726 JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / SEPTEMBER 2009
Persistence of Heavy Drinking and Ensuing Consequences
at Heavy Drinking Colleges*
TOBEN F. NELSON, SC.D.,† ZIMING XUAN, S.M.,† HANG LEE, PH.D.,† ELISSA R. WEITZMAN, SC.D.,†
AND HENRY WECHSLER, PH.D.†
Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1300 South Second Street, Suite 300,
Minneapolis, Minnesota 55454
ABSTRACT. Objective: The purpose of this study was to examine
drinking levels, related harms, and secondhand effects of alcohol use at
heavy drinking colleges between 1993 and 2005 at colleges with high
levels of drinking in 1993. Method: Students attending 18 colleges
with high levels of heavy episodic drinking (50% of students or more)
from the 1993 Harvard School of Public Health College Alcohol Study
were surveyed in 2005 (n = 4,518). Data collected through mailed and
Web-based questionnaires were compared with responses from students
at the same schools in 1993, 1997, 1999, and 2001 (N = 13,254) using
time trend analyses. Results: Overall, levels of alcohol consumption,
experience of problems, and levels of secondhand effects remained high
among students attending heavy drinking colleges. More than four of five
students at these schools drank alcohol (range: 85%-88%), and more than
H EAVY ALCOHOL USE IS A MAJOR public health
problem among college students. Two in five students
attending 4-year colleges in the United States engage in
heavy episodic drinking (O’Malley and Johnston, 2002;
Wechsler and Nelson, 2001; Wechsler et al., 1994, 1998,
2000, 2002b), and many who drink at these levels experience
serious negative academic, social, and health consequences
as a result (Perkins, 2002; Wechsler and Nelson, 2006;
Wechsler et al., 2002b). Drinking also has consequences for
other students (e.g., disruption of sleep or study; property
damage; and verbal, physical, or sexual violence [Wechsler
et al., 1995b]), and residents of neighborhoods near colleges
(e.g., noise disruptions, property damage, and police visits
[Wechsler et al., 2002a]).
In prior studies, heavy episodic drinking varied widely
by college, but these drinking levels remained consistent
within individual colleges in subsequent surveys (Wechsler
Received: October 14, 2008. Revision: June 1, 2009.
*This research was supported by Robert Wood Johnson Foundation grants
to Henry Wechsler.
†Correspondence may be sent to Toben F. Nelson at the above address or
via email at: tfnelson@umn.edu. Ziming Xuan and Henry Wechsler are with
the Department of Society, Human Development, and Health, Harvard School
of Public Health, Boston, MA. Hang Lee is with the Biostatistics Center,
Massachusetts General Hospital and Harvard Medical School, Boston, MA.
Elissa R. Weitzman is with the Division of Adolescent Medicine, Children’s
Hospital Boston, and the Department of Pediatrics, Harvard Medical School,
Boston, MA.
726
half engaged in heavy episodic drinking (range: 53%-58%). The stability
of drinking behavior occurred among subgroups of students as well. The
few statistically significant changes occurred mainly between 1993 and
1997. A decline in driving after any drinking between 1997 and 2005
was observed, but no similar decline was found in two other measures
of drinking and driving. Conclusions: Heavy drinking and associated
problems continue unabated, with few exceptions, at colleges that are
most in need of intervention: those with high levels of heavy episodic
drinking. Addressing student alcohol use at heavy drinking colleges may
require stronger, more consistent, and more comprehensive approaches,
with increased emphasis on the alcohol environment. (J. Stud. Alcohol
Drugs 70: 726-734, 2009)
et al., 1994, 2002b). Students who attended schools with
high levels of heavy drinking were more likely to take up
heavy episodic drinking in college than students who attend
schools with lower drinking levels after accounting for their
drinking history before college (Weitzman et al., 2003).
Given the consequences of heavy drinking, special efforts
to understand and prevent heavy drinking are warranted at
schools with the highest drinking levels.
We have previously reported that little change has oc-
curred in drinking levels and related problems among a ran-
dom sample of American colleges between 1993 and 2001
(Wechsler et al., 2002b). Recent findings from the Monitor-
ing the Future study and the National Survey on Drug Use
and Health show that drinking behavior has remained consis-
tent since 2001 among respondents to these national surveys
who reported attending college (Johnston et al., 2007; Sub-
stance Abuse and Mental Health Services Administration,
2007). These reports discuss colleges in general but do not
focus on the group of colleges most in need of intervention:
those with the highest levels of heavy episodic drinking. The
present article examines change between 1993 and 2005 at a
sample of 18 colleges with high drinking levels that took part
in the College Alcohol Study. We specifically examine trends
at individual colleges within the sample and among groups
of students by gender, race/ethnicity, age, class, residence,
fraternity or sorority membership, and participation in athlet-
ics; these are characteristics demonstrated in prior research
to be associated with heavy drinking. We are particularly
 NELSON ET AL.
interested in determining whether drinking levels and related
problems change during this period among college students
in general and student subgroups at these schools with high
drinking levels.
Method
Sample
Four-year colleges located in the United States that had
participated in the Harvard School of Public Health College
Alcohol Study (Wechsler et al., 1994, 1998, 2000, 2002b)
were divided into tertiles according to the level of heavy
episodic drinking at the college: high (greater than 50% of
students reported heavy episodic drinking), medium (33%-
50%), and low (less than 33%), based on the results of a
survey conducted in 1993 (Wechsler et al., 1994). Heavy
episodic drinking was defined as consuming five or more
drinks in a row (four drinks for women) in a 2-week period
(Wechsler and Nelson, 2001, 2006; Wechsler et al., 1995a),
consistent with previous analyses of these data (Wechsler et
al., 1994, 1998, 2000., 2002b), other large-scale surveillance
studies (Johnston et al., 2007; Naimi et al., 2003), and rec-
ommendations for surveillance research on drinking behavior
(National Institute on Alcohol Abuse and Alcoholism, 2004;
World Health Organization, 2001). The mean college-level
proportion of heavy episodic drinking was 58.8% of students
in 1993, 53.9% in 1997, 57.5% in 1999, and 55.7% in 2001
in the heavy drinking colleges. In contrast, at colleges with a
medium level of heavy episodic drinking, 43.3% were heavy
episodic drinkers in 1993, 42.4% in 1997, 41.8% in 1999,
and 41.8% in 2001. For colleges with the lowest level of
heavy episodic drinking in 1993, 25.2% were heavy episodic
drinkers in 1993, 26.7% in 1997, 27.9% in 1999, and 28.0%
in 2001.
Colleges with high levels of heavy episodic drinking from
the 1993 survey were more likely to field National Collegiate
Athletic Association Division I athletics teams (70% high,
51% medium, 45% low; χ2 = 5.54, 2 df, p = .06), were more
likely to be located in rural areas or small towns (54% high,
20% medium, 24% low; χ2 = 12.54, 2 df, p = .002), and
were likely to be located in the Northeast or North Central
regions of the United States (71% high, 53% medium, 27%
low; χ2 = 17.79, 2 df, p = .007). They did not differ with
respect to enrollment size, admissions selectivity, religious
affiliation, or whether they were public or private.
The heavy drinking colleges from the 1993 survey were
the source for selecting 10 colleges that took part in the “A
Matter of Degree” (AMOD) program—an intervention to
reduce student heavy episodic drinking and related harms
(Weitzman et al., 2004). The 32 remaining schools in that
group were used as a comparison group for the evaluation of
that program. Those 42 colleges were contacted to take part
in a student survey on alcohol use in 2005. Colleges with
727
medium and low levels of heavy episodic drinking were not
surveyed in 2005. Twenty-eight colleges participated in the
2005 survey. The AMOD colleges were excluded from the
present analysis, because they were the subject of a special
intervention program that is described in a separate article
documenting declines in drinking and related harms at some
schools in that program (Weitzman et al., 2004).
Fourteen colleges did not participate in the 2005 survey,
primarily because they had already taken part in other sur-
veys of health behaviors on their campuses. The participat-
ing and nonparticipating colleges did not differ by college
characteristics, with the exception that six colleges that did
not participate had noncompetitive admissions standards,
whereas none of the participating colleges did. These col-
leges did not differ on measures of any drinking, heavy
episodic drinking, and frequent heavy episodic drinking in
any of the previous survey years, with the lone exception of
any alcohol consumption in the 1993 survey year (88.5%
at colleges participating in the 2005 survey and 90.7% for
those that did not [χ2 = 5.65, 1 df, p = .0175]).
The sample for the present study included students who
attended 18 schools located in 13 states in the United States.
They included 10 colleges with an enrollment of more
than 10,000 students, 4 with 5,001-10,000 students, 3 with
1,000-5,000 students, and 1 with fewer than 1,000 students.
Fourteen of the schools participated in National Collegiate
Athletic Association Division I athletics, three schools played
in Division II athletics, and one school played in Division III
athletics. Four of the schools were located in urban areas,
seven were in suburban areas, six were in small towns, and
one was in a rural area; seven schools were in the Northeast
region of the United States, five were in the North Central,
seven were in the South, and one was in the West. Five
schools were private colleges, and 13 were publicly funded;
2 colleges were affiliated with the Roman Catholic Church,
and the other 16 colleges had no religious affiliation. School
characteristics were obtained from Barron’s Guide to Col-
leges and the U.S. Department of Education (Barron’s, 2006;
National Center for Education Statistics, 2001).
At 18 colleges, 4,518 students completed surveys in
2005 (59% female, 82% white, and 60% younger than the
legal drinking age). Student characteristics were similar to
responses in the other survey years (1993: n = 2,497, 54%
female, 85% white, and 57% younger than the legal drink-
ing age; 1997: n = 2,372, 56% female, 80% white, and 57%
younger than the legal drinking age; 1999: n = 2,153, 58%
female, 81% white, and 60% younger than the legal drink-
ing age; and 2001: n = 1,714, 62% female, 81% white, and
56% younger than the legal drinking age). The total sample
from 4 years of data was 13,254. Because the demographic
characteristics of each student sample varied, a direct stan-
dardization procedure was used in the estimation process, as
described herein.
728 JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / SEPTEMBER 2009
Data collection and response
Data collection procedures for the 1993, 1997, 1999, and
2001 surveys were nearly identical and are described else-
where (Wechsler et al., 1998, 2000, 2002b). Methods for the
2005 survey were similar to previous procedures, with some
exceptions as outlined below.
Administrators at each school provided names, email
addresses, and postal addresses of 500 randomly selected
full-time undergraduate students enrolled during the 2004-
2005 school year. Data were collected using Web-based
and mailed questionnaires during the spring of 2005. Web-
based data collection is a cost-efficient, modern alternative
that appeals to college students, provides valid and reliable
measurement of alcohol use, and produces results that do
not differ from data obtained using other modalities (Ekman
et al., 2006; McCabe et al., 2006; Miller et al., 2002; Pealer
and Weiler, 2003; Pealer et al., 2001). No differences were
observed between student responses by survey modality in
the 2005 survey for heavy episodic drinking, frequent heavy
episodic drinking, driving after any drinking, experience of
five or more alcohol-related harms, and experience of two
or more secondhand effects. These findings are consistent
with previous findings in a separate survey using the same
survey questions (McCabe et al., 2006). A private research
firm that was contracted to administer the survey sent letters
to students to explain the study and notify them that they
would be invited to participate by email. Web survey data
were collected on a hosted secure site, running using secure
sockets layer protocol, to ensure that respondent data were
safely transmitted to the server and maintained according
to the Code of Standards and Ethics for Survey Research
established by the Council of American Survey Research
Organizations (1997-2009). Consistent with previous survey
administrations, the Web-based survey site was closed dur-
ing spring break and the 2 weeks following break to capture
behavior that occurred on campus and avoid responses that
reflected behavior during spring vacation (Wechsler et al.,
1994, 1998, 2000, 2002b). Students who did not respond to
the Web-based survey, or who requested it, were sent a mail
questionnaire. The survey wording, response categories, and
skip patterns in each mode were identical. Most (87.9%)
respondents completed the Web-based survey. Analyses were
adjusted to account for survey mode, using a binary indica-
tor variable for the Web-based survey (1 = yes, 0 = no).
Participation was voluntary, and responses were confidential.
As in previous surveys, students who completed the survey
or returned a response card were entered into a drawing for
cash prizes up to $1,500, with larger prizes offered for early
responders. The study protocol was approved by the Harvard
School of Public Health Human Subjects Committee.
A response rate was calculated consistent with previous
College Alcohol Study surveys (Wechsler et al., 1994, 1998,
2000, 2002b), in which the number of responses, at each
school and overall, was divided by the number of students
sampled after removing students who were not in school
during the survey period, dropped out of school, or changed
their status (e.g., went from full-time to part-time attendance,
were on leave, or were studying abroad) between when the
sample was drawn in the fall semester and when the survey
was conducted in the spring semester. The response rate for
the 18 colleges was 56% in the 2005 survey (range: 39%-
78% by college), 55% in the 2001 survey (range: 39%-71%),
62% in the 1999 survey (range: 53%-75%), and 61% in the
1997 survey (range: 51%-78%). Response rate at the col-
lege level was not associated with heavy episodic drinking
at the college level (r = -.117, p = .33), a finding similar
to previous administrations of the survey (Wechsler et al.,
2000, 2002b). Consistent with previous College Alcohol
Study analyses, response rate was included as a college-level
covariate to guard against potential response bias, but this
procedure did not alter the findings (Wechsler et al., 1998,
2000, 2002b). The analyses were weighted to the known
demographic distribution of each school by gender (male/
female), age (<21/≥21 years), and race (white/nonwhite) to
reduce the threat of selection bias, using year-matched statis-
tics from Barron’s Guide to U.S. Colleges (Barron’s, 2006).
Measures
The 2005 survey questionnaire asked the same questions
about alcohol use, alcohol-related problems, and demograph-
ic characteristics as previous College Alcohol Study surveys
(Wechsler and Austin, 1998; Wechsler and Nelson, 2001;
Wechsler et al., 1994, 1995a, 1998, 2000, 2002b). A series of
dichotomous measures of alcohol use were assessed. Heavy
episodic drinking was defined as five (for men) or four (for
women) drinks or more per drinking occasion at least once
in the past 2 weeks. Frequent heavy episodic drinking was
defined as five (for men) or four (for women) drinks on
three or more occasions in the past 2 weeks. Abstainers were
students who had not consumed any alcohol in the past 12
months. Students who had consumed alcohol in the past 12
months, but did not have a heavy drinking episode in the
previous 2 weeks, were non-heavy-episodic drinkers. Among
past-year drinkers, the measures included consumption on 10
or more occasions in the past 30 days, usually consuming
five or more drinks for men, four or more for women (on a
drinking occasion), and being drunk on three or more occa-
sions in the past 30 days.
Students who drank alcohol in the previous 12 months
were asked about their experience of 12 alcohol-related
problems, ranging from academic difficulties to unplanned
or unprotected sex to physical or sexual assault. Responses
were used to create a series of dichotomous variables reflect-
ing any experience of each problem during the school year.
Dichotomous measures of driving after any drinking and
after five or more drinks were examined among students who
 NELSON ET AL.
drove a vehicle in the past 30 days (Wechsler et al., 2003).
Consistent with previous studies, a dichotomous summary
measure of five or more alcohol-related problems was cre-
ated. All students were asked whether they were passengers
in a vehicle with a high or drunk driver.
All students were asked a series of eight questions about
whether they were affected by other students’ drinking
(secondhand effects) during the school year, including in-
terruptions of sleep and study, verbal or physical assaults,
and destruction of personal property; then, a series of di-
chotomous variables were created (Wechsler et al., 1995b).
A dichotomous summary measure of two or more of these
secondhand effects was examined. Reports about experience
as a victim of sexual assault are shown for female students
only (Mohler-Kuo et al., 2004).
Missing data on the drinking outcomes occurred in ap-
proximately 2% or fewer of cases across the four surveys
and declined over time (heavy episodic drinking = 2.0% in
1993, 1.3% in 1997, 2.2% in 1999, 0.6% in 2001, and 0.8%
in 2005). Sensitivity analyses comparing listwise deletion of
missing outcomes, or including missing outcomes as either
all positive or all negative, revealed no differences in the
overall findings. Missing values on outcomes were deleted
listwise from the analyses.
Students were asked a series of questions about their
personal characteristics (e.g., age, gender, race), their year
in school (e.g., freshman, sophomore), and where and
with whom they lived while they were in school (e.g., in a
residence hall, off-campus, with parents, with a roommate)
(Wechsler et al., 2002b). They were also asked about social
affiliations associated with alcohol use, including member-
ship in a fraternity or sorority and participation in athletics.
Athletes were defined as students who played or practiced
intercollegiate athletics on average 1 or more hours per day
(Nelson and Wechsler, 2001).
Data analysis
Data analysis was conducted using SAS/PC statistical
software Version 9.1 (SAS Institute, Inc., Cary, NC). The
generalized estimating equations (Liang and Zeger, 1992;
Zeger et al., 1988) was used to directly compare the 2005
results with each of the other survey years, using indicator
variables for each year and the year 2005 as the referent, in
the same model. A similar set of analyses was conducted to
estimate the time trend from 1993 to 2005 (with year coded
as 1993 = 0, 1997 = 4, 1999 = 6, 2001 = 8, and 2005 =
12). Because, in some cases, only the 1993 survey year was
significantly different, the same time trend analysis was con-
ducted using the year 1997 as the baseline (with year coded
as 1997 = 0, 1999 = 2, 2001 = 4, and 2005 = 8). All three
sets of analyses are presented. All analyses were adjusted for
school-level response rate and survey modality. In addition
to weighting the sample by gender, age, and race, analyses
729
were adjusted for ethnicity (Hispanic/non-Hispanic), place
of residence (on-campus residence hall/substance-free cam-
pus housing/fraternity or sorority/off campus/off campus
with parents), fraternity or sorority members (member/non-
member), and participation in athletics (athlete/nonathlete),
because the sample characteristics of students varied slightly
by year, although these adjustments did not change overall
findings.
Trends in student characteristics and in heavy episodic
drinking by student characteristics and school-specific trends
over five surveys were examined using the Cochran-Armitage
trend test, relative to a hypothesis of no change over time.
The p values for the direct comparison of each survey
with the 2005 survey and a slope parameter from the test
for trend over time relative to a slope of zero are reported.
All performed tests were two-sided, and a p < .01 was con-
sidered statistically significant. Only statistically significant
findings for both the trend tests and the direct comparison
are discussed.
Results
Student sample
Student demographic characteristics at the 18 colleges
varied slightly over the 4 survey years. The proportion of
men decreased over time (46% in 1993, 44% in 1997, 43%
in 1999, 38% in 2001, and 40% in 2005; trend test Z =
5.06, two-sided p < .01). The racial/ethnic characteristics
were similar over time (averaging 84% white, 4% black, 6%
Asian, 6% Native American and other, and 5% Hispanic over
the five surveys), as was age, with 59% of students younger
than 21 years, 36% ages 21-23 years, and 5% ages 24 years
and older across the five surveys. Significant shifts occurred
in where students live. More students lived in substance-free,
on-campus housing (7% in 1993, 9% in 1997, 8% in 1999,
10% in 2001, and 15% in 2005; Z = -10.89, two-sided p <
.001), whereas fewer lived in non-substance-free campus
housing (42% in 1993, 45% in 1997, 46% in 1999, 37% in
2001, and 35% in 2005; Z = 7.89, two-sided p < .001). The
proportion living off-campus did not change, either for stu-
dents living with a parent (6% averaged across surveys) or
not with a parent (39% averaged across surveys). There was
no change for living in a fraternity or sorority house (4% av-
eraged across surveys) or for members of fraternities or so-
rorities who were not house residents (15% averaged across
surveys). The proportion of students in the sample who
reported participating in intercollegiate athletics declined
significantly over time (19% in 1993, 17% in 1997, 18% in
1999, 15% in 2001, and 14% in 2005; Z = 5.76, two-sided
p < .01). The proportion of students who engaged in heavy
episodic drinking during high school declined significantly
(36% in 1993, 32% in 1997, 35% in 1999, 29% in 2001, and
32% in 2005; Z = 3.26, two-sided p < .01).
730 JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / SEPTEMBER 2009

TABLE 1. Patterns of alcohol use among students attending heavy drinking colleges 1993-2005, in percentagea
Linear trend Linear trend
1993 1997 1999 2001 2005 (1993-2005) (1997-2005)
Variable (n = 2,497) (n = 2,372) (n = 2,153) (n = 1,714) (n = 4,518) slope pb slope pb
Past year drinker 88.5‡ 85.4 86.1 86.7 86.2 <.0001 .1371
Heavy episodic drinker,
past 2 weeks 58.1‡ 53.4 57.1† 53.5 55.8 <.0001 .3786
Abstain, past year 11.5‡ 14.6 13.9 13.3 13.8 <.0001 .1371
Drinker, no heavy episodic
drinking past 2 weeks 30.4 31.9 29.0 33.2 30.3 .2261 .9153
Occasional heavy
episodic drinker 29.8‡ 25.6 24.3 22.5 23.9 <.0001 .0612
Frequent heavy
episodic drinker 28.3 27.8* 32.8 31.1 32.0 .1549 .0051
aPercentages are weighted and standardized to the baseline year by gender, age, and race; badjusted for ethnicity, place of residence,
fraternity or sorority membership, athletic participation, response rate, and survey mode.
Significant difference from 2005, *p < .05; †p < .01; ‡p < .001.

Alcohol consumption Drinking style

Some small, but statistically significant, changes in the
proportions of drinkers occurred between 1993 and 2005,
with most of the change occurring between 1993 and 1997.
In 2005, seven of eight students (86%) at 18 heavy drinking
schools drank alcohol in the past year. This was a significant
change from 1993 (88.5%) (Table 1). Similarly, fewer drink-
ers meant more abstainers (13.8% in 2005 compared with
11.5% in 1993). In 2005, more than half (55.8%) were heavy
episodic drinkers (see Table 1). These levels were signifi-
cantly lower compared with students at the same colleges in
1993 (58.1%), although they were similar to drinking levels
observed in the other survey years. Drinking levels remained
unchanged at these heavy drinking schools between 1997
and 2005. When heavy episodic drinkers are divided on the
basis of how often they drank at the five/four level during
the past 2 weeks into occasional (one or two times) and fre-
quent (three or more times), we find two different patterns
of change. Occasional heavy episodic drinking declined sig-
nificantly from 1993 to 2005. Nonetheless, the proportion of
frequent heavy episodic drinkers at these schools was higher
in 2005, but the difference was not statistically significant.
Among students who drank alcohol, many drank heavily
(Table 2). In 2005, one of four (28%) drinkers consumed
alcohol on 10 or more occasions, and two of five (40%)
drinkers had been drunk on 3 or more occasions in the last
30 days. These levels were significantly higher than the levels
among students attending these same schools in 1993, al-
though little change was observed between 1997 and 2005.
Heavy episodic drinking by year and subgroup
characteristics
Heavy episodic drinking did not change during this period
for most student subgroups (Table 3). The exceptions were
increases among students ages 21-23 years, students living
off-campus away from parents, and students in their senior
year of school. Significant declines in heavy episodic drink-
ing were observed among white students and those involved
in intercollegiate athletics.
Among the groups with the highest levels of heavy epi-
sodic drinking, fraternity or sorority members had consis-
tently high levels of heavy episodic drinking over time, with

TABLE 2. Drinking styles of students who drank alcohol in the past year and attended heavy drinking colleges 1993-2005, in percentagea
Test for linear Test for linear
time trend time trend
1993 1997 1999 2001 2005 (1993-2005) (1997-2005)
Variable (n = 2,177) (n = 2,009) (n = 1,831) (n = 1,480) (n = 3,864) pb pb
10 or more drinking
occasions, past 30 days 21.9‡ 25.4 28.0 27.8 27.6 .0002 .1478
3 or more occasions drunk,
past 30 days 31.8‡ 34.5‡ 38.3 35.8* 39.7 .0006 .0174
When drinking,
usually consume 5+/4+ 52.1 50.9 56.1* 46.9‡ 54.0 .0263 .1760
aPercentages are weighted and standardized to the baseline year by gender, age, and race; badjusted for ethnicity, place of residence, fraternity
or sorority membership, athletic participation, response rate, and survey mode.
Significant difference from 2005, *p < .05; ‡p < .001.
 NELSON ET AL. 731

TABLE 3. Student heavy episodic drinking by year and subgroup characteristics at heavy drinking colleges 1993-2005, in percentage (standard error)a
Cochran- Cochran-
Armitage Armitage
trend test trend test
(1993-2005) (1997-2005)
1993 1997 1999 2001 2005 p p
Variable (n = 2,497) (n = 2,372) (n = 2,153) (n = 1,714) (n = 4,518) Statistic (2-sided) Statistic (2-sided)
Gender
Female 55.7 48.4 52.4 49.8 51.8 1.32 .19 -0.29 .76
Male 60.4 58.1 61.9 57.4 59.8 0.17 .87 -1.47 .14
Race
White 62.2 57.5 62.0 59.0 58.3 2.47 .01 0.33 .74
Nonwhite 40.3 37.1 36.7 29.8 40.8 -0.38 .70 -1.11 .27
Ethnicity
Hispanic 50.2 47.9 58.9 46.9 51.6 -0.29 .77 0.65 .52
Non-Hispanic 58.4 53.9 57.0 53.9 55.0 2.19 .03 -0.19 .85
Age group, in years
<21 56.7 51.4 54.9 51.4 52.1 2.36 .02 1.25 .21
21-23 62.8 58.5 64.2 60.1 64.0 -1.25 .21 -2.41 .02
>23 44.3 45.3 39.8 25.0 40.2 1.42 .16 -0.11 .92
Class
Freshman 54.6 50.6 51.8 49.9 52.5 0.45 .65 0.16 .87
Sophomore 54.2 51.1 54.4 55.3 52.0 0.52 .60 0.67 .50
Junior 60.8 54.7 61.2 52.7 56.0 1.79 .07 0.55 .58
Senior 58.9 59.6 63.0 55.3 63.2 -1.58 .11 -2.28 .02
5th year 70.1 52.5 57.3 61.8 58.4 1.83 .07 -1.24 .22
Residence
Off campus with parent 38.2 29.8 36.3 27.2 31.3 1.48 .14 -0.06 .95
Substance-free on-campus housing 43.6 40.1 37.8 46.3 45.3 -1.18 .24 -1.42 .15
Unrestricted on-campus housing 56.1 53.5 55.4 55.3 54.0 0.80 .43 0.43 .67
Off campus without parent 64.6 58.3 62.8 56.3 63.4 -0.16 .87 -2.13 .03
Fraternity/sorority house 88.3 86.6 84.2 82.2 83.5 1.32 .19 0.77 .44
Fraternity/sorority member
No 52.1 48.1 52.9 50.4 52.1 -0.86 .39 -1.34 .18
Yes 77.5 75.7 74.7 69.4 74.3 1.61 .11 -0.42 .68
Athlete
No 56.7 52.0 56.6 52.8 54.4 1.08 .28 -0.81 .42
Yes 64.1 60.9 59.5 57.0 57.5 2.38 .02 1.11 .27
aPercentages are weighted and standardized to the baseline year by gender, age, and race.

75% of fraternity or sorority members (vs 51% of nonmem-
bers) and 85% of fraternity or sorority house residents (vs
72% of nonresident members) classified as heavy episodic
drinkers averaged across the surveys. There were no signifi-
cant changes within these groups over time.
Alcohol-related problems
Consistent with an overall lack of change in alcohol
consumption, no change over time was observed in most
measures of alcohol-related problems (Table 4). The only
exception was a significant decline in driving after drink-
ing between 1997 and 2005. However, no similar change in
driving after consuming five or more drinks or riding with
an intoxicated driver was observed.
Secondhand effects of alcohol use
There was no pattern of significant changes in second-
hand effects among students who did not engage in heavy
episodic drinking and resided on-campus attending these 18
heavy drinking schools from 1993 to 2005 (Table 5).
Institutional-level trends
An examination of change at each of the 18 participat-
ing schools individually yielded few significant results and
no pattern of change. Findings indicated a statistically sig-
nificant (p < .01) decline in heavy episodic drinking at one
school and no significant increases. A similar distribution of
change over time was observed for abstaining from alcohol
(two colleges with a statistically significant increase and
one college with statistically significant decline), drinking
on 10 or more occasions in the past 30 days (significant
increase = 2), three or more drunken occasions in the past
30 days (significant increase = 2), driving after drinking
(significant decline = 1), driving after five or more drinks
(significant increase = 1), riding with a high or drunk driver
(significant decline = 2), five or more alcohol-related harms
among drinkers (significant increase = 1), and experience
732 JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / SEPTEMBER 2009

TABLE 4. Alcohol-related problems of students who drank alcohol attending heavy drinking colleges 1993-2005, in percentagea
Test for linear Test for linear
time trend time trend
1993 1997 1999 2001 2005 (1993-2005) (1997-2005)
Problem (n = 2,177) (n = 2,009) (n = 1,831) (n = 1,480) (n = 3,864) p b pb
Hangover 68.6 68.3 69.5 69.7 68.0 .9327 .1265
Missed a class 35.2† 37.3* 35.1 37.9* 34.9 .7886 .2600
Fell behind in school work 25.2 26.7* 27.5* 24.9 24.5 .6561 .3636
Did something regrettable 38.3 44.4‡ 44.2‡ 41.2 39.8 .0934 .2479
Forgot where you were or what you did 34.8 35.0* 34.9* 31.8‡ 38.2 .5936 .2660
Argued with friends 24.3 29.2 27.5 25.3* 28.3 .3634 .4619
Engaged in unplanned sexual activities 23.7 27.1 25.9 24.4 26.1 .9582 .2604
Did not use protection during sex 11.4 12.6 10.9 11.5 11.8 .5296 .8576
Damaged property 12.2 15.2 15.1 15.3 14.1 .1594 .4816
Got in trouble with the police 6.2 6.5 7.1 8.0 7.5 .0412 .0504
Got hurt or injured 12.9† 17.6 16.3 17.1 17.5 .0312 .7834
Required medical treatment for an overdose 0.4 1.0 0.6 0.5 1.0 .9242 .1444
Drove after any drinkingc 36.8 42.9‡ 40.2‡ 39.7 36.8 .1986 .0002
Drove after five or more drinksc 13.4 15.3 19.9‡ 16.3 14.6 .2913 .3366
Rode with a high or drunk driver 23.1‡ 25.4‡ 27.9‡ 26.4‡ 22.3 .4837 .9755
Five or more problemsd 21.9 26.5 24.0 25.6 25.5 .0407 .4386
aPercentages are weighted and standardized to the baseline year by gender, age, and race; badjusted for ethnicity, place of residence, fraternity or sorority

membership, athletic participation, response rate, and survey mode; camong students who drove one or more times per month; dexcluding hangover, drove
after five or more drinks, and riding with a high or drunk driver.
Significant difference from 2005, *p < .05; †p < .01; ‡p < .001.

of secondhand effects of alcohol among students who did
not engage in heavy episodic drinking and who resided on-
campus (significant increase = 3). No single school showed
a specific pattern of change over time across measures.
Discussion
Colleges that had high levels of heavy episodic drinking
in 1993 continue to have high levels of drinking and as-
sociated problems 12 years later. When individual schools
were examined, the consistency of drinking behavior was
also evident. This stability occurred despite the continuous
entry of new students at these colleges. Among subgroups
of students, there were also few changes in heavy episodic
drinking. Consistent with the stability of drinking behavior
among students at these heavy drinking colleges, there were
no changes in alcohol-related harms among drinkers or the
secondhand effects of alcohol use by others among non-
heavy-episodic-drinking students residing on-campus over
time.
Statistically significant declines in driving after any drink-
ing were observed between 1997 and 2005. These decreases,
although small, may be a welcome sign of change. Motor
vehicle crashes account for most alcohol-attributable mor-

TABLE 5. Secondhand effects of others’ alcohol use for non-heavy-episodic-drinking students who live on-campus or in a fraternity/sorority house and attend
heavy drinking colleges 1993-2005, in percentagea,b
Test for linear Test for linear
time trend time trend
1993 1997 1999 2001 2005 (1993-2005) (1997-2005)
Secondhand effects (n = 557) (n = 661) (n = 565) (n = 400) (n = 1,168) pc pc
Been insulted or humiliated 31.9 33.1* 33.7* 28.7 27.4 .4678 .0953
Had a serious argument and quarrel 19.0 21.7† 18.7 18.9 16.5 .2442 .0716
Been pushed, hit, or assaulted 12.0* 10.7 12.4 9.7 9.6 .5779 .5019
Had property damaged 12.6 17.9* 12.5 15.4 14.4 .1593 .0286
Had to take care of drunken student 53.3 53.5 54.9 56.6 56.2 .8708 .5042
Found vomit in residenced N/A 58.5* 54.1 50.0* 56.4 N/A .0008
Had studying/sleeping interrupted 63.8 67.8* 67.6* 59.5 63.9 .0071 <.0001
Experienced an unwanted sexual advance 19.7 19.1 19.9 20.3 16.3 .7845 .8859
Victim of sexual assault or rapee 2.6 1.3 1.4 1.0 1.3 .2137 .7215
Two or more of these secondhand effectsd 61.7 60.6 59.8 58.5 56.4 .7779 .1809
Notes: N/A = not available. aIncludes only abstainers or non-heavy episodic drinkers living in campus dormitories or a fraternity/sorority house; bpercentages are
weighted and standardized to the baseline year by gender, age, and race; cadjusted for ethnicity, place of residence, fraternity or sorority membership, athletic
participation, response rate, and survey mode; dthe question about finding vomit in the residence hall was not asked in 1993; the two or more secondhand
effects summary measure does not include finding vomit in residence; eamong female students only.
Significant difference from 2005, *p <.05; †p < .01.
 NELSON ET AL.
tality among college students (Hingson et al., 2005). Any
reduction in drinking and driving may lessen these serious
consequences. However, no similar change was observed in
driving after consuming five or more drinks, a volume of
consumption linked with the national legal blood alcohol
concentration driving limit of .08 (National Institute on
Alcohol Abuse and Alcoholism, 2004), or in riding with an
intoxicated driver in this sample.
Several limitations of this research should be considered.
Self-reported survey questionnaires are subject to response
bias. Nevertheless, these methods are common in alcohol-use
studies and are considered valid and reliable for surveillance
research (Cooper et al., 1981; Frier et al., 1991; Midanik,
1988). Another limitation of this research is the comparison
of current and high school alcohol consumption, which is
subject to recall bias or memory decay. However, this effect
should be similar across survey years. The response rate
for the College Alcohol Study has declined since the initial
survey and may affect the results of the survey. It is difficult
to predict the direction of bias this trend may have produced.
The smaller sample may decrease measurement precision
and make it more difficult to detect true change over time.
However, heavy episodic drinking across the 4 survey years
has remained remarkably consistent, and efforts to examine
the influence of response rate statistically have revealed
little effect. In addition, subgroups of students and colleges
exhibit consistent patterns of heavy episodic drinking over
time. To help protect against nonresponse bias, college re-
sponse rate was included in analyses, and data were weighted
to represent the known demographics of each college. The
study is also limited by the attrition of colleges from the
sample. Although participating and nonparticipating colleges
did not differ in previous surveys, nonresponders may have
changed between 2001 and 2005. It is also possible that the
few significant changes that were observed were simply the
result of chance alone.
The 2005 survey administration was conducted using a
mixed mode (Web-based and mailed) methodology. The dif-
ference in methodology between administrations may have
obscured real changes over time. However, this methodology
is recommended for use with college students and is valid
and reliable for collecting data on alcohol use (Ekman et al.,
2006; McCabe et al., 2006; Miller et al., 2002; Pealer and
Weiler, 2003; Pealer et al., 2001). We found no differences
between survey mode in drinking behavior, harms, or sec-
ondhand effects in the 2005 survey or in a previous survey
with a different group of colleges (McCabe et al., 2006).
The findings of the present study indicate that alcohol use
continues to be a serious problem at colleges that had high
levels of heavy episodic drinking in 1993. At these schools,
five of six students drank alcohol (ranging from 85%-88%),
and more than half of the students engaged in heavy episodic
drinking (ranging from 53% to 58%) in 1997, 1999, 2001,
and 2005. Data on what, if any, interventions were tried at
733
these schools between 2002 and 2005 were not collected;
therefore, explaining the lack of progress is difficult. There
are several possible reasons for the continuation of heavy
drinking. These colleges may not be trying to combat student
heavy drinking, although in a national study of 747 col-
leges conducted in 2002, most 4-year colleges in the United
States reported some activities to reduce student drinking
(Wechsler et al., 2004). It is also possible that these heavy
drinking colleges are either not doing enough or are not us-
ing effective prevention policies or programs. A task force
convened by the National Institute on Alcohol Abuse and
Alcoholism (2002) made recommendations for colleges to
address student drinking. These included methods for reach-
ing individual students, such as cognitive-behavioral skills
training, brief motivational enhancement, and challenging
alcohol expectancies. They also included implementation
of, public information campaigns about, and enforcement
of laws to prevent alcohol-impaired driving and underage
drinking, restrictions on alcohol retail outlets, increasing
prices and excise taxes on alcoholic beverages, and re-
sponsible beverage service policies at on- and off-campus
venues. More research is needed to understand what col-
leges are doing to combat heavy episodic drinking, whether
they are using recommended interventions—the barriers to
implementing recommended interventions—and whether
their efforts effectively reduce alcohol use and related con-
sequences. It is also possible that colleges, by themselves,
cannot implement effective prevention efforts. To enact these
approaches, administrators and prevention specialists may
need additional support from other groups, including local
and state policymakers, law enforcement, boards of trustees,
parents, current students, and alumni. The National Institute
on Alcohol Abuse and Alcoholism Task Force on College
Drinking (Task Force of the Advisory Council on Alcohol
Abuse and Alcoholism, 2002) recommended that colleges
engage community stakeholders in their efforts to address
student drinking. More research is needed to identify meth-
ods for helping colleges and college communities with the
difficult question of “how” to do this important work.
References
BARRON’S EDUCATIONAL SERIES. Barron’s Profiles of American Colleges, 27th
2007 Edition, Hauppauge, NY: Barron’s Educational Series, 2006.
COOPER, A.M., SOBELL, M.B., SOBELL, L.C., AND MAISTO, S.A. Validity of al-
coholic’s self-reports: Duration data. Int. J. Addict. 16: 401-406, 1981.
COUNCIL OF AMERICAN SURVEY RESEARCH ORGANIZATIONS (CASRO) CASRO
Code of Standards and Ethics for Survey Research (2006), Port Jeffer-
son, NY: Council of American Survey Research Organizations, 1997-
2009 (available at: www.casro.org/codeofstandards.cfm).
EKMAN, A., DICKMAN, P.W., KLINT, A., WEIDERPASS, E., AND LITTON, J.E.
Feasibility of using web-based questionnaires in large population-based
epidemiological studies. Europ. J. Epidemiol. 21: 103-111, 2006.
FRIER, M., BELL, R.M., AND ELLICKSON, P.L. Do Teens Tell the Truth? The
Validity of Self-Reported Tobacco Use by Adolescents. RAND Note,
Document No. N-3291-CHF, Santa Monica, CA: RAND, 1991.
734 JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / SEPTEMBER 2009
HINGSON, R., HEEREN, T., WINTER, M., AND WECHSLER, H. Magnitude of
alcohol-related mortality and morbidity among U.S. college students
ages 18-24: Changes from 1998 to 2001. Annual Rev. Publ. Hlth 26:
259-279, 2005.
JOHNSTON, L.D., O’MALLEY, P.M., BACHMAN, J.G., AND SCHULENBERG, J.E.
Monitoring the Future: National Survey Results on Adolescent Drug
Use: Overview of Key Findings, 2006, NIH Publication No. 07-6206,
Bethesda, MD: National Institute on Drug Abuse, 2007.
LIANG, K.-Y. AND ZEGER, S.L. Longitudinal data analysis using generalized
linear models. Biometrika 73: 12-22, 1992.
MCCABE, S.E., DIEZ, A., BOYD, C.J., NELSON, T.F., AND WEITZMAN, E.R.
Comparing web and mail responses in a mixed mode survey in college
alcohol use research. Addict. Behav. 31: 1619-1627, 2006.
MIDANIK, L.T Validity of self-report alcohol use: A literature review and
assessment. Brit. J. Addict. 83: 1019-1030, 1988.
MILLER, E.T., NEAL, D.J., ROBERTS, L.J., BAER, J.S., CRESSLER, S.O., METRIK,
J., AND MARLATT, G.A. Test-retest reliability of alcohol measures: Is
there a difference between internet-based assessment and traditional
methods? Psychol. Addict. Behav. 16: 56-63, 2002.
MOHLER-KUO, M., DOWDALL, G.W., KOSS, M.P., AND WECHSLER, H. Correlates
of rape while intoxicated in a national sample of college women. J. Stud.
Alcohol 65: 37-45, 2004.
NAIMI, T.S., BREWER, R.D., MOKDAD, A., DENNY, C., SERDULA, M.K., AND
MARKS, J.S. Binge drinking among U.S. adults. JAMA 289: 70-75,
2003.
NATIONAL CENTER FOR EDUCATION STATISTICS (E.D. TABS). Enrollment in
Postsecondary Institutions, Fall 2001 and Financial Statistics, Fiscal
Year 2001, Publication No. NCES 2004-155, Washington, DC: Institute
of Education Sciences, Department of Education, 2003 (available at:
http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2004155).
NATIONAL INSTITUTE ON ALCOHOL ABUSE AND ALCOHOLISM. NIAAA Council
Approves Definition of Binge Drinking. NIAAA Newsletter, winter No.
3, p. 3, NIH Publication No. 04-5346, Bethesda, MD: National Institute
on Alcohol Abuse and Alcoholism, 2004.
NELSON, T.F. AND WECHSLER, H. Alcohol and college athletes. Med. Sci.
Sports Exerc. 33: 43-47, 2001.
O’MALLEY, P.M. AND JOHNSTON, L.D. Epidemiology of alcohol and other
drug use among American college students. J. Stud. Alcohol, Supple-
ment No. 14, pp. 23-39, 2002.
PEALER, L. AND WEILER, R.M. Guidelines for designing a web-delivered col-
lege health risk behavior survey: Lessons learned from the University
of Florida Health Behavior Survey. Hlth Promot. Pract. 4: 171-179,
2003.
PEALER, L.N., WEILER, R.M., PIGG, R.M., JR., MILLER, D., AND DORMAN,
S.M. The feasibility of a web-based surveillance system to collect
health risk behavior data from college students. Hlth Educ. Behav. 28:
547-559, 2001.
PERKINS, H. Surveying the damage: A review of research on consequences
of alcohol misuse in college populations. J. Stud. Alcohol, Supplement
No. 14, pp. 91-100, 2002.
SUBSTANCE ABUSE AND MENTAL HEALTH SERVICES ADMINISTRATION (Office of
Applied Studies). Results from the 2006 National Survey on Drug Use
and Health: National Findings, DHHS Publication No. SMA 07-4293,
Rockville, MD: Substance Abuse and Mental Health Services Admin-
istration, 2007.
TASK FORCE OF THE NATIONAL ADVISORY COUNCIL ON ALCOHOL ABUSE AND
ALCOHOLISM. A Call to Action: Changing the Culture of Drinking at
U.S. Colleges, NIH Publication No. 02-5010, Bethesda, MD: National
Institute on Alcohol Abuse and Alcoholism, 2002.
WECHSLER, H. AND AUSTIN, S.B. Binge drinking: The five/four measure. J.
Stud. Alcohol 59: 122-123, 1998.
WECHSLER, H., DAVENPORT, A., DOWDALL, G., MOEYKENS, B., AND CASTILLO,
S. Health and behavioral consequences of binge drinking in college: A
national survey of students at 140 campuses. JAMA 272: 1672-1677,
1994.
WECHSLER, H., DOWDALL, G.W., DAVENPORT, A., AND RIMM, E.B. A gender-
specific measure of binge drinking among college students. Amer. J.
Publ. Hlth 85: 982-985, 1995a.
WECHSLER, H., DOWDALL, G.W., MAENNER, G., GLEDHILL-HOYT, J., AND LEE,
H. Changes in binge drinking and related problems among American
college students between 1993 and 1997: Results of the Harvard School
of Public Health College Alcohol Study. J. Amer. Coll. Hlth 47: 57-68,
1998.
WECHSLER, H., LEE, J.E., HALL, J., WAGENAAR, A.C., AND LEE, H. Secondhand
effects of student alcohol use reported by neighbors of colleges: The role
of alcohol outlets. Social Sci. Med. 55: 425-435, 2002a.
WECHSLER, H., LEE, J.E., KUO, M., AND LEE, H. College binge drinking in
the 1990s: A continuing problem. Results of the Harvard School of
Public Health 1999 College Alcohol Study. J. Amer. Coll. Hlth 48:
199-210, 2000.
WECHSLER, H., LEE, J.E., KUO, M., SEIBRING, M., NELSON, T.F., AND LEE, H.
Trends in college binge drinking during a period of increased preven-
tion efforts: Findings from 4 Harvard School of Public Health College
Alcohol Study surveys: 1993-2001. J. Amer. Coll. Hlth 50: 203-217,
2002b.
WECHSLER, H., LEE, J.E., NELSON, T.F., AND LEE, H. Drinking and driving
among college students: The influence of alcohol-control policies. Amer.
J. Prev. Med. 25: 212-218, 2003.
WECHSLER, H., MOEYKENS, B., DAVENPORT, A., CASTILLO, S., AND HANSEN, J.
The adverse impact of heavy episodic drinkers on other college students.
J. Stud. Alcohol 56: 628-634, 1995b.
WECHSLER, H. AND NELSON, T.F. Binge drinking and the American college
student: What’s five drinks? Psychol. Addict. Behav. 15: 287-291,
2001.
WECHSLER, H. AND NELSON, T.F. Relationship between level of consumption
and harms in assessing drink cut-points for alcohol research: Commen-
tary on “Many college freshmen drink at levels far beyond the binge
threshold” by White et al. Alcsm Clin. Exp. Res. 30: 922-927, 2006.
WECHSLER, H., SEIBRING, M., LIU, I.C., AND AHL, M. Colleges respond to
student binge drinking: Reducing student demand or limiting access. J.
Amer. Coll. Hlth 52: 159-168, 2004.
WEITZMAN, E.R. AND NELSON, T.F. College student binge drinking and the
“prevention paradox”: Implications for prevention and harm reduction.
J. Drug Educ. 34: 247-265, 2004.
WEITZMAN, E.R., NELSON, T.F., LEE, H., AND WECHSLER, H. Reducing drinking
and related harms in college: Evaluation of the “A Matter of Degree”
program. Amer. J. Prev. Med. 27: 187-196, 2004.
WEITZMAN, E.R., NELSON, T.F., AND WECHSLER, H. Taking up binge drinking
in college: The influences of person, social group, and environment. J.
Adolesc. Hlth 32: 26-35, 2003.
WORLD HEALTH ORGANIZATION. International Guide for Monitoring Alcohol
Consumption and Related Harm, WHO Document No. WHO/MSD/
MSB/00.4, Geneva, Switzerland: Department of Mental Health and
Substance Dependence, World Health Organization, 2001.
ZEGER, S.L., LIANG, K.Y., AND ALBERT, P.S. Models for longitudinal data: A
generalized estimating equation approach. Biometrics 44: 1049-1060,
1988.