Drinking before sporting events in Australia: An ecological momentary assessment study

Abstract

Background

This study investigates the individual and event-level correlates of drinking prior to attending Australian Football League (AFL) games among a sample of Australian spectators.

Materials and Methods

A total of 30 adults (20% female, mean age = 32) completed a series of questionnaires (n = 417) before, during, and after an AFL match on a Friday, Saturday, or Sunday. Cluster-adjusted regression analyses were conducted to examine the impact of individual-level (age, gender, drinking habits) and event-level factors (time and day of game, location of viewing the game, viewing with friends or family) on drinking prevalence and the number of drinks consumed prior to the game.

Results

41.4% of participants engaged in drinking before attending an AFL match with a mean of 2.3 drinks consumed by those who reported pre-game consumption. Those aged 30 and over were significantly more likely to engage in pre-game consumption (OR = 14.44, p = 0.024) and consumed significantly more pre-game (B = 1.39, p = 0.030). Drinking before the game was significantly more likely before night games than daytime games (OR = 5.24, p = 0.039). Those who watched the game on-premise consumed significantly more before the game than those who watched the game at a private residence or at home (B = 1.06, p = 0.030). Those who watched games with family also drank significantly less prior to the game than those who attended without family (B=-1.35, p = 0.010).

Conclusion

Addressing the contextual factors associated with drinking before the sporting events, such as the time of the game, may assist with efforts to reduce risky alcohol consumption and related harm.

Keywords:

• Alcohol consumption
• ecological momentary assessment
• sport spectators
• Australian Football League

Introduction

Sports spectators drink at riskier levels than the general population and experience a greater risk of adverse consequences, especially while attending sporting events (Kingsland et al., 2015; Lloyd et al., 2011; Nelson and Wechsler, 2003; Poortinga, 2007). Moreover, alcohol is persistently promoted by sports clubs and at sports events (Sartori et al., 2018), and is ingrained in the cultural practices of many sporting codes (Munro, 2000; Palmer, 2011). Australian Rules Football League (AFL), one of Australia’s most popular sporting codes, exemplifies this nexus between alcohol use and sport (Palmer, 2011). In 2017, 83% of AFL clubs (i.e. 15 out of 18) had at least one alcohol industry sponsor (Sartori et al., 2018) and risky drinking among AFL spectators—at a community and professional level—has been widely reported (Thompson et al., 2006).

High levels of alcohol use at sporting events in Australia, and globally, are associated with a range of harmful outcomes, including increased rates of hospital admissions, assaults, and violence (Livingston, 2018; Lloyd et al., 2011; Palmer, 2011; Pradhan et al., 2021). Further, drinking before sporting events shares similar risks to pre-drinking generally, increasing evening-level heavy drinking and the risk of experiencing alcohol-related adverse consequences (Kuntsche & Labhart, 2013; LaBrie et al., 2016; Lawrence et al., 2012; Leavens et al., 2019; Miller et al., 2016). Thus, pre-game consumption is an important and amenable risk factor for heavy drinking and alcohol-related consequences among sport spectators.

Drinking before sporting events—sometimes called ‘tailgating’, ‘pre-gaming’, or ‘frontloading’—is suggested to be common practice internationally (Merlo et al., 2011). While there is a substantial body of literature concerned with pre-game alcohol consumption among sport spectators, most research has focused on college student drinking in the United States (US) (e.g. Glassman et al., 2007; Hustad et al., 2014; Nelson et al., 2010). Several studies have examined the demographic (i.e. age, gender, ethnicity, student status, etc) correlates of consumption among sport spectators (e.g. Barry et al., 2013; Glassman et al., 2007; Glassman et al., 2010; Haun et al., 2007; Lawrence et al., 2012; Merlo et al., 2011), the relationship between engagement in pre-game drinking and breath alcohol concentration (e.g. Barry et al., 2013; Erickson et al., 2011), and the impact of tailgating on subsequent consumption and/or negative consequences (e.g. Lawrence et al., 2012; Merlo et al., 2011). However, most studies have utilized cross-sectional methodologies to assess “typical” pre-game consumption among sports spectators at one time point (e.g. Glassman et al., 2007; Merlo et al., 2011; Nelson et al., 2010), and as such, there is a dearth of event-level research.

Previous research has shown that the social, environmental, and physical contexts of drinking occasions are important determinants of alcohol consumption (Stanesby et al., 2019). This includes factors such as where, when and with whom alcohol is consumed, as well as the affective and motivational states of the drinker (Stanesby et al., 2019; Stevely et al., 2020). Most studies that have considered contextual factors focused on measuring negative consequences, crime, or other harmful behavior associated with sport spectator drinking, rather than the drinking itself (e.g. Menaker & Chaney, 2014; Menaker et al., 2018; Popp et al., 2020). For instance, Manaker and colleagues (2014; 2018) found that alcohol-related stadium ejections in the US are significantly more frequent during night events, than during the day. Erickson et al. (2011) examined the blood alcohol concentration (BAC) of attendees at US football (NFL) games and found that average BAC was significantly higher during night and Monday games, compared to other time periods. However, as the study compared BAC across different samples of attendees, it was not possible to assess the impact of time or day on within-person variations in alcohol use.

Nonetheless, the retrospective methods used in most previous studies suffer from issues (e.g. recall and response bias) which can be addressed through Ecological Momentary Assessment (EMA) (Shiffman et al., 2008). EMA provides the opportunity to examine within-person variation in sport spectator drinking, providing insight into the occasion-level contextual factors associated with sport spectator drinking and pre-game consumption (Shiffman, 2009). To do so, this study aims to estimate predictors of (1) engagement in pre-game consumption, and (2) the mean number of drinks consumed pre-game utilizing repeated measures EMA data.

At the individual level, we assess the impact of age, gender and AUDIT-C; while at the occasion-level we examine the impact of the time and day of the game, whether participants watch the game with friends or family, and where they watch the game. Based on previous research (Ferris et al., 2019) we expect that men drink more, and are more likely to pre-game compared to women. Given pre-gaming is often motivated by a desire to save money by consuming alcohol before going out (Miller et al., 2016), we expect that those who watch the game at an on-premise venue consume more alcohol prior to watching or attending AFL games. Finally, by providing a more granular examination of both individual and situational predictors of sport spectator drinking prior to major Australian sporting events, it is hoped that insights relevant to improving alcohol management practices, and in turn, reducing alcohol-related harm in sporting contexts will be developed.

Materials and methods

Study design

Participants were recruited in Melbourne, Australia. Targeted social media (including Facebook and Instagram) advertisements were used in conjunction with a street intercept method at a large sports stadium in Melbourne (i.e. the MCG). Participants were provided with information about the study and completed an online screening questionnaire as well as providing personal details to facilitate re-contact from researchers later. A total of 704 participants entered and completed the screening survey, with most being approached at the MCG and watching their AFL team play weekly. The number of males and females approached was balanced, with 48% being female (Table 1). Participants who completed the screening questionnaire entered a prize draw to win one of five $200-dollar vouchers.

Table 1. Participant demographics.

	Full screening (N = 704)		Included sample (N = 30)
	Mean (SD)	Range	Mean (SD)	Range
Female (%)	48.4		20
Age	30 (11.4)	18–70	32 (11.3)	18–62
Watching AFL (%)
Never	28.6		0
Few times a year	24.9		6.7
Monthly	9.4		20
Weekly	37.2		73.3
Source (%)
MCG	57.7		13.3
Social Media	29.4		36.7
Word of mouth	10.7		50
Unknown	2.3		0
AUDIT-C score	5.6 (2.5)	0–12	6.2 (1.7)	3–9

Note: N = Number of participants, SD = Standard Deviation.

Research assistants telephoned prospective participants to provide further screening and study information; those eligible were asked to provide consent via email. Once consent was received, participants were provided instructions for how to download and use the RealLife Exp application. RealLife Exp is a smartphone application that facilitates EMA by allowing researchers to program notification-initiated surveys at designated times (www.lifedatacorp.com). RealLife Exp is functional on both Android and Apple smartphones.

AFL games are played mostly on the weekends and take place during the day (earliest start time is 1 pm) or in the evening (latest start time is 8 pm). They have four quarters of approximately half an hour duration (approximately 2.5 hours of game time in total including breaks between quarters). EMA data collection occurred over five consecutive weekends in August and September 2019. At 12 pm each Friday during the study period, participants received a survey notification and were asked to allocate one AFL game that they intended to watch for the weekend. Subsequently, EMA collection commenced 10 minutes prior to the allocated AFL match with another survey every break of the game and every 2 hours after the game with up to 10 questionnaires in total. A follow-up questionnaire was also administered at 12 pm the following day.

Sample

Inclusion criteria into the EMA component were (a) being over 18 years old, (b) being a regular drinker, (c) watching AFL at least fortnightly, and (d) owning a smartphone. Exclusion criteria were (a) being pregnant or planning pregnancy. A total of 34 participants participated in the study. However, one participant did not complete any of the pre-game surveys and was excluded from the analyses. For another 3 participants, demographic information was missing (including gender and age). Thus, the final sample included a total of 30 adults (20% female, Mean age ≈32) (Table 1) who completed a total of 58 pre-game surveys over five weeks.

Measures

Screening questionnaire at baseline

At baseline, participants provided their gender and age and answered the three questions of the Alcohol Use Disorders Identification Test—Consumption (AUDIT-C), i.e. frequency, quantity and heavy episodic drinking in the past 12 months? (Bush et al., 1998). Scores per question were summed to create an AUDIT-C score ranging from 0 to 12.

EMA questionnaire

Pre-game consumption

Each pre-game EMA assessment (10 minutes prior to the start of participants’ allocated AFL game) asked: ‘Have you consumed any alcoholic drinks today?’. If participants answered ‘yes’, they were then asked: ‘How many standard drinks have you had today up until now?’. Participants were shown an image which displayed pictures of a variety of common beverages accompanied by the number of standard drinks in each beverage. A standard drink in Australia contains approximately 10 grams of alcohol.

Day (Friday = 1, Saturday = 2, Sunday = 3) and time of the game (before 6 pm = 0 and after 6 pm = 1) were coded manually by the researchers.

Watched the game with family/friends

Participants were asked at each quarter during the AFL games: “What is your relationship to the people you are watching the game with?” (Friends/Family/Colleagues/Friends of Friends/Other). Dummy variables were then created for watched games with friends (=1, otherwise =0) and watched games with family (=1, otherwise = 0).

On-premise vs off-premise location during the game

Participants were asked at each quarter during the AFL games: “Where are you right now?” (Home/Someone else’s home/Pub/Bar/Sports stadium/Other). Viewing the game at a bar/pub (n = 7) and at the stadium (n = 17) was re-coded as on-premise (=1), whereas viewing the game at home (n = 27) or someone else’s home (n = 8) was coded as off-premise (=0). Participants who responded with “other” were asked to specify their location. These responses were then re-coded into the above two categories.

Data analysis

Cluster-adjusted chi-square and independent t-tests were conducted in STATA 16.0 (StataCorp, 2019) to assess significant differences in the number of drinks consumed and engagement in pre-game consumption, based on gender, age, AUDIT-C, time of game, day of game, presence of friends and family at the game, and location during the game. Then, a cluster-adjusted backward stepwise multiple regression model was conducted examining predictors of engagement in pre-gaming and number of drinks consumed before the game using the 58 pre-game surveys from 30 participants.

Results

Overall, pre-game consumption occurred before 41.4% of games. Participants who reported engaging in pre-game consumption consumed an average of 2.3 (SD = 1.6) standard drinks prior to the game. Table 2 shows the relationship between the individual level variables (gender, age, AUDIT-C) and engagement in pre-game consumption and number of drinks consumed prior to the game. The left column provides an overview of the entire sample, while the right presents findings from participants who reported pre-game consumption.

Table 2. Individual factors and pre-game consumption.

	Total pre-game EMAs (n = 58)		EMAs with pre-game drinking (n = 24)
	% drank before the game	Mean number of drinks before the game (SE)	Mean number of drinks pre-game (SE)
Total	41.38	0.95 (0.20)	2.33 (0.33)
Gender
Male	41.86	0.95 (0.34)	2.28 (0.41)
Female	35.71	0.93 (0.52)	2.60 (0.76)
Test value (sign.)	0.11 (0.742)	0.04 (0.968)	−0.37 (0.714)
Age groups (years)
18 to 29	37.14	0.91 (0.36)	2.46 (0.50)
30+	47.83	1.04 (0.44)	2.38 (0.66)
Test value (sign.)	0.43 (0.513)	−0.23 (0.823)	0.10 (0.918)
AUDIT-C
0-5	42.11	1.16 (0.52)	2.75 (0.63)
6+	36.11	0.81 (0.36)	2.23 (0.50)
Test value (sign.)	0.12 (0.728)	0.56 (0.583)	0.65 (0.529)

Note: n = number of EMA surveys. SE = Standard Error. Sign. = p-value.

Table 3 shows the relationship between the contextual factors and pre-game consumption and spending patterns. People who watched night (after 6 pm) games were significantly more likely to engage in drinking prior to the game, compared with those who watched afternoon (before 6 pm) games, but did not necessarily drink significantly more prior to the game. Those who proceeded to watch the game at an on-premise venue (i.e. a sports stadium or pub/bar) consumed significantly more pre-game than those who proceeded to watch the game off-premise (i.e. at one’s home or someone else’s). Further, those who watched the game with family consumed significantly less than those who watched the game without family.

Table 3. Contextual factors and pre-game consumption.

	Total pre-game EMAs (n = 58)		EMAs with pre-game drinking (n = 24)
	% drank before the game	Mean number of drinks before the game (SE)	Mean number of drinks before the game (SE)	Mean amount spent on alcohol pre-game (SE)
Total	41.38	0.95 (0.20)	2.33 (0.33)	18.38 (4.38)
Time of game
Before 6pm	25.81	0.52 (0.39)	2.00 (0.60)	16.50 (9.19)
6pm and later	59.26	1.48 (0.37)	2.50 (0.42)	19.31 (5.77)
Test value (sign.)	4.36 (0.037)	−1.81 (0.078)	−0.68 (0.502)	−0.259 (0.798)
Day of game
Friday	61.11	1.33 (0.35)	2.18 (0.40)	21.00 (7.60)
Saturday	36.36	1.05 (0.40)	2.88 (0.77)	14.50 (6.33)
Sunday	27.78	0.5 (0.25)	1.8 (0.58)	18.80 (9.49)
Test value (sign.)	1.65 (0.207)	2.50 (0.098)	1.17 (0.333)	0.29 (0.751)
Where game was viewed (location)
Off-premise viewing	34.38	0.53 (0.35)	1.55 (0.49)	14.64 (6.89)
On-premise viewing	54.17	1.63 (0.41)	3.00 (0.45)	21.54 (6.84)
Test value (sign.)	1.41 (0.234)	−2.03 (0.049)	−2.20 (0.041)	−0.71 (0.486)
Watched game with friends
Without friends	28.57	0.43 (0.48)	1.50 (0.66)	9.50 (9.21)
With friends	48.65	1.27 (0.35)	2.61 (0.38)	21.33 (5.64)
Test value (sign.)	1.46 (0.227)	−1.43 (0.162)	−1.45 (0.164)	−1.10 (0.288)
Watched game with family
Without family	48.57	1.34 (0.34)	2.76 (0.38)	16.47 (5.69)
With family	30.43	0.39 (0.42)	1.29 (0.59)	23.00 (8.89)
Test value (sign.)	1.23 (0.267)	1.78 (0.084)	2.12 (0.049)	−0.618 (0.544)

Note: n = number of EMA surveys. SE = Standard Error. Sign. = p-value.

First, a backward stepwise multiple linear regression was conducted. Gender, age, AUDIT-C, time of game, day of game, location, and viewing the game with family or viewing the game with friends viewing the game with family were regressed on the amount consumed pre-game (Table 4). Using the Wald chi-square inclusion threshold of 0.1, variables which exceeded the threshold were excluded from the model. Thus, the final model included the predictors age, time of game, viewing location and viewing the game with family. Then, a backward stepwise multiple logistic regression was conducted which included all of the above variables regressed on engagement in pre-game consumption (Table 4). The predictors age, time of game and viewing with family met the inclusion threshold for this model, and thus, all other variables were excluded.

Table 4. Cluster adjusted multiple linear and logistic regressions predicting engagement in pre-game drinking and the number of drinks before the game using backward stepwise elimination.

	Predictors of number of drinks consumed pre-game			Predictors of engagement in pre-game consumption
	Coefficient (Robust SE)	95% CIs	P-values	OR (Robust SE)	95% CIs	P-values
Individual level
Age (30+)	1.39 (0.60)	0.15, 2.63	0.030	14.44 (17.08)	1.42, 146.67	0.024
Contextual level
Time of game (6pm and later)	1.33 (0.44)	0.43, 2.22	0.005	5.24 (4.22)	1.08, 25.38	0.039
Location (on-premise)	1.06 (0.47)	0.11, 2.02	0.030	–		–
Watched game with family	−1.35 (0.49)	−2.36, −0.35	0.010	0.06 (0.08)	0.01, 0.79	0.032

Note: Robust SE = Robust Standard Error. 95% CI = 95% Confidence Intervals.

Similar trends were noted for both engagement in pre-game consumption and the amount consumed. In both cases, after accounting for the other factors, those aged 30 and over consumed significantly more alcohol and were more likely to engage in pre-game consumption. Likewise, the time of the game was a significant predictor of pre-game consumption, with those attending night games (after 6 pm) more likely to engage in pre-game consumption and to consume significantly more alcohol. The location where participants viewed the game was also a significant predictor of pre-game consumption patterns. Those who viewed the game on-premises (at a sports stadium or pub/bar) consumed significantly more alcohol than those who viewed the game off-premise. Finally, attending or viewing games with family appeared to have a protective effect, significantly reducing engagement in pre-game consumption and the amount consumed.

Discussion

This study aimed to further understand the effect of individual and situational factors on the pre-game drinking patterns of AFL spectators. Contrary to previous research (Haun et al., 2007; Ferris et al., 2019; Wilsnack et al., 2000), we did not find significant gender differences in pre-game consumption (quantity or frequency).

Further, the absence of a significant positive association between AUDIT-C scores and pre-game consumption conflicts with previous findings in the general pre-drinking literature (Barry et al., 2013; Howard et al., 2019). However, research on the impact of AUDIT-C on pre-game consumption among sports spectators is limited. Previous studies on trends among college football spectators in the US have shown that fans drink significantly more on game day than in other social contexts (Glassman et al., 2007). Considering AUDIT-C measures ‘typical drinking’ (Bradley et al., 2007), AUDIT-C may not accurately predict pre-game consumption trends among sports spectators. However, mean pre-game consumption and AUDIT-C was low in our sample, and did not exceed risky drinking levels (i.e. >4-5 standard drinks) (Hingson et al.,2017), which may explain the null effects. Nonetheless, further research on the impact of AUDIT-C on drinking among sports spectators on game days is required.

Interestingly, while age was non-significant in the bivariate analyses, after accounting for other factors in the regression models, those aged 30+ more frequently engaged in pre-game consumption and drank more when they did so. This may reflect broader trends regarding the decline in youth drinking in Australia (Pennay et al., 2018). Similarly, consumption among middle-aged Australians remains high which may also be driving the higher rates of pre-game consumption among those aged 30 and over (Livingston et al., 2016). Given we measured age dichotomously (due in part to the small sample size), we may also be missing some of the variation in pre-game consumption by age, and as such, further research is required to explore age effects.

Clearly, the contextual characteristics of consumption are important for understanding pre-game drinking behaviors among spectators, and drinking patterns generally (Stevely et al., 2020). In the multiple regression, whether participants went on to watch the game on-premise (i.e. at pubs or the stadium) or off-premise (i.e. in one’s home or someone else’s) had a substantial impact on the amount of alcohol consumed. This is not surprising considering Australian studies (Miller et al., 2016) report that the higher cost of on-premise alcohol is a significant motive for pre-drinking. Research has shown that off-premise alcohol is, on average, one-third of the cost of on-premise alcohol in Australia (Jiang et al., 2016). Further, to reduce heavy drinking and problems at Australian sports stadiums, the AFL have implemented relatively stringent alcohol management policies that may inadvertently encourage pre-game consumption. For instance, the alcohol content of beer is limited to mid-strength for evening AFL matches and the price of alcohol compared to other on-premise venues is substantially higher (Parry and Hughes, 2016).

Further, the significant effect of time of the game on the engagement in and amount consumed during pre-game alcohol consumption is in line with our hypothesis that most alcohol consumption and acute alcohol related harm occurs during the nighttime, and nighttime drinking is more frequent than daytime drinking (Livingston et al., 2016). Nonetheless, this has significant implications for alcohol management and safety practices, particularly at sports stadiums (discussed below).

Limitations

A strength of the study was the use of a robust repeated measures design (i.e. EMA) which allowed us to capture close to real time patterns in pre-game alcohol consumption. Additionally, we were able to assess (albeit in a small sample) how situational level factors affected pre-game consumption over time. Nonetheless, there are several limitations. EMA studies are prone to compliance issues and attrition, due to a relatively high response burden, as well as difficulties associated with administering surveys via novel and emerging technologies (i.e. mobile phone applications). Further, as this was a small study, the sample size and number of observations was small. This substantially limits the statistical power, and in turn, the ability to accurately infer the significance and effect size of factors, while also reducing the generalizability of the results.

Second, while this study documents trends in alcohol consumption prior to viewing AFL games, the pre-game location of participants was not recorded. Consequently, differences in pre-game consumption may be, at least partially, attributable to where participants consumed alcohol prior to the game. Finally, the last week of data collection coincided with the “finals” round of AFL. While the number of participants who completed the survey in the final round was low (n = 4), given tickets for the finals are limited, there may have been a reduction in the proportion of participants who were able to attend (and drink) at the game, potentially altering these participants’ pre-game drinking behavior. Despite these limitations, this study is the first to assess the correlates and patterns of pre-game consumption among AFL spectators and provides insights for future researchers interested in studying sport spectator drinking in Australia.

Implications

Studies in the US have shown that alcohol related incidents at college football stadiums are significantly higher during night games than day games (Menaker & Chaney, 2014; Menaker et al., 2018). Likewise, the increased frequency of pre-game consumption among Australian AFL spectators may suggest that night games pose a greater risk for hazardous consumption and alcohol related incidents. This emphasizes the need for robust and responsible alcohol management practices at sports stadiums and bars that host or televise night games.

Similarly, the protective effect of viewing AFL games with family members may provide some insight into strategies that the AFL, and venues that host the AFL, could consider for reducing alcohol related harm. By further endorsing sporting events as family-friendly, alcohol consumption and alcohol harms may be reduced.

Lastly, addressing alcohol sponsorship within the AFL is one avenue for reducing hazardous consumption by spectators. Alcohol promotion is a consistent determinant of risky drinking at sporting events internationally (Rowland et al., 2015). By minimizing alcohol promotion at AFL games, the risk of alcohol related incidents at games may be reduced (Kingsland et al., 2015).

Conclusion

This study provides the first tentative evidence on the pre-game consumption patterns of Australian AFL spectators. Based on our findings, it is clear that contextual factors, such as where, when and with whom individuals watch AFL games have a substantial impact on the amount of alcohol consumed before the game. Overall pre-game consumption was low in our sample. Nonetheless, any pre-game consumption increases the risk of increased adverse consequences as more alcohol is likely to be consumed at the event level, and therefore at sporting events also (Kuntsche & Labhart, 2013). Still, further research is required with a larger, representative sample to better understand the risk and protective factors associated with pre-game consumption before sporting events in Australia.

Author contribution

All authors contributed in a significant way to the manuscript and authors have read and approved the final manuscript.

Declaration of interest

The authors report there are no competing interests to declare.

Data availability statement

The data that support the findings of this study are available from the corresponding author, DAL, upon reasonable request.

Additional information

Funding

This research was supported by a La Trobe University internal grant. Dan Anderson-Luxford was supported by the graduate research scholarship (LTUPRS) from La Trobe University. Kelly van Egmond was 3 supported by the full fee research scholarship (LTUFFRS) and postgraduate research scholarship 4 (LTUPRS) from La Trobe University, Cassandra J. C. Wright PhD, was funded by a NHMRC Early 5 Career Fellowship (1161246), and Amy Pennay is supported by an Australian Research Council 6 Discovery Early Career Researcher Award (190101074).