Using Facebook to deliver a social norm intervention to reduce problem drinking at university

Authors

  • Brad Ridout,

    Corresponding author
    1. Faculty of Health Sciences, The University of Sydney, Sydney, Australia
    • Correspondence to Mr Brad Ridout, c/- Dr Andrew Campbell, Faculty of Health Sciences, The University of Sydney, Building C42, PO Box 170, Lidcombe, NSW 1825, Australia. Tel: 0061 2 9351 9762; Fax: 0061 2 9351 9540; E-mail: brad.ridout@sydney.edu.au

    Search for more papers by this author
  • Andrew Campbell

    1. Faculty of Health Sciences, The University of Sydney, Sydney, Australia
    Search for more papers by this author

  • Brad Ridout PhD Candidate, Andrew Campbell PhD, Senior Lecturer.

Abstract

Introduction and Aims

University students usually overestimate peer alcohol use, resulting in them ‘drinking up’ to perceived norms. Social norms theory suggests correcting these inflated perceptions can reduce alcohol consumption. Recent findings by the current authors show portraying oneself as ‘a drinker’ is considered by many students to be a socially desirable component of their Facebook identity, perpetuating an online culture that normalises binge drinking. However, social networking sites have yet to be utilised in social norms interventions.

Design and Methods

Actual and perceived descriptive and injunctive drinking norms were collected from 244 university students. Ninety-five students screened positive for hazardous drinking and were randomly allocated to a control group or intervention group that received social norms feedback via personalised Facebook private messages over three sessions.

Results

At 1 month post-intervention, the quantity and frequency of alcohol consumed by intervention group during the previous month had significantly reduced compared with baseline and controls. Reductions were maintained 3 months post-intervention. Intervention group perceived drinking norms were significantly more accurate post-intervention.

Discussion and Conclusions

This is the first study to test the feasibility of using Facebook to deliver social norms interventions. Correcting misperceptions of peer drinking norms resulted in clinically significant reductions in alcohol use. Facebook has many advantages over traditional social norms delivery, providing an innovative method for tackling problem drinking at university. These results have implications for the use of Facebook to deliver positive messages about safe alcohol use to students, which may counter the negative messages regarding alcohol normally seen on Facebook. [Ridout B, Campbell A. Using Facebook to deliver a social norm intervention to reduce problem drinking at university. Drug Alcohol Rev 2014;33:667–73]

Introduction

Research has consistently demonstrated that university students tend to overestimate both how much alcohol their peers drink (i.e. descriptive norms) and how much they approve of heavy drinking (i.e. injunctive norms). As normative beliefs are one of the strongest determinants of student alcohol use [1], students ‘drink up’ to these misperceived norms. An increasing body of research has shown that the social norms approach [2], which involves correcting inflated perceptions of drinking norms, can be effective in reducing problematic alcohol use in university populations [3,4].

Previous research has indicated that media exposure has an impact on normative beliefs regarding alcohol use among adolescents [5]. In addition, today's youth are confronted with an abundance of alcohol-related content on social networking sites (SNS), such as photos and comments from friends, as well as advertisements and alcohol-related games. Recent findings by the current authors reveal portraying oneself as ‘a drinker’ is considered by many students to be a socially desirable component of their Facebook identity, perpetuating an online culture that overrepresents and normalises binge drinking [6]. Other recent studies have confirmed that alcohol-related Facebook content mostly refers to alcohol in a positive context [7] and that these descriptive norms influence the perceived drinking norms of university students [8] and the willingness of younger adolescents to drink alcohol [9].

Social norms theory proposes that if true normative feedback is delivered effectively and is well received by the target sample, they will be less likely to conform to the false norm, mediating decreases in actual alcohol use. Analyses of previous social norms interventions have found that personalising normative feedback to individual students is important to their success [10–12].

As with other areas of preventative health, social norms interventions are increasingly being conducted online due to their cost-effectiveness in reaching large populations [13]. A Cochrane review of 22 controlled social norms studies found that feedback delivered by Web or computer was as effective as face-to-face feedback in terms of its effect on reducing alcohol misuse and impacted across a broader set of alcohol-related outcomes [12]. Online interventions since this review, however, have been inconsistent in terms of their effectiveness [14,15], which has been explained by methodological weaknesses, such as taking means of untransformed skewed data and not including a control group [16].

To date, online social norms interventions have used either university webmail or custom-built webpages to present feedback to students. However, a recent study into online communication usage among university students revealed that use of university webmail is rapidly declining, inverse to the use of Facebook [17]. Currently, the most popular SNS in the world, Facebook, is also the second most trafficked worldwide website overall behind Google [18], with over one billion active users [19]. It is estimated that 97% of undergraduate students now have active Facebook accounts [20] and use the site for an average of 1.5 h per day [6]. This presents researchers with a promising modality for engaging with students online. SNSs have been successfully used to deliver health information in other areas of e-Health promotion [21], so it is somewhat surprising that SNSs have yet to be used to deliver social norms interventions. The current study aimed to address this gap in the literature, as well as other methodological issues of previous studies, by testing the feasibility of delivering a social norms intervention using Facebook.

It is proposed that using SNSs, such as Facebook, to deliver social norms information has a number of advantages over using an external website or regular e-mail. First, it allows feedback to be provided to participants in the form of a private message that is identical to a message that would be received from a ‘Facebook friend’, and in the environment most often used to digest textual information regarding their peers’ drinking behaviour. It also has the additional benefit of remaining visible whenever the participant uses their Facebook inbox so that they may refer back to it. Another limitation of previous online social norms interventions is that there has been no way in which to confirm that participants have actually read their normative feedback. While many e-mail clients allow for ‘read receipts’ to be sent back to senders, these are not reliable as they require the receiver to give their explicit permission and do not work between all e-mail systems. Facebook can assist in addressing this issue as messages are automatically marked as ‘seen’ to indicate when they have been viewed by the recipient, without requiring their explicit permission.

The overall aim of the current study was to correct misperceptions among a sample of first-year university students regarding the normative drinking behaviour of their proximal peers (i.e. students in their current unit of study/class), with the intent to raise motivation for potentially hazardous drinkers in particular, to reduce their frequency and quantity of alcohol consumption. Research has shown support for the influence that proximal referents (i.e. friends, classmates) have on personal alcohol use [22,23] compared with distal others (i.e. ‘the average student’) [24], making them a salient reference group for social norms interventions.

It was predicted that of participants identified as drinking at potentially hazardous levels according to the Alcohol Use Disorders Identification Test (AUDIT) [25], those who received personalised social norms feedback would report less monthly alcohol consumption (quantity and frequency) at a 1 and 3 month follow-up compared with a screen-only control group.

Methods

Participants

Participants in the initial screening questionnaire were 244 first-year University of Sydney students (all recruited from the same unit of study) who took part in the study in partial fulfilment of course requirements. Students were given the option of completing an alternative task if they did not wish to participate (16 women and four men out of 264 students selected this option). Recruitment was via an e-mail circulated to students containing a link to the SurveyMonkey online screening survey. The mean age of the sample was 19.05 years [standard deviation (SD) = 1.78] and the majority were women (78%) and Caucasian (56%). The procedures of this study were approved by the Human Research Ethics Committee of the University of Sydney (Protocol no. 14429).

Measures and procedure: screening questionnaire

Participants completed an online initial screening questionnaire that included the AUDIT, the Graduated-Frequency Measure (GF) [26], and a range of social norms and demographic questions. The consumption-related questions contained in the AUDIT were used to determine actual descriptive norms for the sample (how often participants had a drink containing alcohol, how many drinks they have on a typical day when drinking and how often they have six or more drinks on one occasion). For perceptions of classmates' alcohol use, these three questions were reworded and presented as a pair with the corresponding question regarding own behaviour, as per the Social Norms Guidebook [27]. A pair of five-interval Likert-type response questions (strongly disapprove to strongly approve) asked participants for their own and perceived injunctive norms regarding heavy drinking: ‘How do you feel about students drinking heavily in public?’ and ‘How do you believe other students in your unit of study feel about students drinking heavily in public?’ At the conclusion of the screening questionnaire, participants gave informed consent to become ‘friends’ with the study on Facebook, so that all future correspondence regarding the study could be delivered via Facebook private message.

Measures and procedure: intervention

Baseline

Ninety-eight students from the 244 screened were identified by the AUDIT as reporting potentially hazardous alcohol use (scoring 8 or above) and were randomly allocated to either the intervention or control group using the random number function of Microsoft Excel ® (Microsoft, Redmond, WA, USA). Participants in the intervention group received social norms feedback the day after the 1 week screening questionnaire participation period ended in the form of a Facebook private message. Messages were generated using Microsoft Word® (Microsoft) macros and manually pasted into the Facebook private messaging system. They included statements comparing the participants' perceptions of classmates' use and approval of alcohol with actual descriptive and injunctive social norms calculated from their classmates' survey questionnaire responses [e.g. ‘You said that you have six or more standard drinks weekly and that you think a typical student in this unit of study has six or more standard drinks weekly. Actually, of the students in this unit who drink alcohol, most (84%) have six or more standard drinks ONCE A MONTH OR LESS']. A percentile rank of how the participant's alcohol consumption compared with other students in their unit of study was also included. As per World Health Organization (WHO) recommendations [28], the message stated the participant's AUDIT score and a brief explanation of associated health risks and information on reducing these risks.

While Facebook is able to confirm that feedback has been ‘seen’, the current study endeavoured to confirm that it was also understood by concluding the feedback message with a hyperlink to a brief online form asking participants to interpret and type in the figures they received regarding their own and their classmates' alcohol use and approval of heavy drinking. These forms were checked by the research team, and any error in participants' responses was addressed with immediate follow-up in the form of a second short Facebook message, informing them that there were errors in the reporting back of their results and asking them to complete the form again. Fifteen out of the 47 intervention participants had at least one error in their initial baseline feedback forms. This figure dropped to three participants for the 1 month follow-up feedback form.

One month follow-up

One month after completing the initial questionnaire, participants in the intervention and control groups received a Facebook private message containing a hyperlink to the 1 month follow-up questionnaire that asked them to complete the GF based on their previous month's alcohol consumption. Participants in the intervention group received a second set of social norms feedback via a Facebook private message the day after the 1 week period given to complete the 1 month follow-up questionnaire ended. This message stated how many standard drinks they had consumed in the previous month (according to their 1 month follow-up GF response) and their current percentile rank (according to initial sample descriptive norms). These figures were then compared with their own figures at baseline, stating whether the participant had increased or decreased their levels of alcohol consumption. The message also included appropriate information regarding the impact any increase or decrease had on their health risk based on the WHO manual ‘Brief Intervention for Hazardous and Harmful Drinking’ [28]. The feedback message again concluded with a hyperlink to a brief online form asking participants to type in the figures stated in their Facebook message to confirm that they had read and understood their feedback.

Three month follow-up

Three months after completing the initial questionnaire, participants in the intervention and control groups received a Facebook private message containing a hyperlink to the 3 month follow-up questionnaire, which consisted of the AUDIT and the corresponding reworded social norms questions regarding perceived descriptive norms, the GF, and the pair of injunctive norm questions regarding approval of students drinking heavily in public. Participants in both the intervention and control groups received social norms feedback via a Facebook private message the day after the 1 week period given to complete the 3 month follow-up questionnaire ended. This message contained comparisons of their most recent self-reported alcohol use with the descriptive norms of their classmates who drink, which were established by the initial screening questionnaire. The feedback also stated whether the participant had increased or decreased their levels of alcohol consumption since the initial screening questionnaire and relevant information from the WHO manual ‘Brief Intervention for Hazardous and Harmful Drinking’ [28]. Participants were not asked to confirm they had read this final feedback message.

Results

Prior to any social norms feedback being sent out, checks were carried out to ensure that the intervention and control groups did not differ on any reported demographic or alcohol variables of interest (all non-significant, see Table 1). Three participants failed to complete the follow-up surveys and were therefore excluded from analysis due to missing data. The 95 remaining participants were 80% women and ranged in age from 17–24 (M = 18.93, SD = 1.22). As participants were allocated to the intervention and control groups based on them screening positive for potentially hazardous drinking, data collected on the key outcome variable (GF) were substantially skewed in the positive direction. In order to allow for parametric analysis, a log10 transformation was carried out [29]. Descriptive statistics are based on original values.

Table 1. Demographic characteristics and alcohol use of the study groups at baseline
ScaleM
Intervention group (n = 47)Control group (n = 48)Test of independence
  1. AUDIT, Alcohol Use Disorders Identification Test; GF, graduated-frequency measure; M, mean; SD, standard deviation.
Gender, % female78.781.3χ2 = 0.095
Age, mean (SD), years18.96 (1.35)18.90 (1.10)t = 0.244
Living arrangement, %  χ2 = 0.478
Living at parents'/relative's87.287.5 
Living on campus in a college4.32.1 
Living in a sharehouse6.48.3 
Living with a partner2.12.1 
Living on your own0.00.0 
Ethnicity, %  χ2 = 0.338
Caucasian72.377.1 
Asian21.316.7 
Middle Eastern4.34.2 
Pacific Islander2.12.1 
Religion, %  χ2 = 3.518
Christian59.656.3 
Jewish4.36.3 
Hindu2.10.0 
Buddhist0.02.1 
Agnostic6.42.1 
No religion27.733.3 
AUDIT, mean (SD)12.60 (4.20)12.33 (4.30)t = 0.301
GF (quantity), mean (SD)39.99 (34.99)41.24 (47.96)t = −0.145
GF (frequency), mean (SD)8.75 (6.72)7.37 (7.37)t = 0.955

A repeated measures analysis of variance using orthogonal contrasts revealed a significant interaction between intervention group and follow-up time on alcohol consumption quantity, F(2,186) = 5.818, P < 0.01. A Helmert contrast comparing baseline quantity with an average of follow-ups confirmed that as predicted, the intervention group reduced their monthly drinking quantity at the follow-ups significantly more compared with the control group, F(1,93) = 10.446, P < 0.01 (see Figure 1).

Figure 1.

Change in quantity of drinks consumed over 3 month intervention.

A second repeated measures analysis of variance using orthogonal contrasts revealed there was also a significant interaction between intervention group and follow-up time on alcohol consumption frequency, F(2,186) = 6.281, P < 0.01. A Helmert contrast comparing baseline frequency to an average of follow-ups confirmed that as predicted, the intervention group reduced their monthly drinking frequency at the follow-ups significantly more compared with the control group, F(1,93) = 10.569, P < 0.01 (see Figure 2).

Figure 2.

Change in frequency of drinking days over 3 month intervention.

Analyses were carried out in order to test whether the accuracy of social norm perceptions by the intervention group improved following the receipt of personalised social norm feedback in comparison with the control group that did not receive this feedback. Difference scores were calculated between the Likert score [1–5] selected by each participant at baseline on each of the four social norms questions, and the corresponding Likert score representing the actual social norm for each question. The same calculation was carried out for Likert scores selected by participants at the 3 month follow-up, allowing for ‘change in accuracy’ scores to be calculated by subtracting the 3 month follow-up difference scores from the baseline difference scores.

Given that the social norms data are ordinal and therefore non-parametric, a Mann–Whitney U test of independent samples was performed. Results indicated that the intervention group improved their accuracy significantly more than the control group on three of the four social norms questions: ‘How many standard drinks do you think a typical student in this unit of study has when drinking?’ (U = 822.5, P < 0.05); ‘How often do you think a typical student has in this unit of study has six or more drinks on one occasion?’ (U = 564.5, P < 0.001); and ‘How do you think other students in this unit of study feel about students drinking heavily in public?’ (U = 734.5, P < 0.01). Mean ranks related to the overall sample of 95 are shown in Table 2.

Table 2. Mean ranks (n = 95) for change in accuracy of social norm perceptions (difference from actual norm on 1–5 Likert scale) for study groups at baseline and 3 month follow-up
Social norm questionIntervention (n = 47)Control (n = 48)Mann–Whitney U
  1. *P < 0.05; **P < 0.01; ***P < 0.001. Lower rank indicates better accuracy (i.e. smaller difference between perceived and actual social norms).
How often do you think a typical student in this unit of study has a drink containing alcohol?45.9550.011031.5
How many standard drinks do you think a typical student in this unit of study has when drinking?41.5054.36822.5*
How often do you think a typical student has in this unit of study has six or more drinks on one occasion?36.0159.74564.5***
How do you think other students in this unit of study feel about students drinking heavily in public?39.6356.20734.5**

Discussion

This is the first study to show that the messaging function of Facebook can be successfully utilised to correct misperceptions of peer alcohol use and approval, resulting in clinically significant reductions in alcohol consumption. These findings suggest that informing above-average drinkers of the actual drinking norms among their proximal peers (those in the same unit of study) and how their perceived norms and own drinking behaviour compare with these norms influences a change in drinking behaviour. This was achieved in the current study both in terms of how often they drank alcohol and how many standard drinks they consumed on each occasion. This lends support to the increasing body of literature that has found brief online social norms interventions effective in reducing problematic drinking behaviour [12,15] and other e-Health research that has highlighted the role of personalisation in the efficacy of SNS interventions [21]. Results also suggest that corrections to misperceived norms are lasting, as evidenced by the significant increase in accuracy of norm perceptions of the intervention group at the 3 month follow-up compared with the control group.

Facebook has many advantages over traditional social norms delivery, as just discussed, and given its ubiquitous nature, provides an engaging and innovative method for tackling the challenge of university drinking. The exponential uptake of smartphone technology means that most students now have access to Facebook on their mobile phones, with Facebook use on mobile devices presently exceeding that on computers [30]. Given that many young people use Facebook while out socialising, there may be potential for students to be reminded of their personalised social norms feedback while actually in a situation where they are drinking. Future research could explore the feasibility of automatic Facebook notifications to encourage this kind of ongoing engagement with social norms feedback.

Independent of the use of Facebook, this study has also successfully addressed the limitations of many previous social norms interventions by including a control group, transforming skewed data before taking measures of central tendency, not conflating the social norms effect by including other intervention strategies (although some brief information was required in accordance with WHO recommendations) and confirming participants actually read and understood their personalised feedback. To the authors' knowledge, this is the first study to address the latter concern, and while it may not always be feasible to require all participants in an intervention to report back understanding of their personalised feedback, it does raise the question of whether more reliable changes in perceived norms could be achieved by way of guaranteed engagement with normative feedback, no matter what the intervention delivery method. It should be noted that the response rate was uniquely high for this kind of study. While it is possible that the use of Facebook messaging for sending links to follow-up surveys facilitated this, it is more likely due to the fact that students did not receive course credit for their participation unless all follow-up surveys were completed. Further research is needed to determine the role Facebook played in the current results, perhaps by comparing the efficacy of social norms delivery by Facebook with another modality (such as university e-mail).

In considering the strengths of the current study, there are limitations that must be acknowledged. First, as the study focused on misperceptions of drinking behaviour of proximal peers (people in their current unit of study that they are more likely to know, as opposed to ‘the average student’), the ‘actual’ social norms figures required for the feedback messages could not be calculated until all baseline responses had been collected. Participants were given 1 week to complete the online screening questionnaire, resulting in up to a 1 week gap between completing the questionnaire and receiving personalised feedback. The use of proximal peers from a single class also meant that the sample was smaller in size compared with previous social norms interventions [12] and contained significantly more women than men (due to the demographics of the unit of study surveyed), so results cannot yet be generalised to both genders. Future studies are now needed to validate the current findings with larger and more diverse samples.

While previous studies have focused on how SNS content can perpetuate a binge drinking culture [6,7,31,32], these results have implications for the use of SNS to promote positive messages about safe alcohol use, which may counter the negative messages regarding alcohol normally seen on sites like Facebook. With Facebook becoming an everyday part of the university experience where social and educational functions collide, there is a great opportunity for further research into how the social aspects of this dynamic and highly trafficked SNS (which were not utilised in this study) might be used to effect social change regarding binge drinking.

In conclusion, the value of this study lies in the innovative use of SNS in a social norms intervention and evidence for the efficacy of the messaging function of Facebook in changing normative beliefs and reducing alcohol consumption in at-risk students. Given the power of peer influence and the preventable public health problems that binge drinking causes, correcting misperceptions regarding the prevalence and social approval of binge drinking behaviour using SNS is an inexpensive and effective strategy that could potentially bring widespread benefit to university populations.

Acknowledgements

The researchers would like to thank the study participants, Dr Rob Heard for his assistance with statistical analysis and Dr Melanie Nguyen for her assistance in recruitment. Mr Brad Ridout would like to thank the donors of the DBH Scholarship that funds his PhD candidature.

Ancillary