Purpose of the study

India and Australia were chosen for this study based on the contrasting alcohol consumption features – 1) dry (India) versus wet (Australia) drinking cultures, 2) proportion of drinkers and/or per capita consumption increasing (India) versus decreasing (Australia) over the past decade, and 3) substantially different socio-cultural norms towards alcohol consumption. These features are discussed at length in the next section.

Harms related to underage drinking are a major problem in Australia [16] and a growing concern in India [1], and are likely influenced by exposure to alcohol marketing [1]. Most of the research exploring the association between alcohol marketing on SNS and alcohol use among young people has been conducted in the USA [12, 15], the UK [6, 14], and Australia [10, 17–21]. However, these studies explored these associations primarily on Facebook, with work involving other SNS such as Twitter and YouTube is sparse. Further, work in other national contexts such as India appears to be lacking, with a few studies identified [22, 23]. Jones et al. (2016) found significant and positive associations between reported exposure to alcohol advertising and branding on Facebook and reported drinking frequency and volume among 16–24 year old Facebook users [19]. Similarly, another study reported that exposure to Internet advertising was significantly associated with frequency of alcohol consumption among 12–17 year old Australians, with these associations varying across age and gender sub-groups [20].

As stated earlier, alcohol companies use marketing strategies on SNS that are tailored to specific national contexts and users’ responses to such marketing content differ by national settings [22–24]. These studies identified common strategies used for alcohol promotion across social media in both countries. These included prompts to engage in time- and event-specific drinking (e.g., “it’s Friday, and it’s beer-o’clock”), alcohol sponsorship of sporting, music, and fashion events, cocktail recipes and food-drink combinations, competitions, brand-related giveaways, and the use of memes. Other strategies were largely country specific, such as posting content relating to inspirational talks, livelihood skills, and sexually suggestive content on Indian social media sites versus references to a brand’s tradition or heritage on Australian sites. Notably, some of the identified strategies (e.g., brand-sponsored events and posts relating to competitions and giveaways) were traditionally more popular amongst younger people. User engagement was assessed through responses to content posted by brands, for example, user-generated messages, images, and videos posted on brands’ social media pages.

However, cross-national comparisons examining the effects of exposure to alcohol marketing on SNS and alcohol use among young people and across different SNS are scant. To extend this work, the present study investigated associations between 13 and 25 year olds’ exposure to and interaction with SNS-based alcohol marketing and alcohol use (assessed as usual consumption) in diverse national contexts (India and Australia), and across varying SNS. The selected SNS were all popular amongst young people, allowed for a variety of comparisons with existing work from the USA [12, 15], the UK [6, 14], and Australia [10, 17–21] and represented SNS not yet examined within the Australian and Indian literature. This information is important to guide the development of regulatory frameworks to minimize any harmful use of alcohol among young people that may result from exposure to alcohol marketing.

Socio-cultural norms and drinking prevalence and patterns

India and Australia have substantially dissimilar socio-cultural contexts and histories, and represent widely contrasting drinking cultures. In India, 11% of males and 1% of females aged 15–19 years and 29% of males and 2% of females aged 20–24 years report consuming alcohol in the past 12 months [25]. In contrast, in Australia 23% of those aged 12–17 years and 62% of those aged 18–24 years consumed alcohol in the past 12 months, with consumption rates being similar for males and females [26]. Similarly, heavy drinking is less prevalent in India than Australia. For example, about 4% of 18–24 year old Indians are classed as ‘heavy drinkers’, defined as consuming at least 50 g of pure alcohol in a single session at least once a month [27]. By comparison, in the same age group, 42% of Australians reported consuming 5 or more standard drinks (equivalent of 50 + g of alcohol) on a single drinking occasion at least once a month [26].

These differences in national consumption rates are likely due to a range of factors. Socio-cultural norms in India are less accepting of alcohol [28], especially in the context of female drinking [29], and religious practices that proscribe alcohol use (14% of the Indian population is Muslim) [28, 30]. The legal drinking age in India ranges from 18 to 25 years, with sales banned to the whole population in certain states [31]. However, with urbanization and industrialization, economic transition, exposure through extensive alcohol marketing, increased availability, and relaxation of overseas trade rules; alcohol use is increasing in India [31, 32]. Hence, both traditions and social change influence drinking culture in modern India [28, 33, 34]. In contrast, although alcohol has been a ubiquitous feature of Australian culture from its colonial era onwards, per capita consumption has been decreasing over the past 10 years [26, 35]. Consequently, although drinking is more prevalent in Australia compared to India, the drinking trajectory is on the rise in India and on decline in Australia [25, 26].

SNS use and regulatory environment

Reflecting the size of the overall populations, the sheer number of social media users varies greatly between India and Australia. Twitter has an estimated 23 million users in India (2% of the population) and 3 million in Australia (13% of the population), YouTube has about 40 million users in India (3% of the population) and 14 million in Australia (58% of the population), while Facebook has about 108 million users in India (9% of the population) and 11 million in Australia (46% of the population) [36–38]. Although different SNS vary in terms of user demographics, SNS use overall is more prevalent in the younger age groups. For example, in India about 46% of Facebook users are estimated to be 18–24 years of age and a further 11% are 17 years or younger [39], while 70% of YouTube users are below 35 years of age [40]. Similarly, 36% of Australian Facebook users belong to the 13–24 years age group [38] and about 51% of Australians aged 14–17 years reported using YouTube in 2013 [41]. Among Australian SNS users, Twitter is used by 25% males and 14% females. About 33% of them belong to the 18–29 years age group [42]. Although demographic data on Indian Twitter users are not publically available, it is suggested that about 80% of Twitter users are male [43].

The Advertising Standard Council of India (ASCI) regulates alcohol advertising in India. Although it bans both direct and indirect alcohol advertising in traditional media, it does not cover online alcohol advertising [44]. Therefore, alcohol advertising on SNS remains unfettered and is thus extensive in India [31]. In Australia, alcohol advertising (including on digital media) is self-regulated by the alcohol industry via the Alcohol Beverages Advertising Code [45]. However, the current code only relates to the content of alcohol advertisements, does not consider issues around placement of advertisements on digital media (including where there are no age restrictions), and has very weak enforcement powers [46]. Hence, in both countries SNS are largely unregulated platforms for alcohol companies to promote their products.

Respondents

Eligible respondents were those aged 13–25 years who had lived in India or Australia in the past 12 months, had Internet access, and could understand written English. The choice of this age range was guided by the World Health Organization’s and United Nations’ definition of adolescence and youth [47].

Two online surveys were developed, one for each country, that included items relating to demographic characteristics, drinking patterns (defined below), and involvement with alcohol marketing on Facebook, YouTube, and Twitter. There did not appear any academic study and/or national surveys that sought information relating to the objectives of the present survey. Hence, relevant items from existing studies and national surveys were reviewed and included in the present surveys. To cover the topics identified in the objectives, additional items were developed and included in the surveys. Questions on country-specific types of alcohol differed between the surveys because the types of alcoholic beverages consumed differ between countries. A standard drink chart was provided to inform respondents of standard drink measures (one standard drink = 10 g of alcohol in both countries) to facilitate accurate self-reporting of alcohol consumption levels. The surveys were pilot tested with 20 eligible respondents from each country (10 each in the 13–17 and 18–25 years age groups). The surveys were delivered using Qualtrics Survey Software between March and October 2016.

Survey advertisements were posted on social networking sites, sports clubs’ official social media pages, and youth organizations’ official social media pages, to recruit the respondents. Sports clubs and youth organizations also sent out study invitations via email to their members. Alcohol- and drug-related websites and fora (e.g., Family drug support, Bluelight, and Pill Reports) and a word-of-mouth strategy were also utilized to increase recruitment. Ethics approval was obtained from a University Human Research Ethics Committee and all respondents provided online informed consent. Qualtrics survey software’s ‘anti ballot-stuffing’ option was used to prevent the same IP address from accessing the survey more than once, to prevent duplicate survey entries. Further, as no incentive was offered to the respondents, it was unlikely that an individual respondent would complete the survey more than once, thus further decreasing the chances for duplicate survey entries. Providing an incentive for survey participation would also require collecting respondents’ potentially identifiable information, such as an email address, which could compromise their anonymity. As alcohol consumption is a relatively sensitive issue in India, especially among young people, we chose to keep the participants’ information non-identifiable.

Based on a power calculation, 385 respondents were needed for the survey, from each country (95% CI, with a 5% margin of error and response distribution of 50%) [48]. In total, 680 people opened the survey link of whom, 631 (330 Indians and 301 Australians) respondents who met the eligibility criteria were included in the analysis. The survey response rate was 93% (631/680).

Measures and statistical analyses

The results are presented in Table 1 stratified by country, gender, and age group (13–17 years, 18–25 years). Age was split at the boundary of the legal purchase age of alcohol (18 years) in Australia (and some Indian states). In addition, marked differences in patterns of alcohol use were anticipated across the age range. Differences in usual drinks consumed were assessed using Mann-Whitney U tests due to the skewed data distributions. Results were compared between age-groups and between genders within each country and finally between countries. Descriptive data on SNS interactions (Table 2) and other differences in proportions were evaluated with Fisher’s exact test due to the low frequencies in some cells.

To examine the relationship between respondents’ alcohol use and involvement with alcohol marketing on SNS, we used a standard model building approach (Hosmer and Lemeshow, 2000). Adjusting for key demographic factors (age, gender, and education), each variable was entered separately with the outcome variable (usual consumption levels). Those with a p value of <.20 were retained for multivariate analysis [19]. This process was conducted separately for the Indian and Australian data.

Six multiple linear regression analyses were conducted to assess the associations for the outcome variable for each country for each of the three SNS. Therefore, to adjust for multiple significance testing we only interpret values of p ≤ .008 (0.05/6). For each analysis, multivariate outliers were identified and excluded using Mahalanobis distance (p < .001); excluded cases ranged from 0 to 3 per analysis. Due to the high correlation between age and education level, the multivariate analyses were repeated excluding education. No variable moved across the threshold (p ≤ .008) for interpretation in the second analyses compared with the first. Only the first set of analyses is reported.

The following variables were included in the multivariate analyses. The daily hours spent on the Internet were analyzed as the mean of the range of the survey response levels: < 1 (coded as 0.5), 1–2, 3–4, 5–8, and > 8 (coded as 8.5 h thus winsorizing the data). Perceived trends in alcohol advertising and other alcohol-related marketing efforts on SNS in the last 12 months were coded into four categories (increased, unchanged/no difference, decreased, and don’t use this channel). Frequency of noticing alcohol brand logos on merchandise shown on SNS was dichotomized (due to small cell sizes when analyzed by age-group and sex) as ‘very often or often’ and ‘sometimes, rarely, never’. Information on whether respondents had attended events in the last 12 months (social, music, sports, and other) sponsored by alcohol companies via advertising on SNS (sometimes termed ‘real-world tie-ins’) was analyzed as a dichotomized variable again due to small cell sizes when stratified (attended any event - yes/no).

The frequency of receiving suggestions on each of the three SNS to like or to follow alcohol company pages or to participate in alcohol-related events was coded as ‘less than once a week’, ‘weekly’, ‘every fortnight’, ‘monthly’, and ‘don’t use this channel’. Information on the degree to which respondents posted and viewed alcohol-related content on each SNS was dichotomized (due to small cell sizes) as ‘very often or often’ and ‘sometimes, rarely, or never’. Finally, the degree to which the respondents’ friends or contacts post alcohol-related content on each SNS was coded in the same fashion. SPSS version 22 software was utilized for data analysis. Missing values were managed by using a listwise-deletion approach.

Demographics and alcohol consumption

In the Indian sample (n = 330), 36% were female and the median age was 20 years (IQR 17–22 years). In the Australian sample, 62% were female and the median age was 20 years (IQR 18–23). Most respondents who completed the Australian survey identified themselves as Australian citizens (82%), whereas nearly all (99%) who completed the Indian version were Indian nationals.

The majority of the Indian (93%) and nearly all Australian respondents (99%) reported having an Internet connection at home. The median daily Internet use was estimated at 1.5 and 3.5 h among younger and older Indians (across gender), respectively. In Australia, usage was 3.5 h across demographic groups, except for older males where it was estimated at 6.5 h (Table 1). Overall, the median time on the Internet was significantly lower in India than Australia (median 1.5 vs 3.5 h U 6.8, p < .001).

The mean age of first full serve of alcohol was 16.5 (SD 3.5) years in India and 15.2 (SD 2.8) years in Australia, with the earliest age being eight years in both countries. The lifetime prevalence of alcohol use was lower among younger Indians (47% for males and 52% for females) compared to their Australian equivalents (82% for males and 69% for females). Similar patterns were observed for recent drinking (i.e., in the last 4 weeks), across demographic groups and countries (Table 1). The overall prevalence for lifetime and recent drinking was significantly lower in India than Australia (Fisher’s Exact test p < .001 for both measures).

The median usual drinking levels were significantly higher among males than females in both countries (India median 5.5 vs 3.5: Australia median 3.5 vs 2.0; Table 1). Further, older Indian males drank significantly more than their younger counterparts (5.5 vs 3.5 drinks) and older Australian females had significantly higher usual drinking (2.0 vs 1.0 drinks) than their younger equivalents. Overall alcohol consumption quantity was significantly higher in India than Australia (median 4.5 vs 2.75 drinks U 10.6 p < .001). A post-hoc analysis controlling for Internet use, age, and gender found that Indian usual alcohol consumption was still significantly higher than Australian usual consumption (F = 13.0 (1508) p < .001).

Exposure to and interaction with alcohol marketing on SNS

Descriptive data on the variables related to exposure to alcohol marketing on SNS and included in the multivariate analyses are presented in Table 2. Across all these variables, Indian respondents reported significantly greater involvement with alcohol-related marketing. Perceived trends in alcohol advertising differed by country. In Australia, the younger participants reported seeing significantly increased amounts of alcohol advertising and other alcohol-related marketing efforts on SNS in the last 12 months. In contrast, older Indians noticed a larger increase than younger Indians.

In both India and Australia, older respondents reported receiving more frequent suggestions on SNS to like or to follow pages or to participate in events with alcohol-related content compared to younger respondents. Finally, on most variables there was greater sharing or interaction on Facebook, followed by YouTube and then Twitter, and this was consistent across demographic groups and countries.

Associations between usual alcohol consumption and alcohol-related interactions on SNS

Multivariate analyses were conducted to determine which influences were significantly associated with usual alcohol consumption quantity. In total, six multivariate models were developed (two countries by three SNS platforms, as items were specific to each SNS).

In addition to age, gender, and education, the Indian univariate analyses identified the following eight variables for inclusion in the multivariate analyses: (i) daily hours spent on the Internet; (ii) noticing alcohol brand logos on merchandise shown on SNS; (iii) attending alcohol industry sponsored events, (iv) sharing own alcohol-related information on SNS; (v) receiving suggestions to interact with alcohol-related content; (vi) perceived changing trends in alcohol advertising; (vii) respondents’ SNS friends/contacts sharing alcohol-related information on the particular SNS; and (viii) noticing alcohol-related information on SNS.

All three Indian multivariate models were significant, and accounted for 23–28% of variation in usual alcohol use. Only one variable, age, was associated with usual alcohol quantity across the three analyses, controlling for all other factors (p < .004-p < .003). In the Facebook and Twitter models, consumption increased by about 1/3rd of a drink for each year increase in age, but in the YouTube model it decreased by the same amount for each year of age.

Each of the overall models for the Australian analyses was significant, accounting for between 13 and 40% of the variance in alcohol use. In addition to age, gender, and education, univariate analyses identified seven variables for inclusion in the multivariate analyses: (i) noticing alcohol brand logos on merchandise shown on SNS; (ii) attending alcohol industry sponsored events, (iii) sharing own alcohol-related information on SNS; (iv) receiving suggestions to interact with alcohol-related content; (v) perceived changing trends in alcohol advertising; (vi) respondents’ SNS friends/contacts sharing alcohol-related information on the particular SNS; and (vii) noticing alcohol-related information on SNS.