High-Risk Drinking in College: What We Know and What We Need To Learn
Research Considerations
Numerous and varied factors contribute to and sustain heavy episodic drinking among college
students. They include the student and his or her background; the developmental tasks that
accompany this phase of life; the peer group on campus; the particular college’s rituals and
traditions; and the community environment, including alcohol supply and alcohol marketing
practices. The heterogeneity that exists among institutions and students means that findings from
studies conducted on one campus, or even on a few campuses, may not be generalizable to other
institutions.
College students and their drinking behavior have now been studied for at least 50 years. The
pioneering work of Straus and Bacon (1953), Drinking in College, is a milestone in this
area, and served as a starting point for hundreds of subsequent studies. Since then, substantial
methodological progress has been made in measuring college student drinking.
Like any complex human behavior, drinking among college students is best approached as part of a
multifaceted system with many elements. Research processes try to balance this inherent complexity
against the need to construct specific study designs that are feasible and that address specific
research questions. The major approaches used to study drinking behavior in college include
exploratory research, descriptive research, explanatory research, and evaluation research (Dowdall
and Wechsler, 2002).
Exploratory Research
Exploratory research attempts to understand some problem or area of study in a preliminary way.
Research designs for exploratory research often rely on direct observation of a limited number of
examples of what is to be studied—in this case, drinking behavior in college. Focus groups, which
bring together a small group of subjects to engage in a guided discussion about a limited issue,
are an example of exploratory research. Although exploratory studies do not and are not intended
to provide precise estimates of the prevalence of a problem or to test specific theories, they can
be invaluable in mapping domains of study or clarifying the feasibility of specific research
strategies and tactics.
Descriptive Research
Descriptive research usually attempts to examine a few well-developed constructs in detail in an
effort to estimate the frequency with which certain behaviors, such as drinking, or relevant
characteristics occur in a given population. Descriptive research should consider and attempt to
adjust for any bias that could distort a study’s findings such as skewed participation by a
particular population subgroup. An example of descriptive research is a survey of the prevalence
of specific drinking practices.
Explanatory Research
Explanatory research attempts to investigate the causes of particular phenomena, not simply to
describe them. Such research carefully tests causal hypotheses (Campbell and Stanley, 1963; Cook
and Campbell, 1979; Dowdall et al., 1999). It also rules out rival explanations, including:
Maturation (i.e., does drinking decrease after college because students leave a
drinking-supportive environment or because of an age-graded phenomenon that affects students and
their nonstudent peers equally?)
Chance (i.e., does an association between two variables exceed what would be expected
based on their independent distributions in the population?)
Methodological artifacts (i.e., is the relation between two variables attributable, in
part, to common methodological influences, such as social desirability biases in self-report, or
inadvertent common content overlap among measures being correlated?)
Spuriousness (i.e., are two variables related not because of a causal or other
one-to-one association between them but because they are both influenced by a common third
variable?) For example, students who drink heavily tend to report lower grades and have more
academic problems; however, much of this association may be explained by precollege academic
achievement (Wood PK et al., 1997).
Directionality (i.e., is a variable a cause or a consequence of drinking?) For example,
a study that attempts to demonstrate a causal influence of fraternity or sorority membership on
student drinking needs to address the rival hypothesis of self-selection. That is, students who
join Greek organizations may already be heavy drinkers who seek the heavy-drinking lifestyle
associated with fraternity or sorority membership.
Evaluation Research
Evaluation research attempts to explore whether a given program, such as one instituted to lower
alcohol use by college students, is achieving its stated goals. Process evaluation examines
whether the program is operating as planned, whereas outcome evaluation assesses whether the
program achieves the impact or effects that were planned for it.
Issues in Research Design
The specific study designs used by researchers can affect the findings that are produced (Dowdall
and Wechsler, 2002). Issues that must be considered in the design of research to measure drinking
among college students include:
Unit of analysis that is used (e.g., the student, categories of students, the college);
Types of students and universities selected for study;
Diversity among institutions;
Whether the study design is longitudinal or cross-sectional;
Type of data collected;
Sampling techniques and the size of the population samples; and
Validity of the data, including the reliability and validity of self-reports.
Although the validity of self-reported alcohol use and problems is a legitimate concern, most
validation studies show that self-reports are basically accurate. Not surprisingly, given the
limitations of human memory, reports of recent alcohol consumption tend to be more accurate than
reports of long-term patterns of use (Babor et al., 1987; Harris et al., 1994; Midanik, 1988).
Just as binge drinking rates vary widely, so do the institutional characteristics of the
thousands of colleges and universities in the United States. The broad organizational diversity of
higher education poses challenges in study design as institutional factors may influence research
findings. The researcher must ensure that a study sample is representative of the college or
university (or subgroup of students) that is being studied. Care must be taken not to generalize
from samples that are not representative, and comparisons over time must be made cautiously as
higher education and the college student population in the United States have changed over the
years (Dowdall and Wechsler, 2002).
Ideally, researchers would like to study change over time (longitudinal studies) to identify
trends in a behavior, such as drinking among college students. Issues of cost and practicality
often limit use of this option. Although there are few longitudinal studies of college drinking,
those that exist elucidate patterns of stability and change over time, such as the increased risk
of alcohol-related problems in middle age associated with earlier alcohol-related problems in
college (Vaillant, 1996).
Data Collection
Surveys, using both questionnaires and interviews, and focus groups have been used with varying
success to collect information about alcohol use among college students. Surveys can be conducted
face to face, by mail, by telephone, and over the Internet.
Several large national data sets (see Epidemiology of Alcohol Use Among College Students)
are available for researchers conducting secondary analyses of data already collected and stored
in accessible databases. An excellent summary of these generally high-quality data sets has been
published (Larson et al., 1995). In addition, the Internet has made access to these databases much
easier, and many of them have their own Web sites.
Linking Alcohol Use to Harmful Outcomes
Research on alcohol use among college students tends to employ measures indicating heavy episodic
use. This pattern of drinking has come to be known as binge drinking, a term akin to “eating
binges” and “shopping binges” that denotes a particular behavioral pattern occurring over a
relatively short period (see Glossary of Alcohol Terminology).
Research studies that link alcohol use in college to health and behavioral consequences have
examined primary effects for the heavy-drinking college student, such as missing a class, being
unable to study, or having unprotected sex, and secondary effects from others’ drinking, such as
being physically assaulted or having study time interrupted. Self-report measures have been used
to assess academic consequences, drinking and driving issues, and abuse or dependence.
Several short screening forms for alcohol-related problems that were originally
developed for clinical practice have been used in college studies. These include
the CAGE Questionnaire, which is composed of four fairly severe indicators of
alcohol dependence (Mayfield et al., 1974), and the Michigan Alcoholism Screening
Test, composed of 24 items sampling various problems and dependence symptoms
(Selzer, 1971). More recently, assessments of alcohol dependence symptoms and
problems tailored to college students and other young adults have been developed.
These include the 23-item Rutgers Alcohol Problem Index (White and Labouvie,
1989) and the 27-item Young Adult Alcohol Problems Screening Test (Hurlbut and
Sher, 1992). Summaries of these and other problem drinking assessments can be
found in NIAAA’s Treatment Handbook Assessing Alcohol Problems: A Guide for
Clinicians and Researchers (NIAAA, 1995). It is important to note that the
self-reported prevalence of self-perceived alcohol problems (e.g., “Do you think
you have an alcohol problem?”) is relatively low even among those who report
experiencing severe consequences of drinking. This suggests the need to assess
objective problems (e.g., blackouts, fights, and injuries) as well as a student’s
self-assessment of his or her drinking problems. Research designed to assess
alcohol abuse and dependence among college students is a high priority because
of the critical importance of identifying students who may need formal treatment
for alcohol problems (see Glossary
of Alcohol Terminology).
Strategies for Filling Gaps in Knowledge: Methodology
The studies of college drinking conducted over the past 50 years are uneven. Although many meet
high scientific standards, not all have adhered to rigorous research designs, and weak studies
have often drawn inferences that cannot be strongly supported. This variability in study quality
affects both internal validity (i.e., how confidently inferences can be drawn about the
relationships among variables in the study undertaken) and external validity (i.e., how
generalizable the findings are to different campuses or subpopulations of students). In recent
years, research on college drinking has become more sophisticated and rigorous. Solicitations
(e.g., program announcements, requests for applications) from NIAAA, other Federal agencies, and
private foundations could accelerate improvements in the quality of research on college drinking
by ensuring that projects important to the field meet high scientific standards.