The socio-ecological context of U.S. college student drinking: A latent class analysis

Abstract

Background:

U.S. college student drinking typologies often consider quantity and frequency but not the socio-environmental contexts in which students obtain alcohol and drink. Understanding context could be important for preventive interventions.

Methods:

We used latent class analysis (LCA), a person-centered approach to understanding behavior patterns, to identify drinking typologies among 1390 college student drinkers from a representative survey at two interconnected private colleges in the Northeast. Classes were derived from drinking frequency and quantity as well as how students obtain alcohol, where they drink, and their perceptions of peer drinking. Resulting classes were correlated with demographic and developmental characteristics, participation in campus activities and connectedness, and alcohol consequences and protective behaviors.

Results:

Four distinct drinking profiles emerged. ‘Tasters’ (n = 290) included infrequent and low quantity drinkers who drank in dorms with alcohol provided by others. ‘Bargoers’ (n = 271) included low quantity and moderate frequency drinkers who purchased their own alcohol and drank at bars. ‘Partiers’ (n = 483) included moderate frequency and quantity drinkers who obtained alcohol from several sources and drank in many locations. ‘Bingers’ (n = 345) included high frequency and quantity drinkers and binge drinkers, who drank in many locations with alcohol obtained from multiple sources. Classes differed in demographics, age of first drink, campus activities and connectedness, alcohol protective behaviors, and alcohol problems.

Conclusion:

Heterogeneous patterns of drinking based on quantity, frequency and social/environmental context emerged and suggested the need for different tailored interventions.

1. Introduction

Alcohol use is prevalent during college (SAMHSA, 2019). Binge drinking, defined as 4 or more drinks for women and 5 or more drinks for men in a two-hour period (NIAAA, 2022) or on a single occasion (SAMHSA, 2019), is associated with physical injury, physical and sexual assault, drunk driving, alcohol dependence, and mental health problems such as depression, anxiety and suicidal ideation during the college years (e.g., Hingson et al., 2017; Patrick et al., 2020). Although most students outgrow binge drinking (Lee & Sher, 2018), some persist and experience alcohol dependence and related problems into middle and later adulthood (Jennison, 2004; O’Neill et al., 2001). Identifying typologies of college drinking and the socio-environmental factors associated with these typologies can provide vital information for the development, tailoring, and implementation of interventions to reduce drinking-related harms.

Although many college students consume alcohol, they vary in the frequency and amount they drink, and not all binge drink. Cross-sectional studies on college alcohol use have predominantly found three classes or “types” of student drinkers: abstainers, low frequency and quantity drinkers, and hazardous or high frequency and quantity drinkers (Beseler et al., 2012; Davoren et al., 2018; Yeater et al., 2018), although some studies have found four classes distinguished by co-occurring drug use (Chiauzzi et al., 2013) and temporality of alcohol use (Cleveland et al., 2012). Binge drinkers are most likely to experience alcohol-related consequences such as missing school, having unprotected and unintended sex, getting into fights, being assaulted and developing alcohol use disorders later in life (e.g., Davoren et al., 2018), and are least likely to engage in protective behaviors to mitigate drinking harms (Barry et al., 2016).

Drinking is embedded in social and ecological contexts, which are important to understanding risk (Braitman et al., 2017; Lewis et al., 2011; Wilkinson & Ivsins, 2017). Bronfenbrenner’s (1979) socio-ecological model posits that behavior is influenced by factors at individual, interpersonal, community, and societal levels. Contextual factors range from the physical environments where students drink (e.g., bars, dorms) to social factors such as who is buying/controlling the alcohol (buying for oneself vs obtaining at a party) and peer norms about drinking (Lipperman-Kreda et al., 2018; Litt et al., 2015; Martens et al., 2006). Compared to drinking in intimate settings such as dorms or restaurants, drinking in large social settings including bars, clubs, and parties is associated with heavier drinking and binge drinking (Braitman et al., 2017; Clapp et al., 2017; Pilatti et al., 2020; Wamboldt et al., 2019). Students who drink in intimate settings tend to consume lower alcohol-content beverages and may not always drink with the intention of getting drunk (Pilatti et al., 2020; Wamboldt et al., 2019).

Students may drink differently based on how they obtain alcohol. Many underage college students rely on friends and acquaintances of legal drinking age, use a fake ID, or obtain alcohol at a party (Fabian et al., 2008). Adolescents who purchase alcohol themselves and obtain from friends or adults are likely to be high-risk drinkers (Jackson et al., 2014). Students who overestimate how much and how often their peers drink also drink more than students who have more accurate perceptions (Collins & Spelman, 2013; Pilatti et al., 2020; Scaglione et al., 2013).

Correlates of drinking typologies may vary across the socio-ecological model. At the individual level, for example, men are more likely than women to engage in heavy and high-risk drinking (Beseler et al., 2012; Gohari et al., 2020; Read et al., 2013; Rinker et al., 2016; Villarosa-Hurlocker and Madson, 2020; Wilsnack et al., 2018) and experience alcohol related consequences (Read et al., 2013; Rinker et al., 2016). Black and Asian students are less likely than White students to engage in heavy and high-risk drinking (Campbell et al., 2021; Wamboldt et al., 2019) and ethnographic data from this study suggest that LGBTQ + students are less likely than cisgender-heterosexual students to drink in large social settings (Wamboldt et al., 2019). Among young adults, higher socio-economic status has been associated with more frequent drinking (Casswell et al., 2003) and elevated odds of binge drinking (McKetta & Keyes, 2020). Developmental factors, including younger age of alcohol initiation (Haardörfer et al., 2021; Hingson & Zha, 2009) and parental alcohol problems (Haardörfer et al., 2021) are predictors of heavy drinking and alcohol problems later in life.

College student drinking patterns may be shaped by community-level factors such as engagement in student activities and feelings of campus connectedness. For example, participation in Greek life has been associated with heavier drinking over the course of college (Capone et al., 2007; Park et al., 2008) and into midlife (McCabe et al., 2018). Similarly, athletes binge drink more than their peers (Green et al., 2014; Tewksbury et al., 2008), although intramural and club sport athletes drink more than varsity athletes (Barry et al., 2015). In contrast, student members of religious or cultural groups (Strano et al., 2004) and political organizations (Lorant et al., 2013) drink and binge drink less than non-members. Alcohol also has been suggested as a mechanism for developing social connections and increasing feelings of connectedness (Brown & Murphy, 2020; Wamboldt et al., 2019).

This paper advances knowledge about associations between social, demographic, attitudinal, and developmental factors and binge drinking among U.S. college students by including contextual factors when determining typologies of drinkers. Among Canadian secondary school students, problem drinkers were more likely to be in “high alcohol-use” schools where drinking was normalized (Gohari et al., 2020). Similarly, an Argentinian study revealed that heavy drinking college students drank in multiple settings (small groups, bars, parties, alone, etc.), and experienced more alcohol-related consequences than those who drank in fewer, more intimate settings (Pilatti et al., 2020). However, to our knowledge, no American college student studies have considered the socio-ecological context, such as where students drink, how they obtain alcohol, and how much they believe their peers are drinking, when developing typologies.

The current study used a person-centered approach to identify socio-ecologically derived latent classes of college student drinkers and socio-ecological correlates of class membership. In contrast to traditional variable-centered approaches like multiple regression, person-centered approaches like Latent Class Analysis (LCA) allow for the identification of qualitatively distinct subgroups of individuals within a population based on a set of manifest indicator variables (Berlin et al., 2014). Including alcohol consumption indicators alongside where students obtain alcohol, where they drink, and their perceptions of peer alcohol use allowed us to consider aspects of the social environment when forming drinking typologies and examining individual- and community-level correlates of class membership. This knowledge may be crucial for better tailoring harm reduction programs and policies to student drinking patterns.

2. Material and methods

2.1. Participants

This secondary analysis was from a cross-sectional, population-based survey of undergraduates from two private interconnected campuses in a large U.S. metropolitan area (Mellins et al., 2017). Survey data were part of a larger mixed methods study examining risk and protective factors affecting sexual health and sexual violence among college students (Mellins et al., 2017). Participants were selected via stratified random sampling from the March 2016 population of 9,616 enrolled undergraduate students aged 18–29 years. We sent email invitations to participate in the web-based survey to 2,500 students, of whom 1,671 (67 %) consented to participate. LCA focused on 1390 students who had consumed one or more alcoholic drinks in the past 12 months (Table 1).

Table 1.

Demographics of the LCA Sample of Drinkers and Comparisons to Non-Drinkers.

Demographic Variables	Past-year drinkers		Non-drinkers
	N	%	N	%	X 2
Total	1390	100.0	228	100.0
Gender identity					7.84*
Male	574	41.3	73	32.4
Female	790	56.8	150	66.7
Gender expansive	24	1.7	2	0.9
Race/Ethnicity					22.79***
Non-Hispanic White	623	44.8	71	32.6
Non-Hispanic Black	110	7.9	25	11.5
Non-Hispanic Asian	295	21.2	72	33.0
Hispanic	213	15.3	26	11.9
Other	126	9.1	24	11.0
Age					24.78***
18–20	732	52.7	156	69.3
21–23	551	39.6	51	22.7
24+	106	7.6	18	8.0
	Sexual Orientation				10.26**
Heterosexual	1092	78.6	180	82.2
Bisexual, lesbian/gay, queer	221	15.9	22	10.0
Other	56	4.0	17	7.8
	Year of School				25.90***
First year (Freshman)	320	23.0	80	35.7
Second year (Sophomore)	332	23.9	60	26.8
Third year (Junior)	368	26.5	51	22.8
Fourth year (Senior) or beyond	369	26.6	32	14.3
	Current Living Situation				14.57**
Dorm room	694	49.9	135	60.0
Suite	445	32.0	52	23.1
Apartment	217	15.6	29	12.9
Other	24	1.7	9	4.0
Difficulty Paying for Basic Needs in College					3.69
Never	727	52.3	111	49.8
Rarely	317	22.8	45	20.2
Sometimes	245	17.6	48	21.5
Often	66	4.7	14	6.3
All of the time	23	1.7	5	2.2
Pell Grant Recipient					12.23**
Yes	288	21.3	68	32.1
No	1067	78.7	144	67.9

Note:

^*p <.05

^**p <.01

^***p <.001.

Among past-year drinkers, half (52.7 %) were 18–20 years old, while nearly 40 % were 21–23, and 7.6 % were 24 or older. More than half (56.8 %) identified as female and 1.7 % as non-binary or gender expansive. Nearly 45 % were non-Hispanic White, 21.2 % were non-Hispanic Asian, 15.3 % were Hispanic and 7.9 % were non-Hispanic Black. Most (78.6 %) were heterosexual but 16 % reported being lesbian, gay, bisexual, or queer. Most students lived in on-campus dorms (49.9 %) or suites (32.0 %). More than half (52.3 %) never had difficulty paying for basic necessities, while 21.3 % had received a Pell grant, signifying substantial family financial need. An approximately equal number of students represented each year in school. Non-drinkers were significantly more likely to be women, Asian, younger (18–20), freshman, and to live in dorms compared to those who ever drank. Past-year drinkers were significantly more likely to contain White students, seniors, and older students (21+) compared to non-drinkers.

2.2. Procedures

This study was approved by the local institutional review board. We obtained a federal Certificate of Confidentiality and a study-specific institutional waiver of the obligation to report sexual assault to protect privacy. Students used a unique email link to access the survey at a location of their choosing (84 %) or the on-campus research office (16 %). The survey took 35–40 min and participants received a US $40 incentive and entry into a lottery to win an additional US $200 gift card. We offered health and mental health referrals during the consent process and after the survey.

2.3. Measures

2.3.1. Indicators

Twelve indicator variables were entered into the LCA. Three Likert items from the Alcohol Use Disorders Identification Test (AUDIT; Saunders et al., 1993), referred to as the AUDIT-C (Bush et al., 1998) and shown to have good validity and reliability in college samples (Barry et al., 2015), assessed past-year frequency of alcohol use (ranging from never to 4 + times per week), typical quantity on a drinking day (1–2 to 10+), and frequency of binge drinking (6 + drinks on a single occasion).

Each item was a separate consumption indicator. Three dichotomous (yes/no) items about how the student typically gets alcohol (select all that apply: purchase themselves, a friend buys, provided at a party); five dichotomous (yes/no) items about where the student typically drinks (select all that apply: dorm, fraternity/sorority house, Special Interest housing, East campus and off campus, bars); and one continuous item about perceptions of the percentage of students on campus who drank in the past 30 days¹ also served as indicators. We grouped drinking locations based on our qualitative research on this campus (Wamboldt et al., 2019).

2.3.2. Socio-ecological correlates, protective factors, and problems

2.3.2.1. Individual correlates.

Demographics.

We asked participant age, gender (male, female, gender expansive [trans man, trans woman, genderqueer or other]), race/ethnicity (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, Hispanic/Latinx and other [American Indian/Alaskan Native, Native Hawaiian/Pacific Islander, or other]), sexual orientation (heterosexual; gay or lesbian, bisexual, queer; other), year in school (freshman, sophomore, junior, senior, and fifth year or beyond), frequency of difficulties paying for basic necessities (never, rarely or sometimes, often, or always), and Pell grant recipient (yes/no).

Developmental correlates.

Prior to the AUDIT, we asked age at first drink of alcohol with “before 12, 13–14, 15–17, 18–20, and 21 or older” as response options. We asked a single item from an abbreviated version of the Adverse Childhood Experiences Scale (ACES; Felitti et al., 1998) about ever living with someone who had alcohol problems, alcoholism, or used street drugs (yes/no).

2.3.2.2. Community correlates

Campus activities.

Participation in the following activities since starting college was assessed: professional group related to major; fraternity or sorority pledge/member; varsity athletics; intramural/club athletics; political/activist group; community service group; religious, cultural, or identity-based group; student government; resident assistant or orientation leader; media organization; musical, theater or arts group; other student group/organization.

Campus connectedness.

Two items measured campus connectedness: “I feel valued as an individual at this school” and “I feel close to people at this school.” Respondents answered on a 5-point Likert scale ranging from Strongly Disagree to Strongly Agree. Items were summed with higher scores reflecting greater feelings of belonging (α = 0.68).

2.3.2.3. Protective behaviors and problems associated with drinking.

Protective Behaviors.

We asked ten items assessing frequency of past 12-month use of protective behaviors such as alternating alcoholic and non-alcoholic beverages, avoiding drinking games, and tracking number of drinks with response options on a 5-point Likert scale ranging from 0 = Never to 4 = Always (American College Health Association [ACHA], 2009). Items were summed so higher scores indicated greater use of protective behaviors (α = 0.81).

Alcohol Problems.

We asked seven AUDIT (Saunders et al., 1993) items about past-year frequency of problems as a result of drinking on a 5-point Likert scale ranging from 0 = Never to 4 = Daily or Almost Daily. Responses were summed with higher scores reflecting more severe alcohol problems (α = 0.76). The AUDIT has good psychometric properties in college samples (Fleming et al., 1991).

2.4. Analytic plan

We examined descriptive information (proportions, variable distributions) of all drinking and contextual indicators to determine how variables should be coded and entered into LCA. We used LCA to identify typologies of drinkers according to twelve indicator variables that reflected the quantity and frequency of drinking and binge drinking, location of drinking, how students obtain alcohol, and perceptions of peer drinking. We iteratively fit classes, starting with a 1-class model and increasing sequentially until model fit criteria indicated the best-fitting model. The best-fitting model had the smallest Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC; Burnham & Anderson, 2004) and a significant bootstrap likelihood ratio test (BLRT), which indicate how many groups should be extracted by testing the parsimony of the current model against the model with one less group (Nylund et al., 2007). We also examined entropy, or how well the model identifies separate latent classes, for the best fitting model; values above 0.80 provide good separation (Berlin et al., 2014). We fit models using maximum likelihood with robust standard errors. Posterior probabilities of latent class membership were derived using Bayes’ Rule, and drinkers were classified for subsequent analysis according to most likely class.

To examine associations between the derived classes and correlates, we used chi square for categorical correlates (demographics, membership in campus groups/organizations) and Analysis of Variance for continuous correlates (alcohol consequences, protective behaviors, campus connectedness). Significant omnibus χ² or F-tests were followed up with pairwise comparisons using Bonferroni correction to adjust for multiple tests.

3. Results

3.1. Latent class analysis results

The four-class solution was the best fit for the data as indicated by the lowest AIC, BIC, a significant BLRT, and an entropy value of 0.80 (see Table 2). Class 1, ‘Tasters,’ included 21 % of the sample (n = 290) who reported low frequency (i.e., less than monthly) and low quantity (1–2 drinks/occasion) drinking with alcohol provided by others (friends/parties) and occurring in dorms; students in this class believed about 63 % of peers drank in the past 30 days. Class 2, ‘Bargoers,’ included 19.5 % of the sample (n = 271) who reported moderate frequency (monthly-weekly) and quantity (1–2 drinks/occasion) drinking. Bargoers more likely than any other class to self-purchase (99.6 %) and drank in bars (90.8 %). Class 3, ‘Partiers’ included 34.7 % (n = 483) of students who reported moderate frequency (monthly-weekly) and moderate quantity (3–4 drinks/occasion) drinking, infrequent binge drinking, alcohol provided by others (>80 % from friends/parties), and drinking in multiple venues (98 % in dorms, 82 % in apartments, 75 % in bars, and 64 % in Greek houses). Class 4, ‘Bingers,’ comprised 24.8 % (n = 345) of students who reported high frequency (e.g., 1 in 5 drank 4 + days/week) and high quantity (65 % consumed 5 + drinks/occasion) drinking, and frequent binge drinking (44.5 % reported weekly binge drinking); they self-purchased (89 %) and drank in multiple venues including bars (90 %), dorms (88 %), apartments (85 %), and Greek houses (70 %). The latter three classes believed that about 70 % of their peers drank in the past 30 days (Table 3). The class probability statistics for the four classes were 0.90, 0.89, 0.89, and 0.88, respectively.

Table 2.

Fit Statistics for Models with Different Numbers of Classes.

	AIC	BIC	BLRT	Entropy
1-class	27150.976	27240.006	–	–
2-class	25593.349	25760.934	1573.136***	0.74
3-class	24878.020	25124.161	738.526**	0.78
4-class	24368.229	24692.927	534.863***	0.80
5-class	24185.534	24588.788	210.754	0.78

Bolded 4-class solution was the best fit for the data.

Table 3.

Differences in Indicator Variable Endorsement by Class.

Indicator Variables	Response options	LCA sample (n = 1390)	Tasters (n = 291)	Bargoers (n = 271)	Partiers (n = 483)	Bingers (n = 345)
Categorical indicators %							X 2
1.How often do you have a drink containing alcohol?	Monthly or less	21.0	71.1a	10.3b	11.6c	0.3d	352.68***
	2–4 times a month	39.8	26.8a	46.1b	57.6c	20.9d
	2–3 times a week	31.9	1.7a	35.8b	29.4c	58.0d
	4 + times a week	7.3	0.3a	7.7b	1.4c	20.9d
2.How many drinks containing alcohol do you have of a typical day when you are drinking?	1 or 2	37.3	71.6a	50.7b	34.6c	2.3d	322.90***
	3 or 4	39.5	24.2a	40.7b	53.7c	32.1d
	5 or 6	16.0	3.5a	8.1b	11.0c	39.9d
	7, 8 or 9	5.5	0.3a	0.4b	0.6c	20.7d
	10 or more	1.3	0.3a	0b	0c	5.0d
3.How often do you have 6 or more drinks in one occasion?	Never	32.2	80.1a	28.5b	28.4b	0c	996.24***
	Less than monthly	36.9	18.9a	57.8b	62.0b	0.9c
	Monthly	19.4	1a	13.7b	9.5b	53.2c
	Weekly	11.0	0a	0b	0b	44.5c
	Daily or almost daily	0.4	0a	0b	0b	1.5c
4–6.Currently, where does your alcohol typically come from? ^	I purchase myself	66.3	22.0a	99.6b	59.0c	87.5d	474.4***
	A friend purchases	51.2	43.6a	5.5b	81.6c	51.0a	415.19***
	Provided at a party	55.8	43.6a	2.6b	91.1c	58.6d	580.53***
7–11.Since entering college, I party/socialize using alcohol and drugs…^	Fraternity/sorority house	48.1	11.3a	31.0b	64.4c	69.9c	308.09***
	Special interest housing	9.8	4.1a	5.2a, b	14.9c	11.0c, b	31.61***
	Bars	71.0	22.3a	90.8b	75.6c	90.1b	468.39***
	Dorms	77.1	47.1a	63.5b	94.6c	88.4d	297.51***
	East or off-campus	72.6	35.4a	66.4b	89.2c	85.5c	312.57***
Continuous indicators M (SD)							F
12.Past 30-day peer alcohol use	%	68.40 (19.15)	63.25 (22.90)a	69.82 (19.63)b	70.56 (16.15)b	68.58 (18.46)b	9.69***

Note:

^*p <.05

^**p <.01

^***p <.001

^{^}Response options were not mutually exclusive; each response has been treated as a separate dichotomous variable. Identical superscripts denote a subset of 4-class solution categories whose column proportions do not differ significantly from each other at p <.05 using Bonferroni correction. Bolded proportions reflect class characteristics that were used to define the classes.

3.2. Individual characteristics associated with class membership

Table 4 shows demographic differences between classes. Bargoers were older than students in the other classes; consistent with this observation, first-year students were overrepresented among Tasters, Partiers, and Bingers, while fourth-years were overrepresented among Bargoers. Men were overrepresented among Bingers. White non-Hispanic students were overrepresented among Bargoers and Bingers while Asian non-Hispanic students were overrepresented among Tasters, and Hispanic students were overrepresented among Partiers. Although class membership did not vary by difficulty paying for basic necessities, Pell grant recipients were overrepresented among Tasters and Bargoers. Table 5 shows that age at first drink differed across classes such that those who had their first drink at 13–14 were most likely to be Bargoers or Bingers. Close to half of Bargoers, Partiers and Bingers had their first drink at 15–17 compared to less than 30 % of Tasters. More than half of Tasters reported first drinking between the ages of 18–20. Classes did not differ in living with someone who misused alcohol or drugs while growing up.

Table 4.

Demographic Differences By Class Membership

Demographic	Tasters (n = 291)	Bargoers (n = 271)	Partiers (n = 483)	Bingers (n = 345)	X 2
Gender identity					88.24***
Male	34.0 % a	39.9 % a	32.2 % a	61.6 % b
Female	62.5 % a	59.0 % a	66.6 % a	36.9 % b
Gender expansive	3.4 % a	1.1 % a	1.2 % a	1.5 % a
Race/Ethnicity					54.30***
Non-Hispanic White	34.3 % a	53.0 % b	41.9 % a	54.3 % b
Non-Hispanic Black	31.8 % a	19.6 % b	21.1 % b	15.2 % b
Non-Hispanic Asian	11.2 % a	7.4 % a	7.9 % a	6.2 % a
Hispanic	14.3 % a,b	10.4 % b	18.5 % a	16.7 % a,b
Other	8.4 % a	9.6 % a	10.6 % a	7.6 % a
Age					188.26***
18–20	61.9 % a,b	19.2 % c	66.0 % b	52.8 % a
21–23	28.5 % a	63.5 % b	31.3 % a	42.0 % c
24+	9.6 % a	17.3 % b	2.7 % c	5.2 % a,c
Sexual Orientation					12.58*
Heterosexual	78.4 % a	79.9 % a	79.0 % a	81.9 % a
Bisexual, lesbian/gay, queer	14.5 % a	17.5 % a	18.1 % a	13.7 % a
Other	7.1 % a	2.6 % a,b	2.9 % b	4.4 % a,b
Year of School					104.04***
First year (Freshman)	27.1 % a	7.0 % b	28.8 % a	24.1 % a
Second year (Sophomore)	25.8 % a	17.7 % a	26.1 % a	24.1 % a
Third year (Junior)	29.9 % a	29.2 % a	24.5 % a	24.3 % a
Fourth year (Senior)+	17.2 % a	46.1 % b	20.5 % a,c	27.5 % c
Current Living Situation					93.86***
Dorm room	55.9 % a	31.3 % b	54.4 % a	54.7 % a
Suite	25.5 % a	34.3 % a	33.8 % a	34.2 % a
Apartment	15.9 % a	31.7 % b	11.3 % a	9.4 % a
Other	2.8 % a	2.6 % a	0.6 % a	1.8 % a
Difficulty Paying for Basic Needs in College					13.98
Never	54.9 %a	49.4 % a	55.4 % a	49.9 % a
Rarely	22.6 %a	23.2 % a	20.7 % a	26.4 % a
Sometimes	16.3 %a	19.1 % a	18.0 % a	17.6 % a
Often	4.2 %a	5.2 % a	4.1 % a	5.9 % a
All of the time	2.1 %a,b	3.0 %b	1.7 %a,b	0.3 %a
Pell Grant Recipient					17.06**
Yes	27.5 %a	26.0 %a,b	18.1 %b,c	16.8 %c
No	72.5 %a	74 %a,b	81.9 %b,c	83.2 %c

Note:

^*p <.05

^**p <.01

^***p <.001

^{^}Response options were not mutually exclusive; each response has been treated as a separate dichotomous variable. Identical superscripts denote a subset of 4-class solution categories whose column proportions do not differ significantly from each other at p <.05 using Bonferroni correction.

Table 5.

Differences in Individual- and Community-Level Correlates by Class Membership.

Correlate	LCA sample (n = 1390)	Tasters (n = 291)	Bargoers (n = 271)	Partiers (n = 483)	Bingers (n = 345)
Categorical correlates %						X2
Individual: Developmental
Age at first drink						131.21***
≤12 years	2.4	3.4a	1.1a	3.1a	1.7a
13–14 years	14.1	7.2a	16.2b	10.8a	22.9c
15–17 years	45.3	28.9a	46.1b	51.3b	50.1b
18–20 years	34.4	51.9a	31.0b, c	32.9c	24.3b
>20 years	3.7	8.6a	5.5a	1.9b	0.9b
Lived with someone who misused alcohol or drugs	13.0	10.6a	17.0a	12.0a	13.1a	5.65
Community: Campus Activities
Greek life	23.3	7.6a	17.4b	22.9b	41.5c	106.40***
Athletics (varsity)	11.1	6.5a	10.2a	8.6a	19.3b	32.0***
Athletics (intramural/club)	19.9	15.1a	17.5a	18.5a	27.6b	18.1***
Religious/cultural/identity	27.4	30.9a,b	19.8c	32.8b	22.8a, c	19.86***
Professional/political group	62.4	59.9a,b	48.7b	72.5c	61.1a	42.76***
RA/leadership	11.6	6.8b	11.4a,b	13.9a	12.5a,b	8.8*
Media	22.1	19.1a	20.2a	26.1a	20.5a	6.9
Music or performance arts	25.3	28.4a	22.8a,b	31.1a	16.6b	24.16***
Continuous correlates M (SD)						F
Community: Campus Connectedness	5.02 (1.95)	5.13a,b (1.98)	5.18a (1.96)	5.04a,b (1.96)	4.75b (1.86)	3.16*
Alcohol Protective Behaviors	28.78 (7.79)	33.19a (9.56)	29.77b (7.35)	29.37b (6.84)	24.35c (5.77)	71.6***
Alcohol Problems	2.16 (3.18)	0.57a (1.30)	1.69b (2.74)	1.87b (2.60)	4.27c (4.15)	93.52***

Note:

^*p <.05

^**p <.01

^***p <.001; identical superscripts within a row do not differ significantly at p <.05 with Bonferroni correction. feedback and edited subsequent drafts.

3.3. Community factors associated with class membership

Table 5 shows that compared to other classes, Bingers were more likely to participate in Greek life while Tasters were less likely to participate in Greek life. Bingers were more likely than other classes to participate in varsity athletics and intramural or club sports. Tasters and Partiers were most likely to belong to religious or cultural groups and Partiers were most likely to report membership in a professional, political or community service organization. Resident Assistants were more likely to be Partiers than Tasters. Tasters and Partiers were more likely than Bingers to be members of a musical theater or performing arts group. Campus connectedness varied across groups, F (4,1574) = 3.5, p =.007, but follow-up comparisons indicated that the only significant pairwise differences were for Bargoers to have significantly higher campus connectedness compared to Bingers.

3.4. Protective factors and problems by class membership

Classes varied in alcohol protective behaviors, F (3,1284) = 71.60, p <.001, and past-year AUDIT problems, F (3,1384) = 93.52, p <.001. Compared to other classes, Tasters reported the greatest use of protective behaviors and the fewest problems while Bingers reported the lowest use of protective behaviors and the most problems.

4. Discussion

The current study used a person-centered approach to examine types of college student drinkers based on alcohol consumption and sociocontextual factors including how students obtain alcohol, where they drink, and perceptions of peer drinking on campus. Similar to a longitudinal study of college drinking (e.g., Cleveland et al., 2012), four distinct classes of college student drinkers emerged: a low frequency and quantity, other buyer (friends/parties), primarily dorms class (Tasters); a moderate frequency-low quantity, self-buyer, bars class (Bargoers); a moderate frequency and quantity, other buyer (friends/parties), multi-venue class (Partiers); and a heavy frequency and quantity, binge drinking, self-buyer, multi-venue class (Bingers). Findings are somewhat consistent with those of Cleveland et al. (2012), who modeled transitions in drinking behavior from the summer before and during the fall of students’ first year in college and found four classes including non-drinkers, weekend non-bingers, weekend bingers, and heavy drinkers (Cleveland et al., 2012). In the current study, classes differed in demographics as well as age of first drink, campus activities and connectedness, protective behaviors, and alcohol problems.

Consistent with prior research (e.g., Beseler et al, 2012; Gohari et al., 2020), we observed a low use (Tasters) and a heavy use (Bingers) class but we expanded on prior studies by contextualizing how and where these classes drank. Tasters drank small amounts infrequently, typically relied on others for alcohol, and drank in dorms. Compared to other groups, Tasters first drank alcohol at later ages, were more likely to participate in religious/cultural and musical/ theater groups, used more protective behaviors, and reported the fewest alcohol consequences. Bingers, in contrast, reported the heaviest drinking and binge drinking, purchased themselves and drank in a variety of venues. Bingers also evidenced risks including an earlier age of drinking initiation (Haardörfer et al., 2021), greatest likelihood of Greek life and athletics involvement (Green et al., 2014; McCabe et al., 2018), lowest feelings of campus connectedness and lowest use of protective behaviors (Barry et al., 2016; Jouriles et al., 2020), and the greatest alcohol consequences (e.g., Hingson et al., 2017) compared to other classes. Consistent with work on racial and gender differences in heavy drinking (e.g., Campbell et al., 2021; Wilsnack et al., 2018), White men were most likely to be Bingers while Asian students were most likely to be Tasters.

Instead of a single moderate class (e.g., Beseler et al, 2012; Gohari et al., 2020), we observed two moderate classes distinguished by ecological factors including how students obtained alcohol and where they drank or partied. Two moderate classes also were found by Chiauzzi and colleagues (2013) although those classes were developed based on alcohol and co-occurring drug use. Bargoers had fewer drinks per occasion, purchased themselves, drank in bars, and tended to be older and non-Hispanic White students; they may have limited their drinking to social settings like happy hours or dates. Partiers, in contrast, consumed more when they drank, relied on others or parties for their alcohol, drank in multiple venues including dorms, apartment, bars, and Greek houses, and were likely to be younger and Hispanic; however, they reported binge drinking infrequently. Partiers may have similar social drinking motives to Bargoers but fewer independent opportunities to acquire or control alcohol, which could reflect the impact of underage drinking policies. Specifically, younger students who want to drink socially but cannot acquire alcohol on their own may drink more and in more venues because they do not know when their next opportunity to acquire alcohol will arise. Despite differences in how they drink, demographics, and campus activities, Bargoers and Partiers did not differ on campus connectedness, use of protective behaviors, or alcohol consequences, suggesting equifinality across these groups in the correlates measured.

At the individual level, both low quantity drinking classes (Tasters and Bargoers) were more likely to receive Pell grants compared to the moderate and heavy quantity classes, which is consistent with studies showing that higher income people are more likely to drink heavily (McKetta & Keyes, 2020) and may reflect the high costs of both alcohol and the possible negative consequences of drinking. For example, students on scholarship cannot risk conduct violations for underage drinking and may be wary of incurring ambulance or other healthcare costs when partying. Developmentally, classes did not differ in living with someone who misused alcohol or drugs while growing up. Prior studies showing that parental alcohol use problems increase risk for heavy use (e.g., Haardörfer et al., 2021) include both genetic and environmental liability whereas living with someone who misused alcohol or drugs may not necessarily include genetic liability. Bargoers and Bingers both had their first drink at an earlier age than Partiers or Tasters but only Bingers went on to drink heavily; this finding indicates that early exposure to alcohol alone does not predict sustained heavy drinking.

At the community level, on average students believed that around two-thirds of their peers had used alcohol in the past month. Although descriptive norms about drinking only differed between Tasters and the other three groups, Tasters perceived that a lower percentage of their peers used than students in other classes, which fits with literature linking drinking norms to actual behavior (DiGuiseppi et al., 2018). A novel finding here was that Bingers had the lowest feelings of campus connectedness. Although alcohol has been identified as a mechanism for developing social connections (Brown & Murphy, 2020; Wamboldt et al., 2019), high alcohol use at these institutions may harm broader feelings of connection or belonging.

4.1. Clinical implications

Bingers may benefit from personalized normative feedback and programming to increase alcohol protective behaviors (e.g., Fachini et al., 2012). Despite low campus connectedness, Bingers were more likely than other groups to be involved with fraternities and athletics. Developing group-level harm reduction interventions that resonate with and can be facilitated by students who hold these affiliations may be one strategy for changing norms about alcohol protective behaviors to reduce concerning consequences associated with frequent binge drinking (Braitman & Lau-Barraco, 2020). Incentivizing and providing opportunities to form relationships with diverse groups of students also may mitigate these concerns (Jorgenson et al., 2018).

4.2. Limitations & future directions

Limitations of our study include that our sample was drawn from two highly-selective private interconnected colleges in an urban area; findings may not generalize to other college settings. Additionally, because data were cross-sectional, we were unable to examine whether and how drinking patterns changed during college, nor establish causal relationships among variables. Future research could examine drinking patterns over time to further understand the circumstances in which students drink (e.g., Cleveland et al., 2012). Prior studies of college alcohol use typologies often have considered the role of motives for use, which we did not assess. Future work on typologies would be strengthened by the inclusion of motives. Future work also could examine a broader array of correlates such as genetic and developmental factors and peer networks and interactions as well as relationships between classes and other measures of functioning such as GPA, risk behaviors, other alcohol-related consequences, mental health, and wellbeing. Data were collected in 2016 and it is possible that drinking typologies and correlates have shifted over time. Finally, the AUDIT binge drinking item about frequency of consuming 6 or more drinks on an occasion may underestimate binge drinking, particularly among women, according to NIAAA and SAMHSA definitions (Olthuis et al., 2011).

4.3. Conclusion

These limitations notwithstanding, the current study highlights heterogeneous patterns of drinking, acquiring alcohol, and partying/socializing among U.S. college students at these institutions. Groups had distinct personal characteristics, differing levels of campus involvement and connectedness, and engaged in protective behaviors and experienced problems to different degrees. Despite differences in how and where students drank, only students who binge drink frequently were experiencing significant harms. These students could benefit from harm reduction interventions developed with and delivered by members of these groups. The campus also could offer more opportunities and structured experiences to encourage students who binge drink frequently to diversify their social networks and increase feelings of connectedness to the campus.

5. Author agreement

We certify that all authors have seen and approved the final version of the manuscript being submitted. We also warrant that this article is the authors’ original work, hasn’t received prior publication and isn’t under consideration for publication elsewhere.

Contributors

JSH and CAM acquired funding, managed and supervised study design, data collection, and all regulatory aspects of the study. MW was responsible for data curation. KW, ZB, AHPM, JSH, and CAM conceptualized the analysis. KW ran all analyses and worked with ZB and AHPM to draft the initial manuscript. All other authors provided feedback and edited subsequent drafts.

Data availability

The data that has been used is confidential.