spaced intervals throughout the study period, we specified a spatial exponential correlation structure to account for temporally decaying correlation effects. We also specified three random effects- individual buyer (to control for differences in individual buyer's ability to purchase alcohol--i.e., buyer effect), community, and establishment nested within community (to control for similarity of establishments located within the same communities). Correlation due to establishments nested within communities results in parameter estimates with a common variance component and must be addressed to prevent underestimation of standard errors (Murray, Clark, & Wagenaar, 2000).

RESULTS

The overall sales rate to the pseudo-underage buyers was 26%. During the first 30 weeks of the data collection period, the sales rate was 29% and during the second 30 weeks, the sales rate was 23%. Because of this, a temporal trend term was included in the final multivariate analysis. Across communities, the overall sales rates ranged from 0% to 47%. Of the 552 establishments where more than one purchase attempt was made, 74% of them sold alcohol at least once.

For on-premise establishments, 14 variables met the statistical criteria of p [less than or equal to] 0.2 in the bivariate analyses. Female and younger appearing servers were more likely than male and older appearing servers to sell alcohol to the buyers. Establishments that are solely restaurants (without bars) had a higher propensity to sell than bars, restaurants-with-bars and other types of establishments. Establishments that have less than a full liquor license (i.e., beer or beer/wine only), are located in non-residential areas, are next door to other alcohol establishments, have good maintenance, do not post warnhag signs, do not sell tobacco, or that had mostly male or an even mix of male and female customers had a higher propensity to sell alcohol to underage buyers. Purchases made before 6:00 pm were more likely to result in a sale than those made later. Finally, linear trends over time for purchase attempt weekend and buyer age were statistically significant. The buy rate trended downward over time and olderappearing buyers were more likely to purchase alcohol.

In the backwards-stepwise regression analyses, only four variables remained statistically significant (p [less than or equal to] 0.05) for on-premise establishments (Table 1). Restaurants-without-bars had a higher propensity to sell alcohol to underage youth than bars, restaurants-with-bars and other types of establishments. Establishments with full liquor licenses were less likely to sell than those with beer-only or beer-wine licenses, and purchases made before 6:00 pm were more likely than those made later to result in a sale. The linear downward trend in the sales rate also remained in the final model.

For off-premise establishments, only three variables met the p [less than or equal to] 0.2 statistical significance criteria in the initial bivariate analyses. None of the server or buyer characteristics were related to likelihood of an alcohol sale. Establishments with 3.2 beer-only licenses were more likely than those with full liquor licenses to sell to underage youth. Establishments without warning signs regarding alcohol sales to minors were more likely than establishments with such signage to make an alcohol sale, as were establishments where there were other customers in line behind the buyer. In multivariate analyses, only posting warning signs and busyness of the establishment were associated with purchase attempt outcome (Table 1).

DISCUSSION

Results from this study suggest that significant progress has been made in reducing alcohol sales to minors over the past decade. Sales rates to underage or pseudo-underage buyers in the early 1990s ranged from 45% to 88% (Forster et al., 1994; 1995; Lewis et al., 1996; O'Leary et al., 1994; Preusser & Williams, 1992; Preusser et al., 1994; Schofield et al., 1994; Vaucher et al., 1995). In our study, we found that the sales rate to pseudo-underage buyers in 1997 was 29% and by the end of 1998 had declined to 23%. Despite this decline, however, individuals who appeared to be age 17 to 20 were still able to purchase alcohol without showing age identification in almost one out of four purchase attempts, and 74% of the outlets sold to youth in at least one of multiple attempts.

Multivariate analyses revealed several establishment characteristics related to likelihood of illegal alcohol sales to minors. For onpremise establishments, restaurants with a distinct bar area were less likely than restaurants without distinct bars to sell alcohol to underage youth (15% versus 40%). Establishments with full liquor licenses were less likely than establishments with wine/3.2 beer licenses to serve alcohol to minors (19% versus 50%). Establishments with distinct bars and flail alcohol licenses may sell larger quantities of alcohol and as a result may be more aware of issues regarding responsible service of alcohol. Purchase attempts made before 6:00 p.m. were slightly more likely to result in a sale than those made after 6:00 p.m. (26% versus 21%). Law enforcement checks are a promising approach to preventing sales to minors (Preusser et al., 1994; Scribner & Cohen, 2001). More intensive enforcement may need to be targeted towards restaurants without distinct bars, establishments with wine/3.2 beer licenses, and during earlier times of the evening.

For off-premise establishments, the only establishment characteristics that remained in the final model were sign postings and the number of customers in line. Establishments that had posted signs stating "no sales to minors" were less likely than establishments without posted signs to sell alcohol to pseudo-underage youth (23% versus 33%). While this relationship might be expected, a previous study found establishments posting warning signs were more likely to sell to youth (Wolfson et al., 1996a). Given the present results, "no sales to minors" signs may remind staff that their establishment does not support sales to underage youth and may also serve as a support for them as they refuse alcohol sales (i.e., can point to the sign and say that they are just following the rules). Establishments that post "no sales to minors" signs may also implement other establishment policies, such as offering staff incentives for checking age identification and training staff on how ask for and check age identification, to help prevent sales to minors. Establishments with customers in line behind the buyer were more likely to sell alcohol to underage youth than establishments with no customers in line (32% versus 24%). Staff at busier establishments may feel pressured to respond to customers quickly and may not take the time to check for identification. Server and manager training focused on consistent implementation of age identification procedures and monitoring by management may be needed to ensure responsible alcohol sales even when establishments are busy.

Unlike previous studies, we did not find an association between the likelihood of illegal sales to minors and server characteristics. Neither the gender of the server or the perceived age of the server was related to purchase attempt outcomes here, in contrast to Wolfson et al. (1996a) and Forster et al. (1995) who found age and gender effects.

This study has several limitations. First, we conducted the study in one large Midwestern metropolitan area. But, our observed sales rate was similar to the sales rate recently found by Freisthler et al. (2003) in one city in California. A second potential limitation is that we used pseudo-underage buyers rather than individuals who are actually underage; however, a panel assessed all buyers and only those consistently rated to look age 17 to age 20 were employed as buyers. As with most studies, we also included only a limited number of potential predictors of sales to youth. Future studies should assess other potential predictors of likelihood of sales to minors such as whether staff have participated in responsible server training, types of establishment policies, and neighborhood characteristics.

The first studies assessing propensity for illegal alcohol sales were conducted during the early 1990s. During the last half of the 1990s, considerable resources were directed towards preventing underage alcohol use in the United States, with much of the intervention focusing on changing environmental risk factors such as youth access to alcohol. A trend in decreasing sales rates to underage persons was expected following this increase in resources. This study contributes to our knowledge about the likelihood of youth being able to purchase alcohol following these efforts. We advanced the methods used in previous studies by using a large sample with multiple repeated measures, in contrast to previous studies that used relatively small samples of establishments in one or two communities and only one or two purchase attempts per establishment. Repeated measures increase statistical power and permit analyses of variance over time in the propensity of establishments to sell alcohol to youth.

Results of this study are encouraging. The overall sales rate for establishments in our study was 26%--lower than sales rates observed during the early 1990s. During the period of this study, the sales rate dropped from 29% during the first 30 weeks to 23% during the last 30 weeks. Although encouraging, young appearing buyers were still able to purchase alcohol without showing age identification in almost one out of four purchase attempts, and perhaps more importantly, 74% of establishments sold to youth at least once across multiple attempts. Achieving significant reductions in teen drinking and morbidity/mortality indicators is likely to require significant further reductions in observed rates of alcohol sales to youth.

Conclusion

Our findings can directly benefit the efforts of law enforcement professionals and owners and managers of alcohol establishments interested in reducing alcohol sales to minors, as well as public health researchers and policy makers Based on our findings, police may need to conduct more enforcement towards certain types of establishments and during earlier times of the evening. Additionally, alcohol establishments should post "no sales to minors" signs, and server and manager training should focus on consistent implementation of age identification procedures and monitoring by management, particularly when establishments are busy. Policy makers can use these finding to support legislation to mandate increased enforcement and more responsible alcohol beverage serviced practices. Finally, our methods using a large sample with multiple repeated measures advance the research in this area.

Author Note

Heather Britt and Traci L. Toomey, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis; William Dunsmuir, School of Mathematics, University of New South Wales, Sydney, Australia; Alexander C. Wagenaar, Department of Epidemiology and Health Policy Research, College of Medicine, University of Florida, Gainesville.

The study was funded by the National Institute on Alcohol Abuse and Alcoholism under grant R01 AA11258, Alexander C. Wagenaar, Principal Investigator. We also acknowledge the contributions of Nicole Cina and field data collection staff, Linda Fletcher, William Baker and Peter Hannan.

Correspondence concerning this article should be addressed to Traci L. Toomey, Division of Epidemiology and Community Health, University of Minnesota, 1300 S. Second St., Suite 300, Minneapolis, Minnesota 55454. E-mail:[e-mail]

References

Borowsky, I. W., Ireland, M., & Resnick, M. D. (2001). Adolescent suicide attempts: Risks and protectors. Pediatrics, 107, 485-493.

Edwards, G., Anderson, R, Babor, T. F., Casswell, S., Ferrence, R., Giesbrecht, N. et al., (1994). Alcohol policy and the public good. New York: Oxford University Press.

Erickson, D. J., Toomey, T. L., & Wagenaar, A. C. (2003, June). Expanding beyond traditional survey methods: Collecting reliable observational data for prevention trials. Paper presented at the 8th Annual Meeting of the Society for Prevention Research, Montreal, Quebec.

Forster, J. L., McGovern, R G., Wagenaar, A. C., Wolfson, M., Perry, C. L., & Anstine, P. S. (1994). The ability of young people to purchase alcohol without age identification in northeastern Minnesota, USA. Addiction, 89, 699-705.

Forster, J. L., Murray, D. M., Wolfson, M., & Wagenaar, A. C. (1995). Commercial availability of alcohol to young people: Results of alcohol purchase attempts. Preventive Medicine, 24, 342-347.

Freisthler, B., Gruenewald, P. J., Treno, A. J., & Lee, J. (2003). Evaluating alcohol access and the alcohol environment in neighborhood areas. Alcoholism- Clinical and Experimental Research, 27, 477-484.

Hannan, P. J., & Murray, D. M. (1996). Gauss or Bernoulli--A Monte Carlo comparison of the performance of the linear mixed-model and the logistic mixed-model analyses in simulated community trials with a dichotomous outcome variable at the individual level. Evaluation Review, 3, 338-352.

Hingson, R. W., Heeren, T., Zakocs, R. C., Kopstein, A., & Wechsler, H. (2002). Magnitude of alcohol-related mortality and morbidity among U.S. college students ages 18-24. Journal of Studies on Alcohol, 63, 136-144.

Horgan, C., Skwara, K. C., & Strickler, G. (2001, February). Substance abuse: The nation's number one health problem. Princeton, NJ: Schneider Institute for Health Policy, Brandeis University.

Howard, D., Qiu, Y, & Boekeloo, B. (2003). Personal and social contextual correlates of adolescent dating violence. Journal of Adolescent Health, 33(1), 9-17.

Johnston, L. D., O'Malley, P. M., Bachman, J. G., & Schulenberg, J. E. (2004). Monitoring the Future national survey results on drug use, 1975-2003. Volume I: Secondary school students (NIH Publication No. 04-5507). Bethesda, ME): National Institute on Drug Abuse.

Jones-Webb, R., Toomey, T., Miner, K., Wagenaar, A. C., Wolfson, M., & Poon, R. (1997). Why and in what context adolescents obtain alcohol from adults: A pilot study. Substance Use & Misuse, 32, 219-228.

Lewis, R. K., Paine-Andrews, A., Fawcett, S. B., Francisco, V. T., Richter, K. P., Copple, B., et al. (1996). Evaluating the effects of a community coalition's efforts to reduce illegal sales of alcohol and tobacco products to minors. Journal of Community Health, 21, 429-436.

Murray, D. M. (1998). Design and analysis of group-randomized trials. New York: Oxford University Press.

Murray, D. M., Clark, M. H., & Wagenaar, A. C. (2000). Intraclass correlations from a community-based alcohol prevention study: The effect of repeat observations on the same communities. Journal of Studies on Alcohol, 61, 881-890.

National Highway Traffic Safety Administration (2004). Traffic Safety Facts 2002 (Report No. DOT HS 809 620). Washington, DC: U.S. Department of Transportation.

O'Leary, D., Gorman, D. M., & Speer, P. W. (1994). The sale of alcoholic beverages to minors. Public Health Reports, 109, 816-818.

Preusser, D. F., & Williams, A. F. (1992). Sales of alcohol to underage purchasers in three New York counties and Washington, D.C. Journal of Public Health Policy, 13, 306-317.

Preusser, D. F., Williams, A. F., & Weinstein, H. B. (1994). Policing underage alcohol sales. Journal of Safety Research, 25, 127-133.

SAS Institute (1999). SAS/STAT Version 8. Cary, NC: SAS Institute.

Schofield, M. J., Weeks, C., & Sanson-Fisher, R. (1994). Alcohol sales to minors: A surrogate study. Preventive Medicine, 23, 827-831.

Scribner, R., & Cohen, D. (2001). The effect of enforcement on merchant compliance with the minimum legal drinking age law. Journal of Drug Issues, 31, 857-866.

U.S. Census Bureau. (2003) American Fact Finder [Web Page]. Retrieved June 2003, from http://factfinder.census.gov/home/saff/ main.html?_lang=en

Vaucher, S., Rehm, J., Benvenuti, J., & Muller, R. (1995). Young teenagers and access to alcohol in a Swiss canton: Evidence from observational testing and from a telephone survey. Addiction, 12, 1619-1625.

Wagenaar, A. C., & Perry, C. L. (1994). Community strategies for the reduction of youth drinking: Theory and application. Journal of Research on Adolescence, 4, 319-345.

Wagenaar, A. C., Toomey, T. L., Murray, D. M., Short, B. J., Wolfson, M., & Jones-Webb, R. (1996). Sources of alcohol for underage drinkers. Journal of Studies on Alcohol, 57, 325-333.

Wagenaar, A. C., & Wolfson, M. (1994). Enforcement of the legal minimum drinking age in the United States. Journal of Public Health Policy, 15, 37-53.

Wolfson, M., Toomey, T. L., Forster, J. L., Wagenaar, A. C., McGovern, P. G., & Perry, C. L. (1996a). Characteristics, policies, and practices of alcohol outlets and sales to youth. Journal of Studies on Alcohol, 57, 670-674.

Wolfson, M., Toomey, T. L., Murray, D. M., Forster, J. L., Short, B. J., & Wagenaar, A. C. (1996b). Alcohol outlet policies and practices concerning sales to underage people. Addiction, 4, 589-602.

Heather Britt and Traci L. Toomey

University of Minnesota, Minneapolis

William Dunsmuir

University of New South Wales, Sydney, Australia

Alexander C. Wagenaar

University of Florida, Gainesville

Table 1. Final Model Multivariate Results Controlling for Buyer Effect,
Community, Nesting of Outlets and Repeated Purchase Attempts

                                      On-premise

                                         LS
                                N     Means (a)   SE (b)
Type of Business
  Bar/Restaurant                660     0.15      0.031
  Restaurant                    323     0.40      0.039
  Athletic venue                 60     0.27      0.058
  Other                          22     0.20      0.091
License Type
  liquor                        622     0.19      0.030
  Wine/strong beer              165     0.28      0.047
  Wine/3.2 beer                  66     0.50      0.064
  3.2 beer only                 172     0.27      0.043
Minor Sales Sign
  Sign present
  No sign present
Busyness (off-premise)
  [greater than or equal to]
    1 person in line
  No one in line
Time of Day
  Before 6 pm                   518     0.26      0.029
  6 pm or after                 547     0.21      0.028
Purchase Attempt Weekend
  Linear trend                 1065      --         --

                                     On-premise

                                 F-test (c)       p
Type of Business
  Bar/Restaurant               10.66 (3,574)   <0.0001
  Restaurant
  Athletic venue
  Other
License Type
  liquor                        7.33 (3,574)   <0.0001
  Wine/strong beer
  Wine/3.2 beer
  3.2 beer only
Minor Sales Sign
  Sign present
  No sign present
Busyness (off-premise)
  [greater than or equal to]
    1 person in line
  No one in line
Time of Day
  Before 6 pm                   5.01 (1,574)     0.026
  6 pm or after
Purchase Attempt Weekend
  Linear trend                 10.10 (1,574)     0.002

                                     Off-premise

                                        LS
                                N    Means (a)   SE (b)
Type of Business
  Bar/Restaurant
  Restaurant
  Athletic venue
  Other
License Type
  liquor
  Wine/strong beer
  Wine/3.2 beer
  3.2 beer only
Minor Sales Sign
  Sign present                 402     0.23       0.034
  No sign present              256     0.33       0.040
Busyness (off-premise)
  [greater than or equal to]   227     0.32       0.039
    1 person in line
  No one in line               431     0.24       0.034
Time of Day
  Before 6 pm
  6 pm or after
Purchase Attempt Weekend
  Linear trend

                                   Off-premise

                                 F-test (c)     p
Type of Business
  Bar/Restaurant
  Restaurant
  Athletic venue
  Other
License Type
  liquor
  Wine/strong beer
  Wine/3.2 beer
  3.2 beer only
Minor Sales Sign
  Sign present                 6.64 (1,339)   0.012
  No sign present
Busyness (off-premise)
  [greater than or equal to]   3.90 (1,339)   0.049
    1 person in line
  No one in line
Time of Day
  Before 6 pm
  6 pm or after
Purchase Attempt Weekend
  Linear trend

(a) Least-square means

(b) Standard error

(c) F statistic, (numerator df, denominator df)