Internet use and adolescent drinking : Does it matter what young people do online?

(1)

Contents lists available atScienceDirect

Drug and Alcohol Dependence

journal homepage:www.elsevier.com/locate/drugalcdep

Full length article

Internet use and adolescent drinking: Does it matter what young people do

online?

Robert Svensson

a,

*

, Björn Johnson

b

a_{Department of Criminology, Malmö University, SE-205 06 Malmö, Sweden} b_{Department of Social Work, Malmö University, SE-205 06, Malmö, Sweden}

A R T I C L E I N F O

Keywords: Lifetime alcohol use Drunkenness Internet use Peer inﬂuence Unstructured activities

A B S T R A C T

Background: In this study we examine whether the association between internet use and drinking could be different for different types of internet activities among adolescents. We also adjust for a number of theoretically relevant factors such as peer influence, unstructured activities, impulsivity and parental monitoring. Method: The data are drawn from four cross-sectional surveys from the years 2016–2019 in eight municipalities in southern Sweden. The sample consist of 3733 adolescents in year 9 of compulsory education, aged 14–15. Results: The results show that there is an association between internet activities and drinking and that there are differences depending on what young people do online. Self-presentation and online sociality are both positively associated with drinking, whereas news consumption and playing games are negatively associated with drinking. The results also show that the association between the different internet activities and drinking becomes weaker when adjusting for the control variables.

Conclusion: This study suggests that more research is needed to examine the correlations between diﬀerent forms of internet activities and drinking among adolescents in more detail.

1. Introduction

Alcohol use among young people is a risky behavior that is asso-ciated with a number of negative consequences, including an increased risk for accidents, health-related problems, involvement in crime and other behaviors that breach societal norms, etc. (Room et al., 2005; Viner, 2005; Viner and Taylor, 2007; Babor, 2010; Emmers et al., 2015). The risk of developing alcohol dependence later in life is also greater for individuals with an early debut in alcohol use (McCambridge et al., 2011;Emmers et al., 2015).

Existing research in this area has provided knowledge on the most central risk factors for alcohol use among young people, such as peer inﬂuences, low parental monitoring, impulsivity etc. (e.g.,Curcio et al., 2013; Jones et al., 2014;Borsari and Carey, 2001; Cruz et al., 2012; Osgood et al., 1996;Svensson, 2004;Beck et al., 2004). Another factor that has also been raised as a central issue in alcohol research in recent years is that young people are increasingly spending their leisure time online– which has in turn been emphasized as a partial explanation for the observed decline in alcohol use among adolescents (Pennay et al., 2015;Pape et al., 2018).

An increasing number of researchers have come to direct an interest at the question of whether there is an association between internet use

and adolescent drinking (e.g., Moreno and Whitehill, 2014; Moreno et al., 2016,2018). Several cross-sectional studies have found a weak but positive association between internet use and alcohol use among adolescents (e.g.,Epstein, 2011;Coëffec et al., 2015), indicating that the more time young people spend online, the more they drink. Most of these studies are based on cross-sectional designs (Brunborg et al., 2017;Mu et al., 2015;Epstein, 2011;Sampasa-Kanyinga and Chaput, 2016), but studies based on longitudinal designs have also produced evidence of a positive association. Chiao et al. (2014), for example, found a significant association between internet use and heavy drinking but internet use was not associated with light drinking (1–2 times a month). In another longitudinal study, Brunborg and Burdzovic Andreas (2019)found that an increase in time spent on social media was moderately associated with an increase in episodic heavy drinking. Most of the studies mentioned above have used an overall measure of internet use, measuring for example how many hours a week adoles-cents spend online doing different things such as chatting, gaming, shopping, searching for information etc. (e.g.,Mu et al., 2015;Brunborg et al., 2017;Brunborg and Burdzovic Andreas, 2019).

Only a few studies have examined whether diﬀerent types of in-ternet activities are diﬀerently associated with adolescent alcohol use (e.g., Brunborg et al., 2017). In a cross-sectional study among 270

https://doi.org/10.1016/j.drugalcdep.2020.108138

Received 23 March 2020; Received in revised form 16 June 2020; Accepted 18 June 2020

⁎_{Corresponding author at: Department of Criminology, Malmö University, SE-205 06 Malmö, Sweden.}

E-mail addresses:robert.svensson@mau.se(R. Svensson),bjorn.johnson@mau.se(B. Johnson).

Available online 24 June 2020

(2)

adolescents aged 13–17, Epstein (2011)found that those who drink alcohol spend more time on social media and shop more online. No differences were found in computer use for schoolwork and gaming between drinkers and non-drinkers. In a Swedish cross-sectional study among 7089 adolescents aged 15–16,Larm et al. (2019)found no as-sociation between total time spent on computers per day and drinking. However, they found social media use/chatting to be associated with an increased probability of drinking, whereas computer gaming, especially on weekends, was associated with a decrease in probability of drinking. On the other hand,Coëffec et al. (2015)found alcohol use to be more common among video gamers than among non-gamers. Since the re-sults of these studies differ with respect to the relationship between adolescents’ online activities and whether or not they drink alcohol, it is important to improve the knowledge about the activities young people engage in online and how these are related to drinking behavior.

Activities on the internet can be categorised in diﬀerent ways. One group of activities can be categorised as“socially oriented”, i.e. chatting or using social media (Facebook, Instagram, Snapchat etc.). These ac-tivities give an indication about how social an individual is on the in-ternet (Epstein, 2011; Larm et al., 2019; Brunborg and Burdzovic Andreas, 2019;Mu et al., 2015). We would argue that it is relevant to group the socially oriented activities into two categories, i.e. interac-tions with others (e.g., chatting) and self-presentation (e.g., posting information about yourself on Instagram). Further, other categories of activities on the internet, which are less socially oriented, could be grouped into playing games (Epstein, 2011;Coëﬀec et al., 2015) and news consumption/searching for information (Mu et al., 2015;Epstein, 2011).

How can it be that the more time young people spend on the in-ternet, the more they drink? One explanation that has often been dis-cussed is peer influence. This explanation is based on social learning theory, with the internet being viewed as a forum where young people come into contact with peers who have a positive attitude towards al-cohol, and that this in turn increases the likelihood that adolescents themselves develop such attitudes (Larm et al., 2019;Chiao et al., 2014; Mu et al., 2015;Huang et al., 2014. See alsoSutherland, 1947;Bandura, 1971;Akers, 1998). A second explanation for the association between internet use and drinking may be that intensive socializing online is also associated with more intensive socializing offline (Brunborg et al., 2017;Brunborg and Burdzovic Andreas, 2019). If this offline socializing occurs in the form of unstructured activities, i.e. activities characterized by low levels of social control, in the absence of the constraining in-fluence of adults, the likelihood of alcohol use increases. (e.g.,Osgood et al., 1996).

Another potential explanation for the association between internet use and drinking is that it might be spurious. Impulsivity and sensation seeking have been suggested as potential confounders that may be po-sitively associated with both adolescent drinking and internet use (Brunborg et al., 2017). Low parental monitoring may also be a potential explanation for this relationship (Mu et al., 2015). This might indicate that in homes with low levels of parental control, there is a somewhat higher likelihood that young people will be more active online.

It has been found in both cross-sectional and longitudinal studies that internet use is positively associated with drinking after adjusting for peer inﬂuences and other potential confounders, such as sensation-seeking, impulsivity and leisure activities (Brunborg et al., 2017; Brunborg and Burdzovic Andreas, 2019;Larm et al., 2019;Chiao et al., 2014). It is worth mentioning, however, that most of these studies have employed an overall scale of internet activity.

To sum up, a number of studies have shown that there is an asso-ciation between internet activity and alcohol use (e.g.,Mu et al., 2015) but very few studies have examined whether there are diﬀerences in the nature of this association that are dependent on diﬀerent types of in-ternet activities (e.g.,Epstein, 2011). Some studies have examined the association (mainly focusing on an overall scale of internet activity), adjusting for relevant factors (e.g.,Brunborg et al., 2017), but not for

different types of drinking, nor for different types of online activities. Against this background, the aim of the present cross-sectional study is to examine whether the association between internet activities and drinking– lifetime alcohol use, drunkenness during the past year and drunkenness during the past month– are different for different types of internet activities among adolescents. The study also extends the previous research by examining whether these associations remain significant after adjusting for peer influence, unstructured activities, impulsivity and parental monitoring.

In the study we will test the following hypotheses:

H1. There is a positive association between overall internet activity and alcohol use and drunkenness.

H2. Different internet activities are differently associated with alcohol use and drunkenness. More specifically: (a) more socially oriented internet activities are positively associated with alcohol use and drunkenness, (b) less socially oriented internet activities are negatively associated with alcohol use and drunkenness.

H3. The strength of the associations between internet activities and alcohol use and drunkenness becomes weaker when adjusting for peer inﬂuences (having friends who have been drunk), parental monitoring, impulsivity and unstructured activities.

2. Method

2.1. Participants

The data employed in the study are drawn from the Öckerö project, an evaluation of an alcohol and drug prevention method. The project includes an annual, anonymously completed, cross-sectional online self-report survey, which is conducted at 17 secondary schools in eight small municipalities in the county of Skåne. Skåne is Sweden’s most southerly county with a population of approximately 1.4 million. The eight municipalities have between 13,000 and 19,000 inhabitants, a total of 125,000 altogether. The survey has been conducted in all classes in years 7–9 (i.e. 12–15 years of age), the ﬁnal three years of compulsory education. The survey was conducted at the beginning of the autumn term in each of four successive years, 2016−2019.

In this study, we employ data on adolescents in year nine (aged 14–15) from the years 2016–2019. We decided to use data only for those aged 15 (year 9), since very few of the younger groups have re-ported drunkenness. The diﬀerent surveys included a total of 1018 adolescents in 2016, 1028 in 2017, 1050 in 2018 and 1018 in 2019. The four subsamples combined produce a total sample of 4114 adolescents. The non-response rate for the sample is 13.3 %. Following listwise deletion of missing values, the analyses below are based on 3733 re-spondents (50.0 % girls). The non-response rate is fairly evenly dis-tributed across the included variables. We considered employing an imputation procedure, but since several of the variables included are based on single items, we decided against this. It is also important to note that by comparison with Sweden as a whole, alcohol use among young people is somewhat more common in Skåne. The municipalities included in the study are small and for the most part comprised of rural areas. For this reason, the material cannot be viewed as representative for Sweden as a whole. The research design and study procedures were approved by The Regional Ethical Review Board in Southern Sweden.

2.2. Measures

2.2.1. Drinking

Three binary (prevalence) measures of drinking will be used in this study. Lifetime alcohol use is measured using the following item: Have you ever drunk alcohol (by alcohol we mean medium-strength or strong beer, cider, alcopop, wine or spirits) (response options: no or yes, 1 time = 0 / yes, many times = 1). Drunkenness past year is measured using

(3)

the following item: How many times during the past 12 months have you drunk alcohol so that you have felt intoxicated? (response options: never = 0 / 1 time or more = 1). Drunkenness past month is measured using the following item: How many times during the past month have you drunk alcohol so that you have felt intoxicated? (response options: never = 0 / 1 time or more = 1). The proportion of adolescents in this study who reported having drunk alcohol is 36.6 %; 30.9 % had been intoxicated during the past year and 17.3 % had been intoxicated during the past month.

2.2.2. Internet use

Four items were used to identify diﬀerent dimensions of internet activities. The questions that were posed in the survey are as follows:

How often do you use a computer, mobile phone or tablet to do one of the following activities?

•

Post information about myself on Facebook, Instagram, Shapchat or other social media– referred to as self-presentation

•

Stay in contact with and stay informed about my friends via Facebook, Instagram or similar– referred to as: online sociality

•

Read daily newspapers or check the news – referred to as news consumption

•

Play games– referred to as playing games

Response options: never / about once a month / about once a week and / several times a week / every day. The items will be treated as continuous variables in the following analyses. Finally, an overall index including the four internet activity items will also be used. Using this overall index of internet activity will enable us to compare our results with previous studies that employed an overall index of internet ac-tivity. High values indicate high levels of spending time online.

The items we use in this study measure diﬀerent activities that young people engage in online. One limitation of these items is that the activities they refer to can change over time and some of them may be outdated. Another limitation relates to the response options. We use an ordinal scale (that will be treated as a continuous variable), which does not focus on quantity in the way many previous studies have done, i.e. number of hours per day (e.g.,Larm et al., 2019).

2.2.3. Control variables

Peer inﬂuence is measured by a single item: Has one or more of your best friends to your knowledge been drunk during the past month? (response options: no, no friend/yes, 1 friend/yes, 2−3 friends/yes, more than 3 friends). This variable is dummy coded with: no, no friend as the reference category. 53.4 % of the respondents had no friend who had been drunk, 12.1 % had one friend who had been drunk and 22.8 % had more than 3 friends who had been drunk.

Impulsivity is measured using an index comprised of three items: (1) It often happens that I do things without thinkingﬁrst, (2) I sometimes take risks just because it’s exciting, (3) I sometimes think it’s exciting to do things that might get me into trouble (response options: not true at all/not very true/quite true/completely true). Alpha for the scale is 0.79. High scores indicate a high level of impulsivity.

Parental monitoring is measured using an index comprised ofﬁve items: (1) Do you have speciﬁc times when you have to be home in the evenings? (2) Do you have to ask your parents for permission to go out in the evening? (3) Do you have to contact your parents if you are not home by a certain time? (4) If you are going out in the evening, do you have to tell your parents who you are going to meet? (5) If you are going out in the evening, do you have to tell your parents what you are going to do? (response options: no, never/rarely/sometimes/often/yes, always). Alpha for the scale is .81. High scores indicate high levels of parental monitoring.

Unstructured activities is measured using an index comprised of two items: (1) How many evenings per week do you usually travel to the town center or hang out there, either alone or with friends? (2) How

many days per week do you usually just hang around or drift about without doing anything in particular, either alone or with friends? (response options: 0/1/2/3/4/5/6/7). The two items are strongly cor-related r = .57. High scores indicate high levels of spending time in unstructured activities.

Sex is coded as zero for girls and one for boys.

For a description of alcohol use, internet activities and control variables seeTable 1. Online sociality is the most common activity among young people, followed by playing games and self-presentation. The least common internet activity among the respondents is reading newspapers or checking the news.

2.3. Statistical analysis

To examine our hypotheses, we will conduct both bivariate corre-lations and multivariate regression analysis. Firstly, we examine bi-variate (Pearson r) correlations between thefive different scales of in-ternet activities and the three different measures of alcohol use. Secondly, we run a number of multivariate linear probability models (LPM). Since we know of the shortcomings of logistic regression when

Table 1

Descriptive statistics (N = 3733).

Mean Std.dev. Min Max

Scales

Self-presentation 2.74 1.38 1 5 Online sociability 4.24 1.18 1 5 News consumption 2.29 1.17 1 5

Playing games 3.41 1.38 1 5

Overall scale of internet activity 12.68 2.70 4 20 Unstructured activities 5.59 3.45 2 16

Impulsivity 6.23 2.40 2 12

Parental monitoring 18.64 4.81 5 25 Per cent

Lifetime alcohol use

No 63.4

Yes 36.6

Drunkenness past year

No 69.1

Yes 30.9

Drunkenness past month

No 82.7

Yes 17.3

Peer inﬂuence (friends been drunk)

No friend 53.4

Yes, 1 friend 12.1 Yes, 2−3 friends 11.8 Yes, more than 3 friends 22.8 Self-presentation

Never 21.5

About once a month 30.2 About once a week 19.0 Several times a week 11.7

Every day 17.7

Online sociability

Never 5.8

Every day 61.7

News consumption

Never 30.9

Every day 5.6

Playing games

Never 12.5

(4)

comparing coeﬃcients between models (e.g.,Mood, 2010;Breen et al., 2018; Norton and Dowd, 2018), we decided to estimate our models using the LPM method. The analyses were also estimated using the more traditional logistic regression method focusing on estimates of average marginal eﬀects (AME) and the results followed a pattern si-milar to that obtained using LPM.

In the regression models we use our three dummy coded alcohol measures– lifetime alcohol use, drunkenness past year and drunkenness past month– as dependent variables. In the regression models we start by including the four different dimensions of internet use. As mentioned earlier the different dimensions of internet use will be treated as con-tinuous variables since the pattern is linear in relation to alcohol use. All regressions were also estimated using the four dimensions of in-ternet use as dummy variables and the same results were found (ana-lyses not presented in the article). In the second model we then include our control variables: peer influence, parental monitoring, impulsivity, unstructured activities and sex. According to our hypotheses, we expect tofind that different online activities are differently related to alcohol use and that the association with the different internet activities will decrease when all variables are included in the model. In all the re-gression models we adjusted for the year in which the study was con-ducted (year 2016 is reference category). Multicollinearity was not a problem in our analysis, which produced a highest Variance Inflation Factor Score of 1.57, well below critical levels.

All analyses were conducted in Stata/SE version 13.1.

3. Results

Table 2presents bivariate correlations between the different mea-sures of internet activities and drinking. The overall measure of internet use is weakly positively correlated with lifetime alcohol use (r = .11, p < .001), drunkenness past year (r = .10, p < .001) and drunkenness past month (r = .09, p < .001). However, the different measures of internet activities are differentially correlated with the measures of alcohol use. Both self-presentation and online sociality are positively correlated with all of the three measures of drinking. On the other hand, news consumption and playing games are both negatively correlated with lifetime alcohol use and the two measures of drunkenness. Finally, the internet activities measure that is most strongly correlated with lifetime alcohol use and drunkenness is self-presentation.

Results from the multivariate linear probability model are presented inTable 3. In thefirst model, both self-presentation and online sociality are positively associated with lifetime alcohol use, drunkenness in the past year and drunkenness in the past month. News consumption, i.e. reading newspapers or checking the news on the internet, is negatively associated with the three measures of alcohol use. Playing games is also negatively associated with the three measures of alcohol use. According to the standardized regression coefficient, self-presentation seems to be most strongly associated with the different alcohol use outcomes.

In the second model, peer inﬂuence (number of friends that have been drunk), impulsivity and unstructured activities are all positively associated with all three measures of drinking. Parental monitoring is

negatively associated with drinking. According to the beta-value, peer influence seems to be more strongly associated with drinking than the other factors. While the results show that the associations between self-presentation and alcohol use (all measures) become weaker when ad-justing for the control variables, the coefficients remain significant. Although the association between online sociality and the three dif-ferent measures of alcohol use is significant for two of the three mea-sures of drinking, the regression coefficients become much weaker when adjusting for the control variables. Although the associations between news consumption and playing games and alcohol use de-crease, some of the coefficients remain significant in Model 2. This indicates that adjusting for the control variables decreases the asso-ciation between the different internet activities and alcohol use. Sex is not significantly associated with alcohol use in any of the three models. 4. Discussion

We have examined whether the association between internet use and drinking is different for different types of internet activities (self-presentation, online sociality, news consumption and playing games) and different types of drinking behaviors (lifetime alcohol use and drunkenness) among adolescents. We have also adjusted for a number of theoretically relevant factors such as peer influence, unstructured activities, impulsivity and parental monitoring. As far as we know, no previous study has included these factors in the same analysis. Three hypotheses have been examined.

Ourﬁrst hypothesis, i.e. that overall internet use is positively as-sociated with drinking, was supported by our results. We found a weak positive association between spending time online and the three drinking measures (lifetime alcohol use, drunkenness past year and drunkenness past month). That there is a rather weak association be-tween an overall index of internet activity and drinking has previously been reported from both cross-sectional and longitudinal studies (e.g., Brunborg and Burdzovic Andreas, 2019;Coëﬀec et al., 2015).

Our second hypothesis, i.e. that different online activities are dif-ferentially associated with lifetime alcohol use and drunkenness, was also supported by our results. As expected, we found that socially or-iented internet activities, i.e. self-presentation and online sociality are positively associated with both lifetime alcohol use and the two mea-sures of drunkenness. This indicates that a greater amount of time spent socializing with friends on the internet is associated with a greater probability of lifetime alcohol use and drunkenness in the past year and the past month. We also found that less socially oriented internet ac-tivities, i.e. news consumption and playing games, are negatively as-sociated with both lifetime alcohol use and drunkenness. These results might, for example, indicate that the more time young people spend playing games, the less time they have to socialize with friends offline and the less opportunity they have to drink alcohol. Similarfindings have also been noted in a small number of previous studies (Epstein, 2011;Larm et al., 2019). All in all, our results provide support for the importance of studying internet activities on the basis of different measures and not as a single, collapsed measure that includes a number of different activities, which may lead to associations being masked.

In line with our third hypothesis, we adjusted for peer influence, parental monitoring, impulsivity and unstructured activities. As ex-pected, and in line with previous research (e.g.,Curcio et al., 2013; Jones et al., 2014;Borsari and Carey, 2001;Brunborg et al., 2017), we found all four factors to be strongly associated with lifetime alcohol use and drunkenness. Of the four factors, peer influence is the one that is most strongly associated with lifetime alcohol use and drunkenness. The associations between all of the different internet activities and lifetime alcohol use and drunkenness become weaker when adjusting for peer influence, impulsivity, parental monitoring and unstructured activities, which clearly supports our third hypothesis. Previous studies have also found a weakening of the association between internet use (using an overall measure) and alcohol use (Brunborg et al., 2017;

Table 2

Bivariate correlations (Pearson r) between internet activities and alcohol use (N = 3733). Lifetime alcohol Use Drunkenness past year Drunkenness past month

Overall scale of internet activity .11*** .10*** .09*** Self-presentation .26*** .26*** .23*** Online sociability .18*** .17*** .15*** News consumption −.14*** −.12*** −.12*** Playing games −.09*** −.11*** −.09*** *p < .05. **p < .01. ***p < .001.

(5)

Brunborg and Burdzovic Andreas, 2019). This study, however, shows the same pattern for diﬀerent internet activities and for diﬀerent types of alcohol use.

Although we are unable to draw conclusions about causal re-lationships, we would argue on the basis of these results that there are indications that the four factors that we have controlled for may help us understand why there is an association between different forms of in-ternet activities and lifetime alcohol use and drunkenness. We would argue that self-presentation and online sociality are measures of something else, e.g. how social a person is in general. On this basis, high levels of sociality may also constitute a potential explanation for the relationship between internet activity and drinking (Brunborg et al., 2017). This would mean that intensive socializing online is also asso-ciated with more intensive socializing offline. If this offline socializing occurs in the form of unstructured activities, i.e. activities characterized by low levels of social control in the absence of the constraining in-fluence of adults, the likelihood of alcohol use increases (e.g.,Osgood et al., 1996).

The current study has some important strengths in relation to pre-vious research. Firstly, the study has employed more elaborate mea-sures of different online activities than most previous studies. Secondly, many previous studies have not examined both lifetime alcohol use and drunkenness. We have used three different prevalence measures of al-cohol use. Future research should go further and also examine the frequency of alcohol use in relation to different internet activities. Finally, we have also examined the relationship between different in-ternet activities and daily routines among young people. The role of key theoretical factors– peer influence, impulsivity, parental monitoring, and unstructured activities– has been mentioned in previous studies on internet use and adolescent drinking, but not studied empirically.

However, this study also has a number of limitations that need to be addressed. Firstly, the study is based on cross-sectional data. We are therefore unable to deduce the temporal relationship between internet use and alcohol use. It would be useful in the future to test the causal order between different online activities and different forms of ado-lescent alcohol use and to include other potential mechanisms that can help us understand the relationship using longitudinal data. Secondly, we do not control for the potential effects of being targeted by online advertisements or other alcohol-positive communications online, which have been mentioned as potential explanatory factors in the relation-ship between internet use and adolescent alcohol use (Jernigan and Rushman, 2014). Finally, although we argue that our measures of in-ternet activities have some advantages in comparison with previous research, there is also one important limitation that needs to be

addressed. We have employed a relatively unsophisticated measure of time spent online, and have measured days rather than hours. This being said, the results nonetheless provide an indication of important questions that may need to be addressed in future studies that employ more detailed measures focused on hour-based data (i.e. a more con-crete measure of time spent online).

Contributors

Design of the study: RS and BJ. Conducted statistical analyses: RS and BJ. Wrote theﬁrst draft of the manuscript: RS. All authors have approved theﬁnal version of the manuscript.

Role of funding source

This study was supported by grants from the Public Health Agency of Sweden (02916-2015) and from the County Administrative Board of Skåne.

Declaration of Competing Interest

No conﬂict declared. Acknowledgments

None.

References

Akers, R.L., 1998. Social Learning and Social Structure: A General Theory of Crime and Deviance. Northeastern University Press, Boston.

Babor, T., 2010. Alcohol: No Ordinary Commodity. Oxford University Press.

Bandura, A., 1971. Social Learning Theory. General Learning Press, New York.

Beck, K.H., Boyle, J.R., Boekeloo, B.O., 2004. Parental monitoring and adolescent drinking: results of a 12-month follow-up. Am. J. Health Behav. 28, 272–279.

Borsari, B., Carey, K.B., 2001. Peer inﬂuences on college drinking: a review of the re-search. J. Subst. Abuse 13, 391–424.

Breen, R., Karlson, K.B., Holm, A., 2018. Interpreting and understanding logits, probits, and other nonlinear probability models. Annu. Rev. Sociol. 44, 39–54.

Brunborg, G.S., Burdzovic Andreas, J., 2019. Increase in time spent on social media is associated with modest increase in depression, conduct problems, and episodic heavy drinking. J. Adolesc. 74, 201–209.

Brunborg, G.S., Burdzovic Andreas, J., Kvaavik, E., 2017. Social media use and episodic heavy drinking among adolescents. Psychol. Rep. 120, 475–490.

Chiao, C., Yi, C.-C., Ksobiech, K., 2014. Adolescent internet use and its relationship to cigarette smoking and alcohol use: a prospective cohort study. Addict. Behav. 39, 7–12.

Coëﬀec, A., Romo, L., Cheze, N., Riazuelo, H., Plantey, S., Kotbagi, G., Kern, L., 2015.

Table 3

Relationship between alcohol use and internet activity (N = 3733). Linear Probability Model.

Lifetime alcohol use Drunkenness past year Drunkenness past month

Model 1 Model 2 Model 1 Model 2 Model 1 Model 2

b Beta p b Beta p b Beta p b Beta p b Beta p b Beta p

Self-presentation .08 .23 < .001 .03 .07 < .001 .07 .22 < .001 .02 .07 < .001 .06 .20 < .001 .02 .06 < .001 Online sociability .04 .10 < .001 .01 .04 .007 .04 .09 < .001 .01 .03 .027 .02 .08 < .001 .01 .02 .094 News consumption −.05 −.13 < .001 −.02 −.05 < .001 −.04 −.11 < .001 −.01 −.03 .044 −.04 −.11 < .001 −.01 −.04 .003 Playing games −.02 −.06 < .001 −.01 −.03 .071 −.03 −.08 < .001 −.01 −.03 .016 −.02 −.06 < .001 −.01 −.03 .073 Unstructured activities .01 .10 < .001 .02 .14 < .001 .01 .13 < .001 Impulsivity .04 .22 < .001 .03 .17 < .001 .02 .14 < .001 Parental monitoring −.01 −.09 < .001 −.01 −.08 < .001 −.00 −.06 < .001 Peer inﬂuence (ref: no friend drink)

Yes, 1 friend .16 .11 < .001 .14 .10 < .001 .08 .07 < .001 Yes, 2−3 friends .24 .16 < .001 .26 .18 < .001 .11 .09 < .001 Yes, more than 3 friends .43 .37 < .001 .45 .41 < .001 .36 .40 < .001

Sex (1=Boys) −.02 −.02 .151 −.02 −.03 .087 −.01 −.01 .507

R2 _.10 _.38 _.10 _.39 _.08 _.33

Note: The p-values are calculated using robust standard errors. All regression models are including dummy variables for year of study (2016 as reference category). b = unstandardized regression coeﬃcient, Beta = standardized regression coeﬃcient.

(6)

Early substance consumption and problematic use of video games in adolescence. Front Psychology 6, 501.

Cruz, J.E., Emery, R.E., Turkheimer, E., 2012. Peer network drinking predicts increased alcohol use from adolescence to early adulthood after controlling for genetic and shared environmental selection. Dev. Psychol. 48, 1390–1402.

Curcio, A.L., Mak, A.S., George, A.M., 2013. Do adolescent delinquency and problem drinking share psychosocial risk factors? A literature review. Addict. Behav. 38, 2003–2013.

Emmers, E., Bekkering, G.E., Hannes, K., 2015. Prevention of alcohol and drug misuse in adolescents: an overview of systematic reviews. Nord. Stud. Alcohol Drugs 32, 183–198.

Epstein, J.A., 2011. Adolescent computer use and alcohol use: what are the role of quantity and content of computer use? Addict. Behav. 36, 520–522.

Huang, G.C., Unger, J.B., Soto, D., Fujimoto, K., Pentz, M.A., Jordan-Marsh, M., Valentine, T.W., 2014. Peer inﬂuences: the impact of online and oﬄine friendship networks on adolescent smoking and alcohol use. J. Adolesc. Health 54, 508–514.

Jernigan, D.H., Rushman, A.E., 2014. Measuring youth exposure to alcohol marketing on social networking sites: challenges and prospects. J. Public Health Policy 30, 196–204.

Jones, K.A., Chryssanthakis, A., Groom, M.J., 2014. Impulsivity and drinking motives predict problem behaviours relating to alcohol use in University students. Addict. Behav. 39, 289–296.

Larm, P., Raninen, J., Åslund, C., Svensson, J., Nilsson, K.W., 2019. The increased trend of non-drinking alcohol among adolescents: what role do internet activities have? Eur. J. Public Health 29, 27–32.

McCambridge, J., McAlaney, J., Rowe, R., 2011. Adult consequences of late adolescent alcohol consumption: a systematic review of cohort studies. PLoS Med. 8, 1–13.

Mood, C., 2010. Logistic regression: why we cannot do what we think we can do, and

what we can do about it. Eur. Sociol. Rev. 26, 67–82.

Moreno, M.A., Whitehill, J.M., 2014. Inﬂuence of social media on alcohol use in ado-lescents and young adults. Alcohol Res. Curr. Rev. 36, 91–100.

Moreno, M.M., D’Angelo, J., Whitehill, J., 2016. Social media and alcohol: summary of research, intervention ideas and future study directions. Media Commun. 4, 50–59.

Moreno, M.A., Standiford, M., Cody, P., 2018. Social media and adolescent health. Curr. Pediatrics Rep. 6, 132–138.

Mu, K.J., Moore, S.E., LeWinn, K.Z., 2015. Internet use and adolescent binge drinking: ﬁndings from the monitoring the future study. Addict. Behav. Rep. 2, 61–66.

Norton, E.C., Dowd, B.E., 2018. Log odds and the interpretation of Logit models. Health Service Res. 53, 859–878.

Osgood, D.W., Wilson, J.K., O’Malley, P.M., Bachman, J.G., Johnston, L.D., 1996. Routine activities and individual deviant behavior. Am. Sociol. Rev. 61, 635–655.

Pape, H., Rossow, I., Brunborg, G.S., 2018. Adolescents drink less: how, who and why?A review of the recent research literature. Drug Alcohol Rev. 37, 98–114.

Pennay, A., Livingston, M., MacLean, S., 2015. Young people are drinking less: it is time toﬁnd out why. Drug Alcohol Rev. 34, 115–118.

Room, R., Babor, T., Rehm, J., 2005. Alcohol and public health. Lancet 365, 519–530.

Sampasa-Kanyinga, H., Chaput, J.-P., 2016. Use of social networking sites and alcohol consumption among adolescents. Public Health 139, 88–95.

Sutherland, E.H., 1947. Principles of Criminology, fourth ed. Philadelphia J.B., Lippincott.

Svensson, R., 2004. Gender diﬀerences in adolescent drug use: the impact of parental monitoring and peer deviance. Youth Soc. 34, 300–329.

Viner, R., 2005. Co-occurrence of adolescent health risk behaviors and outcomes in adult life:ﬁndings from a national birth cohort. J. Adolesc. Health 36, 98–99.

Viner, R.M., Taylor, B., 2007. Adult outcomes of binge drinking in adolescence:ﬁndings from a UK national birth cohort. J. Epidemiol. Community Health 61, 902–907.