Students define a "drink" differently than researchers
       Each year, dozens of papers are published
regarding alcohol use among college students.
In the vast majority of these papers, the amount
of alcohol consumed by students is estimated
based on self-report survey data. The data from
such surveys are used to investigate a wide variety
of issues, including thresholds for risky drinking
(Wechsler et al., 2002), consequences of alcohol
consumption (White et al., 2002), normative
drinking levels (Perkins, 2002), and longitudinal
changes in alcohol use (Johnston et al., 2002;
Wechsler et al., 2002).
      Alcohol surveys typically provide students with
a definition of a drink and then ask them to formulate
answers to questions regarding their drinking habits
with that definition in mind. The definitions provided
to students vary widely. Currently, the Harvard
CollegeAlcohol Study (CAS) is the only major
project to use a survey that provides volumes in the
definitions of more than two types of drinks. The
CAS survey asks students to define a standard
drink as a 12 oz beer, a 4 oz glass of wine, 1.25
oz of liquor in a shot, or 1.25 oz of liquor in a mixed
drink (Wechsler et al., 2002). The values for wine
and liquor are somewhat smaller than those used by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), the U.S. Department of Health and Human Services (DHHS), and the U.S. Department of Agriculture (USDA), which define a standard glass of wine as 5 oz and a standard serving of liquor as 1.5 oz (NIAAA, 2000; DHHS and USDA, 1995).  See the figure below for a summary.

       The use of survey data to characterize college drinking requires several assumptions to be made. One such assumption is that researchers and students define the term “drink” in comparable ways. That is, if a student indicates on an alcohol survey that he/she consumed four or five drinks the night before, the assumption must be made that they are using the term drink in a manner consistent with the definitions provided in the survey instrument. Unfortunately, this does not appear to be the case. We recently published a study suggesting that students define drinks more liberally than researchers and government agencies (White et al., 2003). Students were asked to pour water into cups of different sizes and approximate the volumes of a standard beer, shot, or the amount of liquor in a mixed drink. In every task, students’ drinks were significantly larger than the standard drink volumes used by the Harvard School of Public Health, as well as the more liberal definitions used by NIAAA.

        The fact that students overestimate how much alcohol represents a standard drink strongly suggests that they underestimate their own levels of consumption. As a consequence, assessments of college drinking based on self-reported drinking habits probably underestimate the actual drinking levels of college students.
       Why do college students overestimate standard drink volumes? It is entirely possible that they are aware of the correct volumes of standard drinks but simply have difficulty estimating how much fluid to pour into cups to approximate the correct volumes. For instance, a student might know that a standard beer should contain 12 oz of fluid, but have great difficulty recognizing when 12 oz has been added to a cup. In support of this possibility, recent studies indicate that the size and shape of a cup can influence perception of volumes (Wansink and Ittersum, 2003). Alternatively, the super-sizing of drinks might be related to a general lack of knowledge regarding the appropriate volumes of standard alcoholic beverages. A student might pour four or more ounces of liquor into a cup for a mixed drink because they believe that four ounces is the correct amount to add.
       To address these issues, we assessed the relationship between students’ knowledge of standard drink volumes and the amount of fluid that they add to cups when asked to pour standard drinks. Students completed an alcohol survey and answered questions regarding their knowledge of standard drink volumes. They then completed a series of tasks in which they poured fluid into cups of various sizes to approximate the volumes of standard drinks. It was hypothesized that inflated definitions of standard drinks are related to over-pouring in the free-pour tasks.
       With the exception of beer, students incorrectly defined the volumes of standard servings of alcohol. When asked how many ounces constitute a standard glass of wine, shot, or the amount of alcohol in a mixed drink, students overestimated these volumes compared to values used in the Harvard CAS and by the NIAAA. When asked to free-pour water into empty cups to reflect the volumes of standard drinks they filled the cups too high. As in a previous study (White et al., 2003), the volumes poured by students increased with the size of the cups. Significant relationships existed between the amounts of alcohol that students thought should be present in single drinks and how many ounces they free-poured. Such findings suggest that students might over-pour drinks due to a lack of knowledge of commonly used standard drink volumes.

              
      The findings of the study raise questions about the validity of students’ responses to standard survey instruments. For instance, a student who indicates having consumed five drinks in one night might actually have had ten or more standard drinks if the drinks were free-poured. If only a small proportion of the drinks consumed by students were free-poured, then errors in their estimates of standard drink volumes might not be an important factor. However, this does not appear to be the case, at least not for the students in our studies. Roughly half of all drinks consumed by students in the two weeks prior to the studies were free-poured, raising the possibility that their actual alcohol consumption was considerably higher than suggested by their survey responses. In studies employing cutoffs for risky drinking practices (e.g., Wechsler et al., 2002; Presley et al., 1999), slight increases in reported drinking levels can be enough to cause a significant shift in the number of students falling into the risky drinking category (Dawson, 1998). Thus, for researchers interested in risk assessments, closing the gap between student-defined standard drink volumes and researcher-defined standard drink volumes is critical.