The Drinking Patterns and Problems of a National Sample of College Students, 1994
Adapted from: Engs, R. C., Hanson, D.J., Diebold, B.A. (1996) "The Drinking Patterns and Problems of a National Sample of College Students, 1994," Journal of Alcohol Drug Education, Spring, 41(3):13-33. Also found on permanent IUScholarWorks Repository: http://hdl.handle.net/2022/17496
Over 12,000 university students from every state were administered the Student Alcohol Questionnaire during the 1993-1994 academic year. Of all students 72.0% consumed alcohol at least once a year and 20.6% were heavy drinkers (consuming 5 or more drinks per occasion once a week or more)["binge drinking" is referred to as heavy drinking in this article]. A mean of 9.6 drinks per week was consumed by all students in the sample, 31% of males consumed over 21 drinks per week and 19.2% of females consumed over 14 drinks a week. Of the drinkers, 28.4% were heavy and 71.6% were light to moderate drinkers and they consumed a mean of 10.9 drinks per week. A significantly higher proportion of men, Whites, under 21 year olds, Roman Catholics, individuals to whom religion was not important, individuals with low grade point averages, fraternity/sorority members, students attending college in the North East part of the United States, in small communities, private schools and colleges under 10,000 students exhibited heavier drinking and a higher incidence of problems related to drinking. These results are similar to other studies which have been accomplished over the past two decades. The results do not support dramatic changes in the demography of heavier drinkers within most demographic categories.
It was concluded that traditional demographic variables need to be taken into consideration when planning campus educational and prevention programs. In times of limited budgets, the primary target needs to be these high risk students.
Educational efforts, prevention programming and comprehensive policies concerning alcohol consumption at the university level have increased over the past decade, largely due to increased funding from the federal level. However, this funding is likely to decline. With more limited resources, universities may need to find the most efficient strategies for delivering alcohol abuse prevention (Gonzalez 1993). Rather than aiming efforts at the total student body, limited funds would be better spent on programming for those at greatest risk for alcohol abuse.
Society is constantly changing and groups at greatest risk in an earlier era may no longer be so. Moreover, educational and prevention efforts considered important in the late 1980's may no longer be relevant to the mid 1990's. It is common knowledge that there have been numerous changes in the structure of American society. These include changes in gender roles and behavioral expectations, changes in socio- economic status of racial and ethnic groups, increased religious intermarriage, and social pressures for earlier maturity of youth. In addition, our society has seen changes in the law concerning alcohol use and the decrease in rural urban differences(Stark, 1994).
Because of possible changes in drinking patterns within these demographic categories, the purpose of this study was to gather new baseline information which could be used for curriculum development. A second purpose of this descriptive cross-sectional study was to test the hypothesis that demographic variables are less important now than in the past in relation to drinking behaviors. The null hypotheses is that there are no longer any differences in drinking patterns or problems within the different demographic categories.
Review of the Literature
Societal changes in the United States may be reflected in drinking patterns and problems. There is mixed evidence regarding the importance of demographic variables in relation to drinking among the college population. Various personal, academic and social characteristics have been associated with drinking and drinking problems and gender has been one of the most important predictors of these phenomena. The majority of studies have shown that a higher percentage of men drink and experience drinking-related problems than women (Engs and Hanson, 1990; Loughlin and Kayson, 1990; Saltz and Elandt, 1986; Engs and Hanson 1985). In addition, recent studies (Billingham, Post and Gross 1993; Gustafson 1993; and Robinson, Gloria, Roth, and Schuetter 1993), have reported that men generally consume alcohol more frequently and/or in greater quantities than women. Other investigations have disputed this, however, finding little or no difference between males and females (Kodman and Stumack, 1984; Berkowitz and Perkins, 1985; Perkins, 1992).
Another important variable which is predictive of drinking patterns and problems related to drinking is racial or ethnic background. Older studies in the United States have reported non-whites having a higher rate of heavy drinking than whites (Maddox and Williams, 1968). Current research (Wechsler, Dowdall, Davenport and Castillo, 1995; Williams, Newby, and Kanitz 1993; Crowley 1991; Schall, Weede, and Maltzman 1991; and Hanson and Engs 1990) has shown that non-white college students report lower rates of both alcohol consumption and drinking-related problems. On the other hand, one study showed drinking rates equal to that of whites (Conners, Maisto, and Watson, 1989).
The relationship between drinking and religious affiliation suggests the highest proportion of drinkers are typically found among Jews, a slightly a lower rate among Catholics and the lowest among Protestants. This has been found in American (Carlucci, et al. 1993), American and Canadian (Engs, Hanson and Glicksman, 1990) and a Scottish(Mullen, Blaxter, and Dyer, 1986) sample. Some reports (Miller and Garrison, 1982; Engs and Hanson, 1985) also indicate a direct relationship between the lack of importance of religion and frequent or heavy drinking but not all (Reiskin and Wechsler, 1981).
While positive association between both quantity and frequency of drinking with both age and with college year have been documented (Wechsler, Dowdall, Davenport and Castillo, 1995; Engs and Hanson 1985,1989; and Crum, Helzer and Anthony, 1993), some studies have reported either relatively little difference or negative association with college year (Lotterhos, Holbert, and Glover, 1990; Schall, et al. 1991; Gross, 1993). An inverse relationship between student drinking and academic achievement has been reported by numerous studies (Engs and Hanson 1985; Ford and Carr, 1990; Borges and Hansen, 1993). Evidence suggests that pledges or members of sororities and fraternities report greater rates of alcohol consumption and drinking-related problems than non-Greeks (Kodman and Sturmak, 1984; Tampke 1990).
Institutional characteristics are also associated with different drinking patterns. The frequency and quantity of drinking are lower in the South, in urban areas, in large and in public institutions. However, these differences may be waning (Engs and Hanson, 1985; Centers for Disease Control and Prevention, 1994).
A pre-coded anonymous instrument, the Student Alcohol Questionnaire (SAQ), was used to collect data (Engs 1977; Hanson 1972). The questionnaire includes various demographic items; six questions concerning quantity and frequency of wine, spirits and beer consumption; and 19 items regarding possible negative health/personal, social/academic, legal/violence or drinking/driving consequences resulting from alcohol consumption. The SAQ also contains alcohol knowledge and attitude questions.
For this report, 19 items regarding possible consequences of alcohol consumption are reported. The items are listed in Table III. Students were asked to indicate if a given problem had occurred at least once during the preceding year. Six items used to calculate quantity-frequency and mean amount of alcohol consumed were also utilized.
Instructions explained the voluntary nature of participation, as approved by the authors' Institutional Review Boards. The instrument has been widely used or adapted by a number of authors. Some recent examples include Hong and Isralowitz(1989), Maney(1990), Hughes and Dodder(1992), Carlucci, Genova, Rubackin, and Kayson(1993), Flynn and Brown(1991), Gross(1993) and Haworth-Hoeppner, Globetti, Stem, and Morasco(1993) among others in the United States and other countries. The instrument has demonstrated internal consistency reliability of .79 for all items, excluding demographic factors. An updated reliability analysis (Engs and Hanson 1994) has demonstrated Spearman-Brown reliability coefficients of .84 for the Quantity/Frequency and .89 for the Problems Related to Drinking sub-scales. The values of Cronbach alpha reliability were .86 and .92 respectively, for these sub- scales.
This sample is part of an ongoing study of drinking patterns and problems of students attending four-year colleges and universities from every state in the United States that was begun in 1982. A number of studies have been published over the past 15 years from the five times data have been collected(See references for citation by the author's for some examples). Institutions, were selected to form a "quota sample". Colleges were chosen to be representative of all four-year institutions of higher education in terms of financial support(public or private) and size(over and under 10,000 student enrollments). For example, approximately 65% of students attend state supported schools in terms of financial control in the United States(Snyder, 1993). This same proportion of institutions, from each state, were randomly selected from a list of colleges and universities which had health, physical education or sociology departments(Simon, 1987; College Blue Book, 1993; Eta Sigma Gamma, 1992). The department head was then contacted about participation in the study. If an institution declined to participate, another institution with similar demographics, eg, state supported, small community, with over 10,000 students, in the same state was then asked to take part.
For collecting the sample, the authors asked faculty, identified by the departmental chair, who teach general elective survey-type classes for which all students in all academic years are eligible. Examples would include courses such as personal health, first aid, and basic sociology. These faculty were asked to distribute the SAQ to students for in-class completion and to return the completed questionnaires to the researchers. The return rate for complete and usable questionnaires exceeded 97%.
Due to the fact that the whole class was surveyed, this "convenience sample" is limited to students in classes from institutions where instructors were willing to distribute the questionnaire.
The resulting sample consisted of 12,081 students from 168 colleges and universities and included the following Personal demographic characteristics: 38.5% men and 61.5% women, 81.6% white, and 18.4% non-white; 31.8% Roman Catholics, 27.3% Protestants whose religions allowed drinking, 21.5% Protestants whose religion does not allow drinking, 1.9% Jews, and 17.5% none or other. 57.8% were under and 42.2% were over the age of 21 years. The mean age was 20.5. The sample over- represents females and non-whites compared to national statistics regarding the demographics of college students (Snyder, 1993)a.
Among Academic and Social Behavioral characteristics, the sample consisted of 27.7% freshmen, 23.8% sophomores, 24.6% juniors, and 23.9% seniors. 16.2% were pledges or members of a fraternity or sorority.
Institutional characteristics: twenty-four percent of the sample attended school in the Northeastern region of the United States, 28.9% in the North Central region; 30.0% in the Southern, and 17.1% in the Western region (including Alaska and Hawaii) of the United States. Of the total sample, 68.1% were from colleges with enrollments of 10,000 or more, and 31.9% were from colleges with a student body of less than 10,000; 81.7% were from public and 18.3% of were from private institutions; 69.5% were from schools with surrounding communities of populations under 100,000, 20.0% from communities between 100-500,000, and 10.5% were from surrounding communities with populations over 500,000.
A limitation to the study was that the sample over-represents females, non-whites and those who attended public schools compared to the universe of students attending four year institution of higher learning in the United States (Snyder 1993)a. Because of its large size, the sample had high power for detecting significant differences. On the other hand, the large sample size also introduces the chance of type I error.
All calculations were accomplished on the Indiana University VAX computer using the SPSS program (Norusis, 1990).
Mean number of drinks per week
Several methods for calculating the amount of alcohol consumed are in common use. They include calculating the mean grams or ounces of absolute alcohol or the mean drinks or units per week or per day. In self report studies, determining grams or ounces is often an imprecise calculation as it is based upon recall. In addition people often underestimate the amount they have consumed (Thomas, Goddard, Hickman and Hunter 1993).
Therefore, in recent years it has become more common to calculate the mean number of drinks, or units, per week or day of all alcoholic beverages consumed in North America and Great Britain (Lemmon, Tan and Knibbe, 1988; Engs, Hanson, Glicksman and Smythe, 1990; Thomas, Goddard, Hickman and Hunter 1993; Engs 1990; Engs and Hanson, 1994; Gaziano, Buring, Breslow, Goldharber, Rosner, VanDenburgh, Willett and Hennekens, 1993). Calculations for this method are based upon the "rule of thumb" that an average can or glass of tavern beer(12 ounces) is roughly equivalent to an average size glass of wine (5 ounces) or shot of spirits (one and half ounce) in terms of grams (approximately 13) of absolute alcohol(Consumer and Food Economics Institute, 1990).
For the calculations, the instrument assessed the usual frequency and quantity of beer, wine and spirits consumed by student. The frequency and quantity response categories were assigned constant values b. To compute the total number of drinks consumed on a weekly basis, a mean score was calculated by multiplying the re-coded quantity by the re-coded frequency weight for each beverage type and summing the three scores. A One-Way Analysis of Variance and the t-test was used to compare the mean number of drinks within demographic variables. The post-hoc Scheffe test was used to determine where differences occurred.
Calculating "at risk drinking" for males and females
Several recent reports suggest that up to 21 drinks per week for males and 14 drinks for females is considered the maximum safe consumption limit in terms of acute and chronic health consequences(Engs and Aldo-Benson, 1995; Cohen, S., Tyrrell, D.A., Russell, M.A., Jarvis, M.J., and Smith, A.P., 1993; Garg, Wagener, and Madans, 1993; Gaziano, J.M., Buring, J.E., Breslow, J.L., Goldharber, S.Z,. Rosner, R., VanDenburgh, M., Willett, W. and Hennekens, C.H., 1993; Bofetta and Garfinkel, 1990).
Chi-Square analysis was accomplished for males and females comparing the percent who had consumed over the maximum amount recommended for their gender. These students were defined as at risk drinkers. In contrast, males who consumed 21 and females who consumed 14 drinks or less per week during the previous 12 months were considered low risk drinkers.
Quantity/Frequency drinking level
Based upon a method suggested by Cahalen(1969) and adapted by Engs(1977), a quantity/frequency level of drinking was calculated to identify different levels of drinkers for the total group. Individuals were divided into three categories: Abstainers, Light to Moderate, and Heavier Drinkersc. These were analyzed by Chi-Square analysis.
Drinking related problems
Only students who had consumed any alcohol during the previous 12 months, ie, "drinkers", were asked to report on problem behaviors associated with drinking. Students who had not consumed any alcohol during the previous 12 months were asked to skip these items. A mean problem score was calculated for each student by assigning one point for each of the 19 problems experienced at least once during the previous 12 months. These scores were subjected to t-tests, and one-way analysis of variance and post-hoc Scheffe tests. In addition Chi-square analyses were used to determine possible differences in the percentages of students exhibiting each of the 19 problems for gender among low risk and among at risk drinkers. Males consuming over 21 and females consuming over 14 drinks per week were select into the at risk drinking category. Likewise, males and females under this level were selected out for the low risk category.
Quantity-Frequency and mean drinks per week
Of the total group, 72.0% were drinkers. One in five were considered Heavy Drinkers and half were classified as Light/Moderate Drinkers. Of drinkers, 27.4 % were Heavy Drinkers. The mean drinks consumed per week for the total sample of students was 9.6. For drinkers only it was 10.9 drinks per week. There was a significant difference (p < .001, t=8.8 ) between the mean drinks consumed between the total population and the drinkers.