Alcohol abuse is arguably the nation's number one public health problem1, and has an especially powerful impact on youth2. Underage drinking is connected to immediate and long-term health and
behavioral consequences. Driving after drinking is the single strongest contributor to morbidity and mortality for adolescents and young adults1, and most alcohol control measures aimed at youth focus on this behavior3. Youth alcohol use has also been tied to morbidity and mortality from burns, drowning, falls, acute alcohol poisoning, suicide and violent crime4. Overall, annual costs associated with underage drinking are estimated at
Nowhere are the effects of heavy episodic drinking evidenced more than at American colleges. College students binge drink more often than their former high school classmates who do not attend college6. In these settings, where about half of students are under 217, regular use and abuse of alcohol is part of many students' environment. At the average college, two in five students had at least one episode of binge drinking in the two weeks before
This pattern of heavy episodic drinking exists despite a national minimum legal drinking age (MLDA) of 21 in effect since 1984, and the requirement by the Safe and Drug Free Schools Act of 1989 that colleges publish information about laws regulating drug and alcohol use (including the MLDA), spell out the penalties, and periodically evaluate the effectiveness of the institution's policy. The 21- year minimum drinking age is credited with reducing motor vehicle fatalities involving drivers between 18 and 20 by about 800 per year, and drinking among youth by an estimated 10-15 %10-11. Despite these laws, underage drinking on college campuses is widespread and heavy8,9,12,13,14.
Risk factors associated with heavy, episodic, or binge drinking have entailed primarily personal characteristics of students and
residence on campus. Caucasian students, males, and, students 23 or younger are more likely to binge
as are fraternity or sorority house residents15.
Contextual or environmental variables have been associated with binge drinking in general population studies. The use of beer as the alcoholic beverage of choice is related to heavy or hazardous alcohol use16-17.
Environmental factors such as access to alcohol and price have been associated with drinking, heavy drinking, and associated harms18,19. In the case of college students, location of a bar within a mile of campus,
and price of alcohol in the adjoining community are associated with binge drinking9. The association to price has been generally found to be weaker than among high school students and is limited to women and moderate
drinkers19,20. However, these studies did not examine the
way established prices are discounted in college communities where special promotions, free drinks for women, and a set price for
unlimited supplies of drinks are available. Furthermore, because access is generally reported as easy for college students, degree
of ease of access has not been considered as an important factor in binge drinking.
This paper examines the nature of underage drinking among college students and how it differs from the drinking of students who are of legal drinking age. We seek to determine what factors are associated with heavy, episodic, or binge drinking among underage college students. Understanding the relationship of environmental factors, such as access and price, to risky alcohol use may provide insights for the formulation of prevention strategies to counter alcohol abuse and related problems among college students.
This report is based on the 1997 Harvard School of Public Health College Alcohol Study (CAS) Survey. Its methods are described elsewhere in detail8,9,21. The 1997 survey involved 130 of the 140 colleges and universities first surveyed in 1993, a representative sample of 4-year accredited institutions of higher education. The ten colleges not participating in 1997 were unable to provide random samples of student names in time for inclusion. Of the 130 participating institutions, 14 had low response rates (less than 45%), and have been excluded from this analysis. The 116 schools examined are located in 39 states and are a cross-section of American higher education. Each institution provided a random sample of students under CAS guidelines.
Students of non-traditional college age, or 24 and over, were eliminated because they would skew results. These older students are
much more likely to be married, to live off-campus or to attend commuter schools, and to drink less than traditional age.
The 20-page self-administered questionnaire used in 1997 contained questions about alcohol, other substance use, student attitudes and lifestyles, and background characteristics. Standard survey items from comparable studies of youth substance abuse were used whenever possible12.
Measure of Binge Drinking
The CAS gender-specific measure of binge drinking was constructed from responses to gender recency of last drink, and frequency of drinking during the past two weeks. Binge drinking was defined as having "five or more
drinks in a row in the past two weeks" for men and "four or more drinks in a row" for women, based on extensive analyses equating
the risk of alcohol-related health and behavioral problems across gender22.
Frequent binge drinkers were students binged three or more times in the past two weeks, while occasional binge drinkers did so only once or twice. Nonbinge drinkers consumed alcohol in the past year but had not binged in the previous two weeks, while
abstainers had not consumed any alcohol in the past year.
Mailing and Response Rate
Questionnaires were mailed to students between February and April 1997. Three separate mailings
of questionnaires and reminders were timed to avoid spring vacation. Responses were anonymous and participation voluntary. If students returned a separate postcard indicating completing the questionnaire, they were eligible to enter a drawing for a $1,000
cash prize. The overall response rate of students at the 116 colleges was 60%.
Several procedures examined potential non-response bias. We found no significant differences in binge drinking rates of early and
late responders to the survey (x2(1)=0.11, p=0.74). Response rates at individual colleges were not associated the binge drinking rates (r= .06, p=.50).
The study compared the drinking behavior of underage and of-age students. Data analysis was carried out using SAS (version 6.12)23. We used contingency table and chi-square tests to compare the drinking behavior of underage and of-age students. We used odds ratios (OR) to examine the associations between correlates and binge
drinking for underage college students. Multiple logistic regression models were used to provide the OR, adjusted for covariates, as the measures of association between underage binge drinking and correlates such as access to alcohol and price. The response rate of each school was also examined in the logistic regression models to control for the potential confounding due to
Generalized Estimating Equations24-25 (GEE) were used for robust inferences on the estimated coefficients of the multiple logistic regression models using clustered binary outcomes arising from the sampling. The
exchangeable covariance structure seemed the best specification and was specified throughout the GEE.
Characteristics of the Study Sample
The respondents include 7,061 students below 21 years of age, and 4,989 students of ages 21-23. Students who are under 21 years of
age are labeled "underage", and students who are 21 to 23 years of age as "of-age." The two groups differed significantly by gender.
One third (37%) of the underage students were male compared with 43% of the of-age group (x2(1) =52.8, p<0.0001). Because of the
difference in gender among these two groups, we adjusted gender in all of the multiple logistics regression models. We found no
significant racial/ethnic differences between underage and of-age students. Overall, 80% of the students were Caucasian, 8% Hispanic, 5% African American, and 7% Asian/Pacific Islander. Almost all (92%) of the underage students were freshmen or sophomores, while most (80%) of-age students were juniors or seniors.
Frequency and Quantity of Drinking
Fewer underage students drank alcohol, and those who did so drank on fewer occasions than their of-age counterparts. Two in three (63%) underage students reported drinking in the past 30 days, compared to three in four (74%) of-age students
(x2(1)=182, <0.0001). Among those who drank in the past 30 days, underage students were less likely to report drinking on more
than five occasions (39% of underage compared to 42% of of-age; (x2 (1) =10.01, p =0.0015). Both males and females exhibited similar trends.
Although drinking on fewer occasions, underage students consumed more drinks at each occasion (Figure 1). Two in five (42%)
underage students and only 27% of-age students reported usually consuming five or more drinks at each occasion (x2(1) =172, P<0.0001).
Underage students who drank any alcohol in the past year experienced more alcohol-related health and behavioral problems in the past year than of-age students (Table 1). They were significantly more likely to experience 8 of 12 specified alcohol-related problems (OR=1.3-2.0). They were also more likely to experience five or more alcohol related problems (OR=1.2). These analyses were controlled for gender and ethnicity.
Notably, underage students were less likely than of-age students to exhibit one alcohol-related problem (OR=0.5). Among students who
drove cars while at school, underage students were only half as likely as of-age students to drive after drinking.
Venues for Drinking
Attendance at off-campus parties, dormitory social event/parties, fraternity parties and on-campus dances was more often reported by underage students than of-age students (Table 2). Students were also asked to indicate number of drinks they had the last time they attended any of the events. In contrast, of-age students were more likely to attend off-campus bars and on-campus pubs than underage students. Among those attending, underage students more often reported having five or more drinks than of-age students at off-campus parties (x2(1) =50.53, p<0.001) and fraternity parties (x2(1) =15.13, p<0.001). Even at off-campus bars and at an on-campus pub, many underage students reported that they had five or more drinks (33% and 18% respectively).
Access to Alcohol, Price, and Type of Alcohol
More than half (54%) of underage students indicated that it was very easy to obtain alcohol, and 40% said that it was easy. As expected, access was significantly easier among of-age students (x2(2) =971.37, p<0.001). Among the latter, 84% indicated it was
very easy and 15% easy.
Most underage students obtained alcohol from of-age students (80%)*, but one half (50%) obtained it from other underage students. One in four (27 %) obtained it without proof of identity, and one in five (21%) used a fake identity card.
Underage students were more likely than of-age students to get free drinks (25% vs. 5%) and to pay a set price for unlimited drinks or under $1 per drink (32% vs. 11%); whereas of-age students were more likely to pay $1-$3 per drink (63% vs. 34%) and more than $3 per drink (21% vs. 8%) than were underage students.
Beer was most frequently mentioned as the preferred type of alcohol. Only very small differences existed between of-age and underage students in the usual alcoholic beverage consumed. About one half of each group drank beer. Of-age students were more likely to
drink beer than underage students (51% vs. 49%) while underage students are more likely to have no preferred alcoholic drink (10% vs. 8%).
Relationship of Student Background Characteristics to Underage Binge Drinking
In univariate analyses, male students were more likely to binge than female students (50 % vs. 41% ; Or=1.4). African American and
Asian students were less likely to binge drink than Caucasian students (16% and 24% vs. 49%; OR=0.2 and 0.3, respectively). Students at colleges in the West and South were less likely to binge drink than students in the Northeast. (35 % and 42% vs. 46%; OR=0.6 and 0.9, respectively).
Correlates of Underage Binge Drinking
Table 3 presents the correlates of underage binge drinking adjusting for age, gender, ethnicity, geographic region of the school, and response rate. Underage students who live in co-ed dormitories and fraternity or sorority houses are more likely to binge drink (OR=1.7 and 6.2, respectively) than are students who live in single-sex dormitories. Underage fraternity or sorority members were also more likely to binge drink (OR=3.0).
Access to alcohol was also significantly associated with underage students' binge drinking. Students who reported that it was very easy or easy to get alcohol were more likely to binge than students who reported that it was difficult (OR=3.6 and 2.0, respectively).
Students who usually paid less per drink or got their drinks for a set price were more likely to binge. This was not the case for students who usually received their drinks free. Compared to those who paid more than $3, those who paid $1 to $3 and those who paid under $1 or a set fee were more likely to binge (OR=1.5 and OR=2.1 respectively).
Type of drink was related to students' binge drinking. Students who drank beer, liquor, or who had no preferred type of alcohol were more likely to binge (OR=9.5, OR=3.3, and OR=2.6, respectively) than those who drank wine.
College students under the minimum legal drinking age report some important differences in their alcohol use compared with their
legal age peers. Although most underage college students drank alcohol, they were less likely to drink than students over the legal drinking age and they reported drinking on fewer occasions. However, among those who drank, underage students consumed more alcohol per occasion and, perhaps as a result, experienced more alcohol-related problems.
Almost all underage students indicated that it was easy or very easy to obtain alcohol. One in four report that they were able to
purchase alcohol either without proof of identity or with a fake identity card, although most obtained it from older students.
Underage students were more likely to drink in private settings such as dormitory, off-campus, and fraternity parties where access
to alcohol is often unchecked. Although underage students were less likely than of-age students to frequent off-campus bars, one
in three who did reported having five or more drinks. Underage students were also more likely to drink more than their legal age peers in these "supervised" settings.
Although almost all underage students indicated that it was easy to obtain alcohol, those who said that it was very easy were more likely to binge drink. Beyond the easy accessibility of alcohol, cost is an important factor in underage binging. Underage students were more likely to obtain alcohol very cheaply: free, for under $1 per drink, or for a set price for unlimited drinks. Paying a low price per drink (under $1 or a set price for unlimited drinks) was associated with binge drinking. One exception was students who
usually received their drinks free, comprised mostly of nonbingers. Studies on the general population of drinkers have found that higher prices for alcohol beverages lead to significant reductions in drinking and heavy drinking26-27.
The ability of students to obtain alcohol at greatly reduced prices is an important factor in underage binge drinking and deserves
further study. Beer as the usual drink of choice was also strongly associated with binge drinking. One possible explanation is that beer is more likely to be purchased in larger volumes in kegs or pitchers. Other studies have found that beer is disproportionately consumed in hazardous amounts because beer drinkers tend to underestimate the risk of ethanol consumption in this relatively dilute form16,17.
Under the current system of MLDA laws, college alcohol policies, and their enforcement, underage students may be dissuaded from
drinking alcohol to the extent that they drink on fewer occasions and less often in public settings. However, when they do drink they do so in a risky manner, having more per occasion and drinking in private, less controlled situations. Although underage students experience more alcohol-related problems, the only alcohol-related probelm they
experience less often than of-age students is driving after drinking. This may be related to zero tolerance and other drunk driving laws aimed at underage drinkers.
Does this mean we should abandon current MLDA laws as some advocate28? That would mean a likely increase of at
least 800 automobile-related deaths per year in youth between 18 and 213. It would also make alcohol legal for some high school students and thus make it more available to even younger students.
Most current efforts to reduce or prevent high-risk alcohol consumption are aimed at individual students through educational or norm redefinition programs29. Some environmental factors also deserve attention as important areas for intervention. Although we can not expect to make access to alcohol difficult for the underage college population, making it less than
"very easy" may be a start. Since underage students more often drink and binge in private settings, these offer an arena for action. In particular, closer monitoring of fraternity parties, and the enforcement of laws against the sale of alcohol without a license (in the case of parties that charge an admission fee) are two areas to explore. In addition, stricter enforcement of MLDA laws that target underage students who drink in off-campus bars without showing proof of age. Educational campaigns could be aimed at the misguided practice of older students providing alcohol for underage classmates.
Price is another important point of intervention. Areas near college campuses are characterized by high density of alcohol outlets; intense competition for customers; and high- volume, reduced price sales. Eliminating such practices as happy hours, special promotions, and the sale of beer in large volume may also be effective. Further, the common practice of charging an admission fee in private settings entitling guests to an unlimited number of drinks is equivalent to selling alcohol without a license.
Eliminating this practice should be a priority.
These results must be considered in terms of the study's limitations. The CAS is subject to the limitations of self-report surveys. However, such surveys have been considered generally valid in examining alcohol use30-32. Bias may have been introduced through non-response. However, several procedures performed to detect potential bias found none. In the multiple logistic regression models, response rate is not associated with either binge drinking or other covariates. In addition, the binge drinking rates reported in this study are comparable to those found in other national surveys12-14.
This study was supported by the Robert Wood Johnson Foundation. We gratefully acknowledged the assistance of Jeff Hansen for the
preparation of the data and Toben Nelson for carefully reviewing drafts of this manuscript.