Likelihood of Posting Alcohol Related Content on Social Networking Sites – Measurement Development and Initial Validation

Abstract

The vast majority of adolescents and young adults are active on Social Networking Sites (SNSs). SNSs are influential, risk-conducive environments for alcohol use among adolescents and young adults. Specifically, posting or sharing alcohol-related content (ARC) is associated with higher levels of alcohol use. However, it is unknown if sharing different types of ARC associates differentially with alcohol use and consequences. The goal of the current project was to develop a measure of the likelihood of posting key types of ARC posted by adolescents and young adults and to examine their associations with SNS use patterns and actual alcohol-related behavior. Participants were 15 to 20 years of age (n=306; 46.7% male; 56.6% Caucasian/White; 27.0% Asian) who completed a battery of self-report measures. Results from an Exploratory Factor Analysis revealed four types of ARC: (1) Self and Friend Consumption, (2) Memes and Viral Photos, (3) Status Updates: Others’ Drinking and Consequences, and (4) Pictures: Others’ Drinking and Consequences. Participants’ likelihood of posting Self and Friend Consumption was significantly associated with heightened Snapchat use, typical drinks per week, peak drinking, and negative drinking consequences. Whereas youth appear to share more readily alcohol-related viral posts and memes, it seems that the sharing of ARC that is specifically related to the participants’ own use or friends’ use is salient concerning alcohol use and problems. Therefore, interventions might consider sending targeted prevention messages to individuals who share certain types of ARC which are more associated with problematic alcohol behaviors.

1.1. Introduction

In the United States, adolescent and young adult alcohol use remains a major public health concern given that approximately 19% of individuals aged 12 to 20 years report having consumed alcohol in the past 30 days, and 11.4% reporting having consumed alcohol in the form of heavy episodic drinking (HED; defined as having five drinks for men or four drinks for women in the span of two hours; SAMHSA, 2020). Research concerning the initiation and progression of alcohol use indicates that most adolescents and young adults experiment with alcohol and that such experimentation can lead to later hazardous alcohol use during young adulthood (Bolland et al., 2016; Eaton et al., 2012; Hingson et al., 2002, 2003). As such, it is important to examine factors related to alcohol use among this group in order to determine viable prevention and early intervention strategies.

One potential risk factor associated with adolescent and young adult alcohol use is social networking site (SNS) use. The majority of adolescents and young adults are active on SNSs, with research indicating that among individuals aged 18 to 24 years old, 68%, 71%, 78%, and 45% are regular Facebook, Instagram, Snapchat, and Twitter users, respectively (Smith & Anderson, 2018). In addition, a majority of individuals 18 to 24 years of age use Tik Tok (55%; Pew Research, 2021). Moreover, individuals spend a remarkable amount of time on SNSs, with an average of 136 minutes per day (Statista, 2019), a marked increase from 90 minutes of average use in 2012 (Statista, 2019). However, it is important to note that SNS use is an important issue for adolescents and young adults not only due to the number of users and time spent on it, but also due to the way users interact with each other while on these sites.

Prior research suggests that in part due to the significant amounts of time adolescents and young adults report spending on SNSs combined with the multitude of opportunities to communicate about alcohol while using these sites, SNSs should be considered influential risk-conducive environments for adolescent and young adult alcohol use (McCreanor et al., 2013; Moreno & Whitehill, 2014). Further, research indicates that many SNS profiles contain alcohol-related content (ARC), a majority of which portray alcohol use in a positive light (Cavazos-Rehg et al., 2015; Moreno et al. 2013, 2016). Importantly, both experimental and longitudinal studies demonstrate that sharing SNS ARC are significantly associated with increased risky drinking cognitions, alcohol use, and related negative consequences (Hendriks et al., 2021; Hoffman et al., 2017; Geusens & Beullens, 2017, 2021a, 2021b; Litt et al., 2018; Moreno et al., 2016; Trager et al., 2022; Steers et al., 2021). Research further suggests that ARC likely reflects both the poster’s actual alcohol use and their perceived norms for drinking (D’Angelo et al., 2014; Geusens & Buellens, 2016; Westgate et al., 2014). Given the reach and impact that SNSs are likely to have on adolescent and young adult alcohol use, research suggests that SNSs likely represent the “super peer” (Strasburger et al., 2009) in that their influence extends further than a normal peer or friend. Thus, examining ARC posted on SNSs may give us a glimpse into a particularly powerful channel for peer and social influence.

Although posting ARC on SNS is somewhat ubiquitous, research also indicates that there are varying motivations to post ARC on SNS (i.e., entertainment, information, social, identification; Hendriks et al., 2017). Research also indicates that there may be differences in the type of content posted. Hendriks and colleagues (2018) suggest that ARC in which alcohol was in the background of photos were most often posted to SNSs, followed by text or post captions about ARC and ARC that contained alcohol advertisements. However, the same study found that ARC about drunkenness or drinking-games were almost nonexistent (Hendriks et al., 2018). Although this research provided some insight into different types of ARC, they notably did not address whether the ARC posted referred to the participant themselves, close friends, or strangers. Given that research indicates that more proximal peers have a stronger impact on risk behavior (Lewis et al., 2006), it is important to determine if the likelihood of ARC posts differs based on whom the post is about and whether these different categories are associated with alcohol consumption. Thus, research is needed to examine how different types of ARC, including posts about self and others, are associated with frequency of SNS use as well as actual alcohol-related behavior. This research will help us gain a better understanding of what specific types of ARC are associated with general SNS use patterns as well as alcohol consumptions and problems, which in turn has potential to lead to more precise targets for intervention. Specifically, understanding which types of ARC are both more likely to be posted and more likely to be associated with the poster’s drinking and related outcomes has potential to indicate which types of ARC on SNSs may be of more concern and thus more important to target in brief interventions.

The present study had two primary aims. The first was to develop a measure to examine the likelihood of adolescents and young adults posting different types of ARC to SNSs. The second aim was to determine if the likelihood of posting different types of ARC on SNSs were associated with SNS use patterns, alcohol consumption, and consequences among a sample of adolescents and young adults.

2.1. Material and Methods

2.1.1. Participants and Procedure

Data were collected from March 2017 to April 2018 as baseline data from a larger longitudinal experimental study that aimed to understand the relationship between alcohol displays on SNSs and subsequent alcohol cognitions, use, and related negative consequences among 15–20 year olds.

Recruitment efforts for this study were conducted locally in the Seattle-metro area through various methods, such as online and print advertisements, friend referrals, email invitations via a local university’s registrar’s list, and flyers at local coffee shops, libraries, and high schools. Recruitment materials provided a link to the study website, study contact information (i.e., email and phone number), and a direct link to the online screening survey where participants could obtain information about the study. Participants who accessed the online screening survey were required to provide consent in order to proceed to the online screening survey. Of the 1,017 participants who completed the online screening survey, 543 (56.4%) met initial eligibility criteria for the larger study, which included: 1) be 15 to 20 years old, 2) live in the Seattle-metro area, 3) drank at least once within the past 6 months (applied only to 18–20 year olds; no drinking criteria for 15–17 year olds), 4) use Facebook, Snapchat, or Instagram at least weekly, and 5) be willing to attend two in-person sessions as part of a larger experimental study. Participants aged 15–17 were required to have parental consent to participate. Parental consent was obtained from 93% of those 15–17 year olds who expressed interest in the study. Participants who were deemed eligible and chose to participate were invited to schedule the first in-person session at the lab, where they completed a baseline survey, from which the current data is drawn. The baseline assessment was provided prior to randomization to condition and prior to any experimental manipulation. All participants were compensated $30 for the completion of the baseline assessment. Study procedures were approved by the local IRB and no adverse outcomes were reported.

A total of 306 participants completed the in-person baseline session. For the current investigation, only those who reporting having consumed alcohol in the past month (N = 274) were retained for analyses. At baseline, participants in the analytic sample were, on average, 18.44 years old (SD = 1.31) and 46.7% (n = 128) reported being male. The sample was diverse with 56.6% (n = 155) reporting being Caucasian/White, 27.0% (n = 74) Asian, 10.2% (n = 28) more than one race, 2.9% (n = 8) Black, and 2.2% (n = 6) other; 9.2% (n = 26) identified as Hispanic/Latino.

2.1.2. Existing Measures

2.1.2.1. Demographics.

The participants responded to the following demographic questions: age and birth sex (0 = female, 1 = male).

2.1.2.2. Social Media Use.

For the three most popular SNSs among young people at the time of data collection (2017–2018) - Facebook, Snapchat and Instagram (Smith & Anderson, 2018), participants reported how many times per day that they checked the platform (0 to 7+ times).

2.1.2.3. Alcohol Consumption.

The Daily Drinking Questionnaire (DDQ; Collins, Parks, & Marlatt, 1985) was used to assess number of typical drinks per week. Participants were asked to “Consider a typical week in the past month. How much alcohol, on average (measured in number of drinks), do you drink on each day of a typical week?” Weekly drinking was computed by summing the number of standard drinks for each day of the week. For peak drinks per occasion, a single item from the Quantity/Frequency Index (DDQ; Marlatt et al., 1995) was utilized and participants were to “Think of the occasion you drank the MOST in this last month. How MUCH did you drink? Participants responded on a scale from “0” to “25 or more” standard drinks. For all items, a standard drink was defined as: 5 oz. of wine, 12 oz. of beer, 10 oz. of wine cooler, or 1 oz. of 100 proof liquor (NIAAA, 2020).

2.1.2.4. Brief Young Adult Alcohol Consequence Questionnaire (B-YAACQ; Hurlbut & Sher, 1992).

The B-YAACQ consists of 24 items that assess a range of alcohol-related consequences (e.g., spent too much time drinking; felt bad about oneself; become overweight). For each item, participants indicated if in the past month they experienced the consequence due to their drinking (Yes = 1 and No = 0). For the current sample, Cronbach’s alpha was .84 (M = 5.31; SD = 4.11).

2.2. Measurement Development Procedure

The measurement development process was adapted from DeVellis (2011). Specifically, DeVellis (2011) suggests a procedure for reliability, validity, item generation, and item/scale evaluation. The items for the Likelihood of Posting ARC Scale were developed in conjunction with a review of the international literature, inclusive of 2010–2017, related to types of content frequently posted on SNSs as well as stemming from the investigators’ previous work examining SNS alcohol displays. The authors had undergraduate research assistants and staff members who were familiar with SNSs review and examine the face validity the resulting 37 items. Minor wording changes were made prior to finalizing the measure. Higher scores indicate that the participant is more likely to post a given type of alcohol-related content on SNSs based on the following prompt: “In the next month, how likely are you to post each of the following?”. Participants responded to the items using a 5-point Likert scales from 1 = “Very Unlikely” to 5 = “Very Likely.”

2.3. Data Analysis Plan

A number of initial tests examined the fit of the data for the analyses warranted for this investigation. Alpha = .05 was used for all statistical tests.

To examine the structure of the Likelihood of Posting ARC measure, we performed a series of analyses in an effort to retain the best performing items and still retain the breadth of the measure. We proceeded with item development using information from literature, descriptive statistics (means, standard deviations, skewness, and kurtosis), principal components analysis, and Cronbach’s alpha (Cronbach, 1951). In the first phase, we removed items if their means and standard deviations indicated that the items did not vary across participants or represent the full spectrum of response options. Additionally, we examined the normality of the items using skew and kurtosis. During the second phase and for our exploratory factor analysis, we used principal components analysis (PCA) and principal factor axis using a varimax rotation and parallel analysis to help determine the number of components to retain (Zwick & Velicer, 1986). We retained items on components if they were fully saturated (loadings >.5) and not complex (difference of loadings on each component >.2). In this same phase, we removed items if they suppressed the Cronbach’s alpha. We continued this iterative process until the PCA, alpha, and theory indicated a stable structure remained.

3.1. Results

3.1.1. Descriptive Statistics

Participants indicated that on average, they consumed 7.75 standard typical drinks per week (SD = 8.74) and 6.44 (SD = 4.22) drinks on their peak occasion in the past month. In addition, participants reported an average of 5.31 (SD = 4.11) negative consequences in the past month. In general, participants reporting using Snapchat more frequently than Facebook and Instagram per day. See Table 1 for the frequency of use across platforms.

Table 1.

Frequency of Social Media Consumption

	0 times	1 time	2 times	3 times	4 times	5 times	6 times	7+ times
Facebook	8.0(22)	22.1(58)	17.9(49)	12.0(33)	8.0(22)	3.6(10)	1.5(4)	17.5(48)
Instagram	6.3(16)	11.5(29)	19.4(49)	15.9(40)	13.5(34)	6.7(17)	2.4(6)	24.2(61)
Snapchat	2.0(6)	5.6(17)	7.5(23)	5.2(16)	8.5(26)	6.9(21)	3.3(10)	58.4(160)

Note. Number of times per day in the past week that the platform was checked; % (n).

3.1.2. Exploratory Factor Analysis

The measurement development matrix of potential items was examined using Bartlett’s test of sphericity (Bartlett, 1950), which tests if the correlation matrix is an identity matrix, and the Kaiser-Meyer-Oklin test (Kaiser, 1974), which is a measure of sampling adequacy. Bartlett’s test indicated that the matrix was not an identity matrix, χ²(666) = 10732.58, p < .001, and the KMO test (KMO= .83) was in the “meritorious” category (Kaiser, 1974). According to this evidence, measurement development can proceed.

A 37X37 correlation matrix of the potential items (see Table 2 for a full list of items) was used for the initial Exploratory Factor Analysis (EFA). After an iterative series of descriptive statistics, EFA, and Cronbach’s alpha, 26 items remained. The final EFA resulted in a four-factor structure. The final structure was converged upon using information from the maximum likelihood, principal components analysis, and principal factor axis solutions. The resulting factors were named (Self and Friend Consumption; Memes and Viral Photos; Status Updates: Others’ Drinking and Consequences; and Pictures: Others’ Drinking and Consequences) and accounted for 73.92% of the variance. “A picture of yourself holding an alcoholic drink” is an example item from the Self and Friend Consumption subscale. The Meme and Viral Photo subscale addresses various types of alcohol posts that gained popularity online (example item: “A viral photo showing a person who is drinking alcohol.”) The Status Updates: Others’ Drinking and Consequences subscale describes alcohol posts that describe non-personal alcohol behaviors (example item: “A status about the consequences of the drinking of someone you don’t know (i.e., passing out, doing something they regret, missing class or work responsibilities).”). The Pictures: Others’ Drinking and Consequences subscale addresses posts with images of alcohol behaviors (e.g., “A picture of someone you don’t know who is drinking alcohol.”). The factor loadings and Cronbach’s alphas from the final EFA are in Table 2. Due to low levels of endorsement, six items were removed that referenced celebrity ARC. Additional items that were removed are in Table 2. The resulting scales has a Flesch –Kincaid grade level of 7.0.

Table 2.

Exploratory Factor Analysis

	Self and Friend Consumption α = .94	Memes and Viral Photos α = .96	Status Updates: Others’ Drinking and Consequences α = .85	Pictures: Others’ Drinking and Consequences α = .91
1. A status that mentions your own alcohol use.	0.758	0.137	0.185	−0.145
2. A status that mentions that you’re drunk.	0.746	0.018	0.204	−0.146
5. A status that mentions the alcohol use of someone you don’t know.	0.194	0.081	0.772	0.228
6. A status that mentions that someone you don’t know is drunk.	0.118	0.128	0.809	0.294
10. A status about the consequences of a friend’s drinking (i.e. passing out, doing something they regret, missing class or work responsibilities).	0.26	0.148	0.701	0.087
11. A status about the consequences of the drinking of someone you don’t know (i.e. passing out, doing something they regret, missing class or work responsibilities).	0.072	0.137	0.823	0.175
12. A picture of yourself drinking alcohol.	0.856	0.043	0.008	0.071
13. A picture of yourself visibly drunk.	0.772	0.01	0.009	0.111
14. A picture of yourself holding an alcoholic drink.	0.871	0.055	0.07	0.059
15. A picture of you with someone else who is drinking.	0.814	0.11	0.155	0.096
16. A picture of you with someone else who is visibly drunk.	0.752	0.148	0.108	0.148
18. A picture of a friend who is drinking alcohol.	0.805	0.168	0.101	0.227
19. A picture of a friend who is visibly drunk.	0.726	0.154	0.068	0.272
20. A picture of a friend who is holding an alcoholic drink.	0.834	0.17	0.119	0.143
22. A picture of someone you don’t know who is drinking alcohol.	0.181	0.055	0.181	0.895
23. A picture of someone you don’t know who is visibly drunk.	0.121	0.117	0.161	0.890
24. A picture of someone you don’t know who is holding an alcoholic drink.	0.165	0.074	0.188	0.860
25. A picture of someone you don’t know who is passed out from drinking alcohol.	0.019	0.167	0.222	0.735
30. A meme showing a person who is drinking alcohol.	0.181	0.888	0.054	0.052
31. A meme showing a person who is visibly drunk.	0.176	0.885	0.072	0.068
32. A meme showing a person who is holding an alcoholic drink.	0.171	0.881	0.08	0.061
33. A meme showing a person who is passed out from drinking alcohol.	0.17	0.842	0.083	0.102
34. A viral photo showing a person who is drinking alcohol.	0.062	0.901	0.062	0.092
35. A viral photo showing a person who is visibly drunk.	0.055	0.898	0.138	0.082
36. A viral photo showing a person who is holding an alcoholic drink.	0.066	0.891	0.081	0.049
37. A viral photo showing a person who is passed out from drinking alcohol.	−0.005	0.859	0.124	0.086
Removed Items
3. A status that mentions a friend’s alcohol use. (removed due to complexity)
4. A status that mentions that a friend is drunk. (removed due to complexity)
7. A status that mentions a celebrity’s alcohol use. (removed due to low loading)
8. A status that mentions that a celebrity is drunk. (removed due to low mean and low loading)
9. A status about the consequences of your drinking (i.e. passing out, doing something you regret, missing class or work responsibilities). (removed due to complexity)
17. A picture of yourself passed out from drinking. (removed due to low loading and low mean)
21. A picture of a friend who is passed out from drinking alcohol. (removed due to complexity)
26. A picture of a celebrity who is drinking alcohol. (removed due to low mean)
27. A picture of a celebrity who is visibly drunk. (removed due to low mean)
28. A picture of a celebrity who is holding an alcoholic drink. (removed due to low mean)
29. A picture of a celebrity who is passed out from drinking alcohol. (removed due to low mean)

3.1.3. Bivariate Associations with Validation Measures

The new scales were examined with regards to frequency of SNS use, alcohol consumption (drinks per week, peak drinks) and negative consequences. See Table 3 for the correlations, means, and standard deviations. Participants that reported that they were more likely to post content that mentions their own or their friends’ alcohol use (i.e., Self and Friend Consumption) also reported greater use of Instagram and Snapchat as well as more drinks per week, peak number of drinks, and negative consequences. However, being more likely to post memes and viral photos was only associated with using Snapchat more frequently, whereas likelihood of posting about other’s drinking consequences was only correlated with reporting more typical drinks per week and more negative consequences. The Status Updates: Others’ Drinking and Consequence scale was not related to time spent on SNS but was correlated with typical drinks per week and negative consequences.

Table 3.

Correlations between Scales and Validation Scales and Items

	1.	2.	3.	4.	M (SD)
1. Self and Friend Consumption	--				1.56 (0.81)
2. Memes and Viral Photos	.28***	--			1.66 (0.93)
3. Status Updates: Others’ Drinking and Consequences	.28***	.24***	--		1.21 (0.55)
4. Pictures: Others’ Drinking and Consequences	.36***	.29***	.46***	--	1.24 (0.54)
How many times per day did you check Facebook in the past week?	.11	.09	.04	.03	3.06 (2.34)
How many times per day did you check Instagram in the past week?	.14**	.04	.04	.05	3.69 (2.29)
How many times per day did you check Snapchat in the past week?	.21**	.13*	.06	.03	5.53 (2.12)
Past Month Typical Drinks per Week	.32***	.04	.16*	.17*	6.97 (8.59)
Past Month Peak Drinks per Occasion	.27***	.02	.06	.08	6.44 (4.22)
Past Month Negative Consequences	.34***	.04	.12*	.12*	5.31 (4.11)

Note.

^*p < .05

^**p < .01

^***p < .001; M = mean; SD = standard deviation

3.1.4. Regressions

Preliminary analyses revealed non-normal distributions for the alcohol outcomes (typical drinks per week, peak drinks per occasion, negative consequences). Because of the violation of normality assumption and the positive skew of the data, negative binomial regression was selected as the primary analysis strategy for all outcomes (Atkins, Baldwin, Zheng, Gallop, & Neighbors, 2013; Atkins & Gallop, 2007; Hilbe, 2011). Thus, we used the generalized linear modeling approach with the distribution specified as negative binomial (i.e., negative binomial regression) for these outcomes. Birth sex and age were included in all analyses as covariates based on previous associations with alcohol consumption (O’Malley & Johnston, 2002).

Typical Drinks per Week.

For the model examining typical drinks per week, the likelihood ratio for the full model was X² (9) = 57.46, p < .001, which indicated that the overall model was significant. Results of the negative binomial regression indicated that both age and being male were associated with greater drinks per week. Further, while more frequent use of Snapchat was associated with drinks per week, there were no significant associations with frequency of Facebook or Instagram use. Finally, results indicated that of the four new subscales of the Likelihood to Post ARC measure, only the Self & Friend Consumption subscale was significantly associated with typical drinks per week (see Table 4).

Table 4.

Negative Binomial Regressions

Predictor	B	SE B	z-value	p-value	Ratio	(95% CI)
Criterion Variable: Past Month Drinks per Week
Sex	0.70	0.15	4.49	0.01	2.01	(1.48, 2.74)
Age	0.24	0.08	3.20	<0.01	1.27	(1.10, 1.48)
Frequency of Facebook Use	0.04	0.04	1.22	0.21	1.05	(0.97, 1.12)
Frequency of Snapchat Use	0.08	0.04	1.99	0.04	1.08	(1.01, 1.16)
Frequency of Instagram use	0.01	0.04	0.01	0.99	1.01	(0.93, 1.16)
Self & Others Consumption	0.38	0.10	3.48	0.01	1.45	(1.18, 1.80)
Memes & Viral Photos	0.04	0.08	0.49	0.63	1.04	(0.88, 1.23)
Status Updates: Other’s Drinking & Consequences	0.04	0.17	0.26	0.79	1.04	(0.75, 1.46)
Pictures: Others’ Drinking & Consequences	0.10	0.15	0.63	0.53	1.10	(0.82, 1.49)
Criterion Variable: Past Month Peak Drinks per Occasion
Sex	0.58	0.10	6.01	<0.01	1.78	(1.47, 2.15)
Age	0.10	0.05	2.29	0.02	1.11	(1.01, 1.21)
Frequency of Facebook Use	0.01	0.02	0.16	0.87	1.01	(1.00, 1.05)
Frequency of Snapchat Use	0.07	0.02	2.94	<0.01	1.08	(1.02, 1.13)
Frequency of Instagram use	0.01	0.02	1.62	0.98	1.00	(0.95, 1.05)
Self & Others Consumption	0.22	0.07	3.27	0.01	1.24	(1.09, 1.41)
Memes & Viral Photos	0.02	0.05	0.41	0.68	1.02	(0.92, 0.13)
Status Updates: Other’s Drinking & Consequences	−0.03	0.10	−0.31	0.75	0.96	(0.79, 1.19)
Pictures: Others’ Drinking & Consequences	−0.03	0.10	−0.32	0.75	0.97	(0.80, 1.17)
Criterion Variable: Past Month Negative Consequences
Sex	−0.04	0.10	−0.38	0.70	0.96	(0.79, 1.16)
Age	0.07	0.04	1.52	0.13	1.07	(0.98, 1.16)
Drinks Per Week	0.04	0.01	6.72	<0.01	1.04	(1.03, 1.05)
Frequency of Facebook Use	−0.02	0.02	0.09	0.36	0.98	(0.94, 1.03)
Frequency of Snapchat Use	0.05	0.02	2.06	0.04	1.05	(1.00, 1.10)
Frequency of Instagram use	0.02	0.02	0.72	0.47	1.02	(0.97, 1.07)
Self & Others Consumption	0.13	0.06	2.05	0.04	1.14	(1.00, 1.28)
Memes & Viral Photos	0.01	0.05	0.13	0.89	1.00	(0.91, 1.12)
Status Updates: Other’s Drinking & Consequences	0.02	0.10	0.23	0.80	1.03	(0.84, 1.24)
Pictures: Others’ Drinking & Consequences	0.05	0.09	0.54	0.59	1.05	(0.79, 1.14)

Note. n = 274, Ratio = negative binomial incidence rate ratios. Sex coded 0 = female, 1 = male.

Peak Drinks.

The overall model for peak drinks per occasion was significant, X² (9) = 58.89, p < .001. Results examining peak drinks per occasion as the outcome displayed similar patterns to the aforementioned regression such that of all predictors in the model, only being older, male, using Snapchat more frequently, and the Self & Friend Consumption subscale were associated with number of drinks consumed on a peak occasion in the past month (see Table 4).

Alcohol-Related Negative Consequences.

Results indicated that contrary to previous models, age and sex were not associated with consequences. However, drinking more per week, using Snapchat more frequently, and the Self & Friend Consumption subscale were significantly associated with reporting more negative consequences in the past month and the overall model was significant, X² (9) = 77.31, p < .001 (see Table 4).

4.1. Discussion

In an adolescent and young sample (15–20 year olds) who regularly use SNSs and reported past month drinking, this study aimed to develop a measure focused on the likelihood of posting alcohol-related content on SNSs. Findings from the current study indicated four factors in the developed measure: Self and Friend Consumption, Memes and Viral Photos, Status Updates: Others’ Drinking and Consequences, and Pictures: Others’ Drinking and Consequences. When examining if the likelihood of posting different types of ARC on SNSs was associated with SNS use patterns, alcohol consumption, and consequences among a sample of adolescents and young adults, only the Self and Friend Consumption subscale was consistently significantly associated. Thus, in terms of alcohol-related content on SNSs, research should focus on ways to identify content about self or friends. Content regarding strangers, memes, or viral posts may not need to be targeted as ways to trigger alcohol interventions. Further research is warranted to replicate and confirm these findings.

The association between the Self and Friend Consumption subscale and alcohol consumption and problems is particularly interesting as it the only scale that was consistently associated with alcohol-related outcomes. Given that this study sample was all underage, posting about actual alcohol consumption of themselves or close friends might be associated with potential legal and familial consequences (i.e., getting in trouble with the law or with parents/guardians). As such, it is possible that individuals more likely to post this type of content are inherently more risk-taking than individuals who are less likely to share this content. In that way, being willing to post content that shows personal or peer consumption may be part of an underlying risk-taking personality or part of constellation of risk behaviors that may also include being more likely to drink alcohol in the first place. The relationship may be an artifact of the composition of the current study. The participants were under the legal age for alcohol consumption. Posting or sharing alcohol-related content concerning themselves or friends would most likely be self-generated (vs. posting of material created by others). Therefore, the act of sharing it on SNSs could be more risky, and risky behavior links to alcohol consumption (Cyders et al, 2009).

4.2. Clinical/Public Health Implications

Our study provides an important first step in the development of adolescent and young adult online risk profiles with respect to ARC. Our findings may have salient implications for prevention and intervention efforts, as well as for parents, caregivers and educators who are often faced with determining whether certain SNS activity by the youth in their care is a cause for concern. Our results emphasize that, whereas certain behaviors, such as posting photos or statuses of one’s own or friends’ drinking may, indeed, be cause for concern, other behaviors such as posting alcohol-related memes and viral photos may not be, at least as far as associated alcohol consumption and related problems are concerned.

In addition, our results might help specify which SNS users, based on their types of ARC they post, might benefit most from intervention. It has been suggested ARC that individuals post could be used to flag and send targeted prevention messages to those individuals (Moreno et al., 2016). As such, our results suggest that not all types of ARC may need to be flagged, but rather posts specifically related to self and friend consumption may warrant the provision of targeted prevention messages. Given that SNSs have the power to provide users with targeted advertisements based on their posting behavior and personal information, it is possible that this same technology could be harnessed to both identify individuals in need of intervention and deliver the material itself (Ramo et al., 2019).

4.3. Future Directions

The current study provides a base from which a number of important research questions can now be explored. First, it will be valuable for subsequent research to further identify the profiles of young individuals who are most likely to post differing types of ARC, in addition to the profiles of SNS users based on their differing ARC content. Regarding the latter, perhaps young people with a greater array of ARC posted to SNSs are at a greater risk of alcohol consumption and related consequences. Second, it was particularly interesting that in our study, the type of ARC participants were most likely to share – alcohol-related memes and viral photos – was not significantly associated with alcohol use or related consequences. It will be important for future research to replicate these results. Extending from this, it is possible that this type of ARC may not relate to the poster’s alcohol-related behavior, but could still affect the subjective norms and consumption patterns of others exposed to this content (i.e., the poster’s followers). Thus, a useful extension of our work will be to examine the extent to which the four identified types of ARC affect the consumers (followers) of other’s posts. Finally, it is possible that specific SNS platform characteristics, including increased privacy and temporary posts that delete after a short amount of time (Snapchat, Instagram stories), as well as the ability to like others’ posts, thereby demonstrating peer approval for alcohol-related behaviors (Boyle et al., 2018), may differentially affect posting of the four different types of ARCs. In a similar vein, we found that Snapchat, which possess the characteristics described above was the most common SNS for ARC in general, echoing past research (Boyle et al., 2017). Further, given the recent rise in TikTok use among adolescents and young adults as well as it’s unique features (e.g., public posts, video content, etc.), future research might seek to examine TikTok use for ARC and its relationship to the scales validated within this manuscript given that at the of data collection for the present study (2017 to 2018) TikTok was not widely used.

Limitations

This study is not without limitations. Because our sample had to meet several eligibility criteria at screening, our findings are not generalizable to all adolescents and young adults who reside in the United States. Moreover, the current study focused on likelihood of posting ARC rather than actual displays on SNSs. However, triggering an intervention based on likelihood of posting ARC is more feasible and cost effective than having to trigger alcohol interventions based on actual displays. Another limitation is related to the selection and testing of the items for inclusion in the likelihood of posting ARC scale. Given that the items were not fully pilot tested prior to inclusion in this study, it is possible that the measure may missing types of ARC relevant to this sample. Further, the present study did not examine exposure to ARC on SNSs, which may also have important associations with the likelihood of posting ARC as individuals who are exposed to more ARC may be more likely to post it themselves. In addition, participants endorsed the scale items at low levels. The low level might be due to the age of the participants. Since they are under the legal age for alcohol consumption, posting on SNSs or sharing ARC might have social constraints and repercussions. Future studies should consider examining the items in samples of individuals who post higher quantities of ARC or who are older. Due to the age of the participants, some of the potential participants needed parental consent. Therefore, it is possible that for these adolescents aged 15–17, asking their parents to be part of this type of alcohol and social media research led to a selection bias. Of note, it is also interesting that that subscale that address status updates for others’ drinking was not related to Facebook use. Given that the common terminology for the Facebook platform includes “status updates,” this is surprising. Therefore, future research might examine what participants consider a “status update” to provide further clarity to the measure. Finally, future studies should address the associations with Self and Friend Consumption ARC and alcohol outcomes longitudinally as the current study was cross-sectional and could not test temporality or causality.

4.4. Conclusion

The purpose of the current study was to develop a measure of likelihood of posting different types of ARC that can apply to a range of SNS platforms. Results indicate that the newly developed scale is both reliable and valid as a measure of ARC sharing behaviors. Thus, it is recommended that investigators aiming to better understand the associations between posting ARC on SNS and alcohol use and consequences utilize this scale in their research as it provides a more nuanced exploration of which specific types of ARC may be indicators of risk and which may not in a sample of adolescents and young adults.