Multilevel analysis of situational drinking among Canadian undergraduates
Introduction
Drinking is an integral part of campus social life, and the university milieu carries its own set of norms, opportunities and social influences regarding alcohol consumption. Consequently, the campus drinking culture is often described as a “wet” culture, and the university students’ drinking style is generally characterized as a heavy episodic or a binge drinking style (Gliksman, Demers, Adlaf, Newton-Taylor, & Schmidt, 2000; Wechsler, Lee, Kuo, & Lee, 2000).
By far, most studies have attempted to understand campus drinking based on individual-level variables. Student drinking patterns have been related to (1) socio-demographic characteristics such as gender, religiosity, ethnicity and marital status (Berkowitz & Perkins, 1986; Brennan, Walfish, & AuBuchon, 1986a; Prendergast, 1994; Saltz & Elandt, 1986), (2) psychological factors such as personnality, expectancies, attitudes, beliefs and motivation (Klein, 1994; O’Hare, 1998; Prendergast, 1994; Schall, Kemeny, & Maltzman, 1992), (3) child and youth development and family background (MacDonald, Fleming, & Barry, 1991), (4) peer-influence (Brennan, Walfish, & AuBuchon, 1986b; Martin & Hoffman, 1993), and (5) individual experiences of university life such as the perception of the campus drinking culture, involvement in social activities and living arrangement (Brennan, Walfish, & AuBuchon, 1986b; Gfroerer, Greenblatt, & Wright, 1997; Gliksman, Newton-Taylor, Adlaf, & Giesbrecht, 1997; Igra & Moos, 1979; O’Hare, 1990; Perkins & Berkowitz, 1986; Perkins & Wechsler, 1996; Wechsler, Dowdall, Davenport, & Castillo, 1995). Some have attempted to assess the impact of ecological campus factors on individual drinking despite inferential difficulties (Bell, Wechsler, & Johnston, 1997; Chaloupka & Wechsler, 1996).
The statistical models employed by most studies investigating student drinking patterns implicitly assume that the sole influences on drinking behaviour are individual traits and that drinking occasions are fairly invariant within individuals. These assumptions are generally unfounded. Indeed, although heavy drinking episodes are prevalent among university students, not every episode is excessive. The issue of occasion and individual drinking heterogeneity is long-standing. As suggested by Harford (1979), “the consumption of alcoholic beverages is situationally specific, rather than a trans-situational property of specific individuals” (p. 289). Drinking occurs in a variety of situations that are normatively marked. Moreover, as noted by Bacon (1957), “the rules and procedures are on occasion rather specific, but also show enormous variability so that a given individual may follow one set of rules with his family, another with business or professional associates, and a third on holiday occasion….” (p. 179–180). Hence, it might be expected that the variation in drinking is, at least in part, related to the drinking setting.
Individuals do interact with the social environment in which they belong. Sociologically, the drinking setting can be seen as the environment in which the act of drinking occurs. Each setting carries its own set of rules and norms regarding drinking in terms of normal-deviant drinking (appropriateness of drinking) as well as in terms of beverage choice (form of drinking). These are reinforced through social interaction, and thereby normatively regulates the alcohol intake (Klein & Pittman, 1990; Simpura, 1991). Arguably, the apparent relationship between setting and alcohol intake is an artefact of a self-selection process, i.e., heavy drinkers are more likely to find themselves in heavy drinking situations. However, self-selection cannot be the sole determinant because the alcohol intake of any given individual is not invariant across drinking events. Thus the self-selection hypothesis cannot explain the variability in the individual drinking in different settings. Furthermore, it has been clearly established, particularly by research on the modelling of alcohol consumption, that a social influence process is at play in drinking (Quigley & Collins, 1999).
Studies assessing situational influences related to student drinking remain rare and underdeveloped. Nevertheless, the results of these studies suggest that student drinking is related to the setting of the drinking occasion, such as why, where, when and with whom students consume alcohol. Students drink more in bars, in group settings, particularly at large parties, and in same-sex groups (Perkins & Berkowitz (1981), Harford, Wechsler, & Rohman (1983); Perkins & Berkowitz, 1986; Rosenbluth, Nathan, & Lawson, 1978). Clapp, Shillington and Segars (2000) reported that binge drinking occasions occur mainly on Saturday, with friends, in the absence of partners or spouse, in bars as well as in private homes, on dates or socializing and in parties. Harford et al. (1981) suggested that the usual decline in student drinking during college may be related to changes in setting: as students move from group drinking to more intimate settings, they seem to reduce their consumption. Finally, contextual gender differences show that women are more likely than men to drink in family and in restaurant settings (O’Hare, 1990), whereas men drink in a wider range of activities and locations than do women (Biber, Hashway, & Annick, 1980).
The relationship between drinking setting and situational consumption has also been shown among the general population. As for university students, research has stressed the importance of the symbolic meaning, physical, temporal and relational dimensions of the setting, on the situational alcohol intake (Clark, 1984; Demers, 1997; Harford, 1983; Hennessy & Saltz, 1993; Single & Wortley (1984), Simpura (1987); Single & Wortley, 1993; Sykes, Rowles, & Schaefer, 1993).
Previous studies contributed substantially to our understanding of how people drink in settings in which alcohol is commonplace. They provided a thorough description of drinking patterns in various interpersonal environments such as the group size and the group composition (Hennessy & Saltz, 1993; Rosenbluth et al., 1978; Sykes et al., 1993). They also examined the role of individual characteristics in predicting the level of consumption in various settings (Harford, 1979; Single & Wortley (1984), Simpura (1987); Single et al., 1993). Even though they all acknowledge that drinking patterns are the results of interactions among setting and individual, they treated the drinking situations and the individuals at one single level of analysis. When treating individuals as the unit of analysis, they ignore the variation due to the occasion. This can be resolved crudely by studying a single occasion (most recent occasion approach), but again this assumes occasion factors to be invariant. On the other hand, when treating occasion as the unit of analysis, they ignore individual variation. This is indeed problematic given the socio-ecological conception of health behaviours, which argues that similar types of people may not behave in the same way under different settings. The major contribution of this study will be to simultaneously assess the predictive values of individual characteristics and setting characteristics on drinking in a multilevel approach. Using a multilevel modelling of the data to disentangle the contribution of the individual factors and of the situational factors will represent more adequately the determinants of the drinking act (Hox, 1994).
From our perspective, alcohol intake per occasion is best viewed as a function of both individual and situational characteristics. Therefore, alcohol intake on a given occasion might vary according to the setting in which drinking takes place as well as according to the individual's characteristics. In this study, drinking situations and individuals are conceptualized as hierarchical relations in which drinking occasions are nested within the individual, and both the setting of the occasion and the individual characteristics determine the quantity of alcohol consumed. Although the most common application of multilevel models is with nested organizational or geographical structures, such as students within schools, multilevel models can also analyse data that are nested within individuals, such as repeated measures and multiple responses (Duncan, Jones, & Moon, 1996).
Situational factors are conceptualized as the primary determinants of per occasion intake, that is, explaining variations between individuals as well as between various drinking settings (within individual or between occasions variation). Individual factors are conceptualized as secondary determinants of individual drinking patterns. Defined at a higher level, individual factors may only explain residual variance between individuals, as these factors remain constant for a given individual despite the setting in which he/she drinks. These factors may also modify the relationship between drinking setting characteristics and per occasion alcohol intake.
A full theoretical model might include interaction effects to capture the complexity of the interwoven action of situational and individual factors. However, given the scarcity of previous research and the exploratory nature of this study, the main contribution of this paper will be firstly to disentangle and estimate the relative contribution of the setting and of the individual levels on drinking and, secondly, to determine the direct effect of situational and individual characteristics. However, as gender is a major determinant of drinking behaviour (Engs & Hanson, 1990; Wilsnack & Wilsnack, 1997), the interactions between gender and the other factors will be explored.
Section snippets
Sample
The data analyzed in this paper are based on a hierarchical structure of drinking behaviour. From a sample of university undergraduate drinkers, 26,348 drinking occasions (level 1), nested within 6,850 individuals (level 2) were analysed. Each respondent provided information for up to five drinking occasions, defined as the five most recent occasions since the beginning of the academic year (i.e., over an 8 to 12 week period).
These data were derived from the Canadian Campus Survey (CSS), a mail
Results
The results of the multilevel analyses are presented in Table 2. The null model (Model 1) indicates an overall mean alcohol intake per occasion (log) of 1.48 (s.e.=0.59), i.e. 4.4 drinks per occasion in original units, and finds sizeable variance for both setting and individual characteristics. The intraclass correlation is 0.49, indicating that 49% of the variance in alcohol intake per occasion is between individuals and 51% is between occasions.
Model 2 evaluates the effects of the situational
Discussion
This study examined the effects of individual factors, particularly the individual experience of university life, and of situational factors on alcohol intake per occasion. The hierarchical structure of our data, where drinking occasions are nested within individuals, allows us to conduct a multilevel analysis to disentangle the situational and individual effects.
The drinking situation, why, where, when and with whom students drink, appears to have an important effect on situational alcohol
Conclusion
The analysis presented here indicates that the drinking situation has a substantial effect on alcohol intake. It is apparent from our findings that the individual cannot be conceptualized as an autonomous actor making self-governing decisions in a social vacuum. From a prevention perspective, the major implication of this study is that programs or policies aiming to reduce student drinking may be more beneficial if they address both individual and situational factors, concurrently or
Uncited References
Rasbash and Woodhouse (1995).
References (54)
- et al.
Health-related behaviour in contextA multilevel modelling approach
Social Science & Medicine
(1996) - et al.
The influence of households on drinking behaviourA multilevel analysis
Social Science & Medicine
(1998) - et al.
Predicting small-area, health-related behaviourA comparison of smoking and drinking indicators
Social Science & Medicine
(2000) Social settings conducive to alcoholism
The Journal of the American Medical Association
(1957)- et al.
Correlates of college student marijuana useResults of a US national survey
Addiction
(1997) - et al.
Problem drinkers among college studentsA review of recent research
Journal of American College Health
(1986) - et al.
Drinking patterns of male and female collegiansAre the patterns converging?
Journal of Youth and Adolescence
(1980) - et al.
Alcohol use and abuse in college students. I. A review of individual and personality correlates
The International Journal of the Addictions
(1986) - et al.
Alcohol use and abuse in college students. II. Social/environmental correlates, methological issues, and implications for intervention
The International Journal of the Addictions
(1986) - et al.
Hierarchical linear modelsapplication and data analysis methods
(1992)
Binge drinking in collegeThe impact of price, availability, and alcohol control policies
Contemporary Economic Policy
Deconstruction contexts of binge drinking among college students
American Journal of Drug and Alcohol Abuse
Alcohol use in various setting
When at risk? Drinking contexts and heavy drinking in the Montréal adult population
Contemporary Drug Problems
Why are some people healthy and others not?.
Alcohol and other drugs use by Ontario University studentsThe roles of gender, age, year of study, academic grades, place of residence and programme of study
Drugs Education Prevention and Policy
Canadian campus survey 1998
Substance use in the US college-age populationDifferences according to educational status and living arrangement
American Journal of Public Health
Multilevel mixed linear model analysis using iterative generalised least squares
Biometrika
Multilevel statistical models
Contextual drinking patterns among men and women
A contextual analysis of drinking events
The International Journal of the Addictions
Contextual drinking patterns of college studentsthe relationship between typical companion status and consumption level
The structural context of college drinking
Journal of Studies on Alcohol
Modeling social influence on public drinking
Journal of Studies on Alcohol
Young families and alcohol use in Finland and the Soviet Union
Contemporary Drug Problems
Applied multilevel analysis
Cited by (125)
A social contextual review of the effects of alcohol on emotion
2022, Pharmacology Biochemistry and BehaviorAlcohol ‘use’ and ‘abuse’: when culture, social context and identity matter
2020, Current Opinion in Food ScienceCitation Excerpt :Furthermore, among French students, alcohol consumption is perceived more favorably when occurring in social groups than when alone [16], and the number of alcohol glasses reflecting a normal consumption in a group situation is significantly greater than the number of glasses reflecting a consumption denoting an addiction in a solitary context [22]. Taken together, these points highlight both the functional character of context [30,9] and representations [25]. This one contributes to the interpretation of situations [31], and impacts evaluations of drinking behaviors and judgments about individuals adopting these behaviors [17,11•].
Understanding social factors in alcohol reward and risk for problem drinking
2019, Psychology of Learning and Motivation - Advances in Research and TheoryCharacteristics of drinking events associated with heavy episodic drinking among adolescents in the United States
2017, Drug and Alcohol Dependence“It Always Depends on the Context”: Poker Players' Perceptions of Substance Use at and Beyond the Table
2023, Critical Gambling Studies