MASSACHUSETTS INSTITUTE OF TECHNOLOGY AN INVESTIGATION OF FRESHMEN DRINKING HABITS 17.871 POLITICAL SCIENCE LABORATORY GROUP PROJECT BY [NAMES REMOVED TO PROTECT THE INNOCENT] CAMBRIDGE, MASSACHUSETTS [IN THE PAST]1INTRODUCTION College drinking is a major concern to many universities, federal authorities, and parents throughout the United States. Public and private institutions must cope with high insurance rates and numerous lawsuits as a result of problems with underage drinking, especially that done by college freshmen. An important question thus arises: when freshman live on campus, do they drink less than when they live off campus? Schools such as MIT have made the assumption that freshmen who live on-campus will have limited exposure to alcohol relative to those who live off-campus, and many colleges and universities have committed to policies specifying that freshmen must live on campus. However, empirical work has yet to demonstrate that freshmen actually drink less when they are forced to live on campus. Therefore, research correlating the effects of on- or off-campus housing with the drinking habits of freshmen seems justified to shed light on whether a policy of mandatory on-campus housing for freshmen is likely to be efficacious. Though the effects of freshmen residence on drinking habits have not been examined directly, there is a literature documenting various aspects of college drinking. Henry Wechsler, a researcher at the Harvard School of Public Health, has conducted a series of studies on college drinking. Wechsler’s first College Alcohol Study (CAS) was published in 1993 and prompted a dramatic change in the way in which the public viewed college drinking. The study brought the issue of heavy episodic alcohol use, or binge drinking, to the forefront of the public consciousness. In fact, the CAS prompted the Centers for Disease Control and Prevention and the US Surgeon General to declare alcohol abuse the number one public health issue affecting students in the 1990s (U.S. Department of2Health and Human Services, 2004). The Surgeon General even set a national goal of reducing the rate of college binge drinking by 50% by the year 2010 (U.S. Department of Health and Human Services, 2000). Furthermore, the problem of binge drinking gathered international attention when the World Health Organization convened a special conference on Young People and Alcohol in Stockholm in 2001. Finally, Wechsler’s studies have caused many school administrations to initiate or increase problem drinking prevention efforts. However, the effect of these preventive initiatives seems to have been minimal. Wechsler found that the percentage of students consuming 5 drinks or more in a row in the last two weeks changed insignificantly from 40.2% in 1993, to 40.0% in 1999. Additionally, in the 1999 CAS, he noted that under the existing patterns 31.6% of college students’ reported drinking behavior qualified them for a formal diagnosis of alcohol abuse (Wechsler et al., 1998). With these statistics in mind, in this paper we investigate the effect place of residence has on freshmen drinking, and we compare the effect of place of residence with the effects of living in alcohol-free housing and membership in a Greek organization. We first explain our methodology. We subsequently present and discuss the results obtained from the application of our methodology to data from Wechsler’s studies. Finally, we address the implications our findings have on university policies and conclude by defining their relevance in the larger picture of alcohol usage on college campuses. METHODS To answer the posed question, we examined data from the Harvard School of Public Health College Alcohol studies from 1993, 1997, and 1999 (Inter-university3Consortium of Political Science (ICPSR) datasets 6577, 3163, and 3818, respectively). The data were drawn from simple-random sample studies of undergraduate students at 140 four-year United States colleges and universities. From the codebooks for the ICPSR data we selected variables relevant to answering the question and evaluating rival explanations. Using our chosen variables, we created a STATA dataset aggregating the data from 1993, 1997, and 1999. We created a dummy variable to code for the year each observation came from. We dropped the observations of seniors, juniors, and sophomores from our dataset. To generate dummy variables “on” and “off” for freshmen place of residence, we coded place of residence responses in the following manner: those in university halls, dorms or other university housing were coded “on.” Those in fraternity or sorority houses, co-ops or university affiliated housing, and off campus were coded “off.” This coding decision is potentially problematic, as we do not know how many fraternity and sorority houses are physically located on college campuses or considered by colleges to be on-campus housing. To estimate the total alcohol consumed, we multiplied respondents’ number of drinking occasions in the past 30 days with the number of drinks consumed at each occasion. Recoding of the variable for number of drinking occasions in the past 30 days was necessary because the possible responses to the question were ranges of 0, 1-2, 3-5, 6-9, 10-19, 20-39, and 40 or more occasions. We chose the midpoints of the ranges as the estimated number of occasions for each range; for example, 1-2 became 1.5, 3-5 became 4, and so on. The “40 or more occasions” response presented a challenge to our recoding efforts because it had no midpoint. We selected 50 occasions to represent this4range. Similarly, because the drinks per occasion responses were 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9 or more, we recoded 9 or more as “13.” This may seem an indiscriminate technique, but we reasoned in choosing it that responses of 40 or more occasions and 9 or more drinks per occasion would be outliers in the dataset. Another way to have approached the problem would be to recode the occasion and drink responses into qualitative categorical variables of light, moderate, and heavy drinking. However, we desired a more quantitative and less qualitative estimation of drinking for our analysis purposes. We defined drinking “more” to be drinking a percentage of the total alcohol consumed greater than the population percentage of the group doing the drinking. To illustrate this point, if the on-campus