Frequency of drinking games participation and alcohol-related problems in a multiethnic sample of college students: Do gender and ethnicity matter?

Abstract

Introduction:

A drinking game (DG) is a high-risk, social drinking activity that consists of certain rules (i.e., when to drink and how much to consume) designed to promote inebriation and that requires each player to perform a cognitive and/or motor task (Zamboanga et al., 2013). Research suggests that non-White or female students who play DGs are at an increased risk of experiencing alcohol-related problems. Thus, this study examined whether the associations between DG participation and alcohol-related problems were similar for men and women and across ethnic groups.

Method:

College students (N = 7409; 73% women; 64% White, 8% Black, 14% Hispanic, 14% Asian) from 30 U.S. colleges/universities completed self-report questionnaires. Results: Controlling for age, site, Greek membership (i.e., membership in a fraternity or sorority), and typical alcohol consumption, results indicated that the association between DG participation and alcohol-related problems was stronger for men compared to women. With respect to ethnicity, the association between these variables was stronger among Black women than Black men.

Conclusions:

Findings from this large-scale study highlight the need to closely investigate how gender and ethnicity moderate the associations between DG participation and alcohol-related problems. College intervention efforts designed to address high-risk drinking behaviors such as DG participation might consider paying close attention to ethnic minority populations, perhaps particularly Black women.

1. Introduction

A drinking game (DG)¹ is a high-risk, social drinking activity that has certain rules (i.e., when to drink and how much to consume) designed to promote intoxication and requires participants to perform a cognitive and/or motor task (Zamboanga et al., 2013). DGs are unlike other high-risk drinking activities (e.g., prepartying, or drinking before going out) because by following the rules, certain players may be targeted to drink more than others, leading to their more rapid intoxication.

DGs are prevalent among college students; for instance, Grossbard, Geisner, Neighbors, Kilmer, and Larimer (2007) found that nearly half of the students across both of their study samples had participated in DGs at least once in the past year (see also Zamboanga et al., 2014, for a review). However, findings regarding gender or ethnic prevalence rates have been inconsistent. Some studies suggest that men and women participate in DGs at equal rates (e.g., Grossbard et al., 2007; Pedersen & LaBrie, 2006), whereas other work has found higher rates of DG involvement among men (e.g., Cameron et al., 2010; Polizzotto, Saw, Tjhung, Chua, & Stockwell, 2007). With respect to ethnicity, one study (Pedersen & LaBrie, 2006) found higher rates of DG participation among Whites than non-Whites. However, another study (Haas, Smith, Kagan, &Jacob, 2012) found a very modest (albeit significant), positive bivariate correlation between non-White ethnicity and rates of DG participation on prior drinking occasions.

College students are aware of the health risks associated with heavy drinking, but this knowledge does not appear to deter them from playing DGs (Polizzotto et al., 2007). Although research has found positive associations between DG participation and alcohol-related problems (e.g., Grossbard et al., 2007; Hone, Carter, & McCullough, 2013; Polizzotto et al., 2007), some questions regarding this association require further investigation. Because women metabolize alcohol more slowly than men, women who participate in DGs and consume similar amounts of alcohol as men are likely to achieve higher levels of inebriation, which can increase their risk for alcohol-related problems (Cameron, Leon, & Correia, 2011; Cameron et al., 2010; Correia & Cameron, 2010). Indeed, Pedersen and LaBrie (2006) found that although men and women participate in DGs at comparable rates, the association between frequency of DG participation and alcohol-related problems was stronger among women compared to men.

Pedersen and LaBrie (2006) also found the association between DG participation and alcohol-related problems to be higher among non-White students than White students. Because non-White students in their sample had lower rates of DG participation than White students, they noted that lack of familiarity with DGs among non-White students may help explain their increased risk for alcohol-related problems. Although these findings are informative, collapsing students from different ethnic minority backgrounds into a single “non-White” group makes it difficult to ascertain how the association between DG participation and alcohol-related problems might differ across different ethnic groups. Collapsing across minority groups can also mask which ethnic groups may be at greater risk and in need of targeted intervention. Research also suggests that there are differences in alcohol metabolism enzyme activity across racial groups, which could affect the pattern of negative consequences experienced across ethnic groups (U.S. Department of Health & Human Services, 2007).

Using a multisite, multiethnic college sample, the present study builds on prior research by examining the association between frequency of DG participation and alcohol-related problems and by testing whether such relationships are similar across gender and ethnic groups. Based on prior research (Cameron et al., 2010), we controlled for typical alcohol consumption to isolate the unique association between DG participation and alcohol-related problems. We also controlled for age because younger students tend to report higher DG participation rates than older students (Nagoshi, Wood, Cote, & Abbit, 1994; Polizzotto et al., 2007). Finally, we controlled for Greek membership given that (a) students who are members of fraternities or sororities are generally at high risk for heavy alcohol consumption and related problems (Borsari, Hustad, & Capone, 2009; Mallett et al., 2013), and (b) Haas et al. (2012) found a positive (albeit modest) bivariate correlation between Greek affiliation and rates of DG participation on prior drinking occasions.

Based on prior research, we hypothesized that the associations between DG participation and alcohol-related problems would be stronger for women than men. However, given the mixed findings in the literature, we did not advance any a priori hypotheses regarding ethnic differences in these associations.

2. Method

2.1. Participants and procedures

Participants were derived from the Multi-Site University Study of Identity and Culture (Weisskirch et al., 2013). The data analytic sample consisted of 7409 college attending emerging adults (18–25 years; 13% reported membership in a fraternity or sorority; see Table 1 for descriptives) from 30 U.S. colleges and universities who answered most if not all questions pertaining to DG and alcohol behaviors. Researchers recruited participants through flyers and e-mail announcements. In exchange for course credit or entries to win a prize, respondents completed an online survey that took 1–2 h to complete. All procedures were approved by the Institutional Review Board at each site.

Table 1.

Descriptive statistics.

Variable	Total	Men	Women	Whites	Blacks	Hispanics	Asians	F-test
	Sample	(27%)	(73%)	(64%)	(8%)	(14%)	(14%)	(Partial eta2)
	(N = 7409)	(n = 2005)	(n = 5404)	(n = 4741)	(n = 595)	(n = 1065)	(n = 1008)
Age	19.78 (1.61)	19.82a (1.69)	19.76a (1.58)	19.79xz (1.59)	19.92yz (1.66)	19.72y (1.71)	19.69y (1.57)	2.15 (.000)1 2.88** (.001 )2
Frequency of DG participation	1.57 (1.68)	1.82 (1.78)	1.48 (1.64)	1.87 (1.72)	.78 (1.25)	1.16x (1.55)	1.07x (1.47)	56.86***(.01)1 150.04***(.06)2
Typical alcohol consumption3	2.68 (2.69)	3.36 (3.04)	2.42 (2.51)	3.19 (2.77)	1.29 (1.88)	2.04 (2.36)	1.71 (2.33)	181.41*** (.02)1 186.87*** (.07)2
Frequency	.82 (.89)	.91 (.96)	.78 (.86)	.99 (.91)	.40x (.65)	.59 (.79)	.49x (.76)	33.24*** (.01)1 183.05*** (.07)2
Binge drinking	1.05 (1.12)	1.32 (1.19)	.95 (1.07)	1.25 (1.14)	.52 (.85)	.80x (1.00)	.72x (1.02)	162.97*** (.02)1 147.82*** (.06)2
Quantity	.84 (1.06)	1.16 (1.31)	.72 (.94)	1.00 (1.12)	.39x (.77)	.69 (.95)	.52x (.90)	256.66*** (.03)1 108.81*** (.04)2
Alcohol-related problems	2.96 (4.13)	3.61 (4.79)	2.72 (3.82)	3.27 (4.14)	2.18x (3.99)	2.59x (4.02)	2.35x (4.11)	67.50***(.01)1 26.39***(.01)2
Audit total scores4	5.61 (6.05)	6.93 (6.80)	5.12 (5.68)	6.44 (6.12)	3.44x (5.26)	4.60y (5.62)	4.04xy (5.83)	134.53***(.02)1 90.60***(.04)2

Notes. Drinking games (DG). Alcohol Use Disorders Identification Test (AUDIT). Values on the first rows are means per group. Values on the second rows enclosed in parentheses are standard deviations per group. Bonferroni pairwise analyses were conducted. Each subscript letter denotes a subset of gender (^a) and ethnicity (^x,y, and^z) whose means do not differ significantly from each other at the .05 level.

¹Gender comparisons.

²Ethnic group comparisons.

³A score of 4 or 5 on the AUDIT consumption subscale (i.e., AUDIT-C; the present study’s index of typical alcohol consumption) has been found to be useful in detecting “alcohol-related problems” among college students (Dawson, Grant, Stinson, & Zhou, 2005).

⁴An AUDIT total score (AUDIT typical alcohol consumption + AUDIT alcohol-related problems) of 6 or greater has been shown to be effective in detecting high-risk drinking among college students (Kokotailo et al, 2004).

^*p < .05.

^**p < .01

^***p < .001.

2.2. Measures

2.2.1. Frequency of DG participation²

Participants indicated how often they played DGs using an 8-point scale: 0 = I Don’t Play Drinking Games (38.7%), 1 = Less than Once a Month (19.3%), 2 = Once a Month (10.3%), 3 = Two to Three Times a Month (14.2%), 4 = Once a Week (8.7%), 5 = Two to Three Times a Week (6.4%), 6 = Four to Five Times a Week (0.5%), and 7 = Daily or Nearly Daily (0.3%) (1.6% did not respond to this question). This was similar to a response scale used in another college DG study (Ham, Zamboanga, Olthuis, Casner, & Bui, 2010).

2.2.2. Alcohol-related problems

Participants completed the Alcohol Use Disorders Identification Test (AUDIT; Saunders, Aasland, Babor, de la Fuente, & Grant, 1993). The AUDIT is a well-validated, standardized 10-item instrument that measures hazardous alcohol use. Based on our findings³ and previous work on the AUDIT (e.g., Peng, Wilsnack, Kristjanson, Benson, & Wilsnack, 2012; Shields, Guttmannova, & Caruso, 2004), we used the two-factor model in this study (Factor 1: typical alcohol consumption, items 1–3; α = .85; Factor 2: alcohol-related problems, items 4–10; α = .81).

3. Results⁴

3.1. Structural equation modeling

To test our first research question, we regressed the latent variable for alcohol-related problems on a manifest variable measuring the frequency of DG participation, while controlling for typical alcohol consumption (latent variable), age, and Greek membership (Fig. 1). To account for the nesting of participants within the different colleges and universities, we used the sandwich estimator (Kauermann & Carroll, 2001). The model fit the data adequately, χ²(58) = 1678.46, p < .001; CFI = .99; RMSEA = .06 (90% CI = .05–.07). Overall, the frequency of DG participation was not significantly associated with alcohol-related problems,⁵ β = .04, p = .33.

Fig. 1.

Structural equation model testing the association between frequency of DG participation and alcohol-related problems (controlling for typical alcohol consumption, age, and Greek membership). Standardized coefficients are presented. Correlations among the predictors were omitted from the model for ease of presentation. *p < .05.

3.2. Invariance tests

To test our second research question, we conducted two multigroup model comparisons (controlling for age, site, typical alcohol consumption, and Greek membership) and examined differences in the robust chi-squares (MLR) for the constrained and unconstrained models (Muthén & Muthén, 2006). The first test indicated differences in beta weights between genders, Δχ²(14) = 661.66, p < .001. The path was significant for men (β = .18, p = .02) but not women (β = .01, p = .48). The second test indicated differences in beta weights across ethnic groups, Δχ²(22) = 218.43, p < .001. Inspection of the path coefficients indicated that the association between the frequency of DG participation and alcohol-related problems was significant for Black students (β = .20, p = .03), but not significant for the other groups: Hispanics (β = .12, p = .11), Whites (β = .02, p = .75), and Asians (β = .01, p = .95).

3.3. Post hoc gender analyses within each ethnic group

Because we found significant gender and ethnic variances in the association between the frequency of DG participation and alcohol-related problems, we sought to determine whether there were also gender differences within each ethnic group. As such, we conducted another multigroup model (controlling for age, site, typical alcohol consumption, and Greek membership) and added gender and an interaction term, derived from multiplying gender by the centered frequency of DG participation into the model. There was a significant difference in the robust chi-squares (MLR) for the constrained and unconstrained models, Δχ²(18) = 65.31, p < .001. When examining the beta weights of the interaction terms across ethnic groups, the results indicated a significant gender interaction effect for the Black students (β = − .08, p = .03), but not for the other groups: Asians (β = .04, p = .40), Whites (β = .02, p = .53), and Hispanics (β = − .007, p = .75). A simple slope analysis indicated that the association between the frequency of DG participation and alcohol-related problems was significant for Black females (β = .25, p = .02) but not Black males (β = .01, p = .95).

4. Discussion

The findings from this large-scale study suggest that gender and ethnic considerations may be of particular importance with regard to DG participation and its association with alcohol-related problems. In contrast to Pedersen and LaBrie’s (2006) study, we found that when controlling for typical alcohol consumption, age, Greek membership, and site, the association between the frequency of DG participation and alcohol-related problems was stronger for men compared to women, and for women, this association was nonsignificant. Perhaps these discrepancies resulted from their use of an event-level⁶ approach to examine DG behaviors and alcohol-related problems, and/or differences in instruments between studies [i.e., AUDIT versus Rutgers Alcohol Problem Index (White & Labouvie, 1989) and College Alcohol Problems Scale (Maddock, Laforge, Rossi, & O’Hare, 2001)]. Because we were unable to examine the amount of alcohol consumed while playing DGs,² we also do not know whether its association with alcohol-related problems is stronger for women than men.

Moreover, research suggests that compared to women, men tend to be more accepting of negative alcohol consequences for themselves and others (DeMartini, Carey, Lao, & Luciano, 2011). It is therefore possible that men experienced more alcohol-related problems because they may have been socialized into thinking that heavy drinking and activities like DGs are “manly,” may enjoy the competitive nature of the game, and may be more tolerant of alcohol-related problems. Indeed, a study found that certain masculine norms (e.g., enjoying risky activities, striving to win at all costs, or being a “playboy”) were positively associated with drinking to intoxication (Iwamoto, Cheng, Lee, Takamatsu, & Gordon, 2011). Thus, it is possible that masculine norms may also lead to riskier styles of gaming, resulting in more alcohol-related problems. Future research is needed to explore this possibility.

Finally, compared to men, women may experience different types of alcohol-related problems (e.g., interpersonal/socioemotional negative consequences) that may be overlooked using typical measures (Ham & Hope, 2003). Nonetheless, our findings underscore the need to critically examine the moderating role of gender regarding these associations. Future research could seek to elucidate why gender differences emerge. Gaining a better understanding of these reasons may help improve intervention and prevention efforts.

We also found that when controlling for typical alcohol consumption, age, Greek membership, and site, the association between the frequency of DG participation and alcohol-related problems was stronger among Black women compared to Black men, and for Black men, this association was non-significant. These results are somewhat consistent with Pedersen and LaBrie’s (2006) findings in that non-White or female students who play DGs are at risk for alcohol-related problems. Pedersen and LaBrie (2006) noted that lack of familiarity with DGs among ethnic minority student gamers may increase their risk for alcohol-related problems. However, we found modest differences in the frequency of DG participation across ethnic groups (Table 1).

Perhaps there are additional factors that neither we nor Pedersen and LaBrie (2006) accounted for which may explain why Black women who frequently participate in DGs are at increased risk for alcohol-related problems compared to other student populations. That being said, although we did not examine variables that assess “strain,” and although such analyses would fall outside the scope of the specific aims of our study, it is possible that strain factors could be at play. General Strain Theory (GST) posits that those who experience more stress (i.e., strain) are at increased risk for negative affect (Agnew, 2001), and the emotional response to this stress has been found to be associated with heavy alcohol use among Black college women (Walton, Dawson-Edwards, & Higgins, 2014). Perhaps compared to Black men, Black women may be at an increased risk for strain, and as a result, they may also be at an elevated risk for more harmful outcomes when they play DGs.

Future research on DGs and other high-risk drinking behaviors could test GST by examining the risks that stress poses for negative affect, DG participation, and related consequences, and could investigate whether certain ethnic groups (and men and women within these groups) are at an increased risk for stress. Moreover, in addition to assessing information about students’ alcohol use and high-risk behaviors such as DGs, college health personnel could also gather information about the intersectional impact of race/ethnic and gender stressors, college stress in general, and how students cope. It is important to identify vulnerable students and provide them with the necessary tools to help them cope with their stressors in an effective and healthy manner.

Some limitations of the present study are listed here. We collected self-report data; therefore, students may have under- or over-reported their frequency of DG participation and other alcohol-related behaviors. The cross-sectional design precludes any inferences of causality or directionality between DG participation and alcohol-related problems. Despite our efforts to control for typical alcohol consumption, an event-level analysis⁶ of our research questions would have strengthened our study. We also were not able to examine specific ethnic subgroups; thus, our findings may not generalize to these students (e.g., students from Mexican, Filipino, Chinese backgrounds). Finally, the AUDIT does not capture all the health, social, psychological, and academic challenges that students may experience as a direct result of DG participation. Future research could utilize instruments that capture a broad array of problems for which heavy college drinkers and gamers are at risk.

4.1. Conclusions

The present study highlights the relevance of gender and ethnicity with respect to college DG participation and alcohol-related problems. Intervention efforts designed to address high-risk drinking behaviors like DGs on college campuses might consider paying close attention to ethnic minority populations, perhaps particularly Black women. Research examining the role of ethnicity and gender on DG behavior remains limited; as such, we still do not know for sure whether DGs pose a greater health risk for certain segments of the college student population. Considering that many college campuses are becoming increasingly more ethnically diverse, it is important that researchers and practitioners not only understand the association between high-risk drinking behaviors and health outcomes among different populations, but that they also consider the relevance of gender and ethnicity when developing or refining theoretical models of drinking behaviors.

HIGHLIGHTS.

• High-risk nature of gaming can be a function of typical/heavy alcohol use.

• Overall, men who play are at risk for experiencing alcohol-related problems.

• Black women who play are at risk for experiencing alcohol-related problems.