Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 10.
Published in final edited form as: J Health Commun. 2020 Feb 25;25(3):214–222. doi: 10.1080/10810730.2020.1731632

Linguistic Predictors of Problematic Drinking in Alcohol-related Facebook Posts

Lyn M Van Swol 1, Chen-Ting Chang 1, Bradley Kerr 2, Megan Moreno 2
PMCID: PMC7654720  NIHMSID: NIHMS1642926  PMID: 32096449

Abstract

Emerging adults often increase problematic drinking during college. Although they generally do not seek help for problematic drinking, college students discuss their drinking on social media. This study followed college students’ Facebook profiles from the inception of their attendance at a university and identified alcohol-related posts. Within 28 days of their first alcohol-related Facebook post, participants were interviewed to assess problematic drinking (binge drinking episodes and number of drinks). Linguistic analysis of alcohol-related Facebook posts found that use of negative emotion language and swear words were related to problematic drinking, in support of proposed hypotheses. Results are situated within alcohol use disorder and health research examining the link between problematic drinking and anxiety, deviant behavior, and negative emotions.

Adolescents often start drinking alcohol or escalate drinking during the transition to college, with approximately 20% starting heavy, problematic binge drinking in college (Hingson, Heeren, Winter, & Wechsler, 2005; Wechsler et al., 2002). Binge drinking is a contributing factor to causes of death and high-risk behavior for college students, including accidents, suicide, homicide, injuries, and sexual assault (Deas & Brown, 2006; Derman, Cooper, & Agocha, 1998; Eaton et al., 2008; Hingson, Heeren, & Winter, 2006; Hingson, Zha, & Weitzman, 2009). Finding correlates to problematic drinking can help identify college students who can benefit from interventions and support. This is especially important because college students often do not seek out help or get screened for problematic drinking on their own (Foote, Wilkens, & Vavagiakis, 2004). One area where college students actually do advertise their drinking behavior is on social media (Moreno et al., 2010).

College students use social media to make references to drinking (Moreno et al., 2010; Moreno, Parks, & Richardson, 2007; Moreno, Parks, Zimmerman, Brito, & Christakis, 2009). For example, one study found 37% of 18-year old MySpace users made references to alcohol use (Moreno et al., 2010). Now a majority of college students use Facebook or other social media, but they continue to use it to make references to drinking (Duggan, Ellison, Lampe, Lenhart, & Madden, 2014; Lenhart, Purcell, Smith, & Zickuhr, 2010; Moreno et al., 2014), and these posts are related to drinking behavior. Past research has found that alcohol related postings on Facebook were positively related with self-reported use of alcohol (Moreno, Christakis, Egan, Brockman, & Becker, 2012; van Hoof, Bekkers, & Van Vuuren, 2014; Moreno, Cox, Young, & Haaland, 2015) with stronger relationships between posting of photos and profile pictures and problematic drinking (Moreno et al., 2015).

However, not all references to alcohol indicate problematic drinking, and identifying posts that may be correlated to a more serious problem is important. While past research has found a link between alcohol references on Facebook and drinking levels (Moreno et al., 2012), finding specific correlates of problematic drinking within a post, especially in ways that can be applied on a large scale with algorithms, will help fill a gap in the literature. Given the difficulty of coding social media posts, especially at a scale with social media sites like Facebook that have over a billion users (“Facebook Newsroom”, 2019), one solution might be use of automated coding software. Language software has been used in the social sciences to identify the relationship between interpersonal and group processes and language use (Pennebaker, Mehl, & Niederhoffer, 2003; van Swol & Kane, 2019). Holtgraves and Kashima (2008, p. 74) state “language use frequently involves the recoding of implicit, nonlinguistic representations into explicit, linguistic ones” because psychological processes are manifested through language use. A recent review of language use in groups found that group processes and behavioral outcomes like status, cohesion, well-being, and performance were related to language use (van Swol & Kane, 2019). For example, group members that use more positive emotion language in discussion tend to have higher ratings of well-being (van Swol & Kane, 2019).

Linguistic Inquiry Word Count (LIWC) (Pennebaker, Booth, Boyd, & Francis, 2015) is a software program that counts and categorizes words into over 80 categories and provides an index of percentage of language use for each category. Each category has a dictionary of words that represent that category. Categories can include parts of speech (e.g., adverbs, conjunctions), personal categories (e.g., work, family), and psychological processes (e.g., negative and positive affect). For example, the category negative emotion is represented by words like hurt, worried, ugly, kill, annoyed, or crying. Language analysis of alcohol-related Facebook posts could reveal relationships between emotions as manifested in language use and the behavior of problematic drinking. Thus, language analysis of alcohol displays on social media could be an efficient bottom-up means to examine when alcohol displays may be related to problematic drinking. While LIWC measures over 80 different categories with a dictionary of representative words, there are two categories that may be of interest to examining language use in alcohol references in social media.

First, use of negative emotion words may be related to problematic drinking. Use of negative emotion words has been related to coded or self-reported negative emotions such as anger, fear, dejection, depression, and anxiety (Gill, French, Gergle, & Oberlander, 2008; Johnsen, Vambheim, Wynn, & Wangberg, 2014; Settanni & Marengo, 2015; Tov, Ng, Lin, & Qiu, 2013), and researchers have analyzed negative emotion words used in social media posts to specifically assess negative well-being (Settanni & Marengo, 2015). Problematic drinking is often related to negative emotions like anxiety (Anker, Kushner, Thuras, Menk, & Unruh, 2016; Falk, Yi, & Hilton, 2008), and depression and alcohol misuse are highly correlated in emerging adults (Frohlick, Rapinda, O’Connor, & Keough, 2018). For example, Mushquash et al. (2013) tested a reciprocal influence model of depression and heavy drinking and found that depression and feelings of worthlessness made undergraduate women susceptible to heavy drinking. Research with alcohol use disorder (AUD) has found negative emotions as a risk factor for relapse into problematic drinking (Khantzian, 1997; Witkiewitz, 2011). Other research has indicated that anxiety and distress were both significantly positively related to drinking to cope with negative emotions and to conform to peer pressure among college students (Stewart, Morris, Mellings, & Komar, 2006), especially for those with low self-efficacy for avoiding heavy drinking in social situations (Gilles, Turk, & Fresco, 2006). Yet, while research has found an association between social anxiety and alcohol use in non-college samples, some other research results have been mixed for college samples (Ham & Hope, 2006).

Further, alcohol has not infrequently been used as self-medication for mood disorders such as depression and anxiety and for individuals suffering from AUD (Bolton, Robinson, & Sareen, 2009; Swendsen et al., 2000; Turner, Mota, Bolton, & Sareen, 2018). Research on AUD sufferers in an app-based treatment program that examined their language use in an online forum found a significant relationship between use of negative emotions words in the online forum and relapse rates (Kornfield, Toma, Shah, Moon, & Gustafson, 2017). This supports the potential of negative emotions expressed when sharing information about drinking experiences to be related to problematic drinking. Therefore, we hypothesize:

Hypothesis 1: There will be a positive correlation between self-reported problematic drinking – a.) binge drinking episodes and b.) number of drinks – and use of negative emotion language in alcohol-related Facebook posts.

Second, use of swear words are related to lack of agreeableness, deviance, impulsivity, immoderation, and negative emotions like anger and depression (Kornfield et al., 2017; Mehl, Gosling, & Pennebaker, 2006; Robbins, Mehl, Smith, & Weihs, 2013; Rodriguez, Holleran, & Mehl, 2010; Santos & Viera, 2017; Yarkoni, 2010), and given the previous discussion of problematic drinking and negative life outcomes (Deas & Brown, 2006; Derman et al., 1998; Eaton et al., 2008; Hingson et al., 2006; Hingson et al., 2009) and negative emotions (Anker et al., 2016; Falk et al., 2008), we expect a positive relationship between use of swear words and problematic drinking. Heavy alcohol use may be more likely in students who are impulsive and predisposed towards more deviant behavior (Perkins & Berkowitz, 1986), and swear words can reflect deviance and impulsivity more specifically than negative emotions words. Further, problematic drinking reduces inhibitions and increases impulsive behavior (Cohen-Gilbert at al., 2017), which could be related to increased use of swear words, and alcohol-related Facebook posts may occur while the person is intoxicated, and thus, disinhibited to expressions of immoderation or anger. Research on individuals with AUD in an online forum (Kornfield et al., 2017) found a relationship between use of swear words and rates of relapse into problematic drinking. Thus, we hypothesize:

Hypothesis 2: There will be a positive correlation between self-reported problematic drinking – a.) binge drinking episodes and b.) number of drinks – and use of swear words in alcohol-related Facebook posts.

Method

This secondary analysis study used phone interviews and content analysis data from a longitudinal study focused on identifying transitions in Facebook displays and offline health behaviors.

Participants

This study took place at two large state universities in the United States, one in the Midwest and the other in the Northwest. The study received approval from the two relevant institutional review boards. Eligible potential participants included graduated high school seniors planning to attend one of the two participating universities, who were 17–19 years old, English speaking, and owned a Facebook profile. Those who arrived on campus for summer early-enrollment programs were excluded. There were 127 participants who had alcohol-related Facebook posts during their freshman, sophomore, or junior year and completed an interview to determine their number of binge drinking episodes.

Potential participants were identified through random selection from the registrars’ lists of incoming first-year students at both universities. Recruitment involved several steps, starting with a pre-announcement postcard. Over a three to four week period, potential participants received contacts via up to four rounds of e-mails, phone calls, or Facebook messages.

Procedure

During the consent process, potential participants were informed that this longitudinal study focused on substance use and involved phone interviews and creation of a social media link with a research team Facebook profile. The research team indicated that participants’ Facebook profile content would be viewed, but that the research team would never post on participants’ profile. Participants were also asked to maintain open security settings with the research team’s Facebook profile for the duration of the study.

Phone Interviews.

Participants completed phone interviews lasting 30–60 minutes apiece. This approach was chosen because some participants lived more than an hour away from the primary research site, and because phone interviews have been successfully used to study stigmatizing topics such as risky health behaviors (Fortney et al., 2004; Meyer, Rossano, Ellis, & Bradford, 2002).

The baseline interview, conducted during the summer preceding the first year of college, included questions to obtain demographic data. The subsequent interview on which data about problematic drinking from this study is based took place for participants who had an initial displayed alcohol reference or an initial displayed alcohol reference related to intoxication or problem drinking on social media. When a coder identified one of these two types of alcohol references on the Facebook profile, the participant met criteria for an interview and was contacted. In order to link the social media post to behavior within a short time frame, interviews were conducted within 28 days after the reference was evaluated.

Content Analysis.

Seven trained coders used content analysis to evaluate alcohol references on participants’ social media profiles. Each coder completed a training period lasting 6–8 weeks. The first step was to review an established coding manual (Moreno, Egan, & Brockman, 2011). Subsequently, trainees shadowed the content analysis process, completed supervised practice coding, and reviewed and discussed their work with trained coders. Competency was assessed using interrater reliability which was good (Fleiss’ kappa = 0.80 for the presence or absence of alcohol references on profiles and Fleiss’ kappa = 0.73 for agreement among all coders for number of alcohol references) (Kimberlin & Winterstein, 2008). To become familiar with the material, trained coders completed a baseline evaluation of each participant’s Facebook profile for 3 months before each participant’s study enrollment. At the start of the university academic year, Facebook profile content was evaluated every 4 weeks for the preceding period of four weeks.

The research team adapted a previously developed codebook to evaluate displayed alcohol references, namely those referencing behaviors, on Facebook (Egan & Moreno, 2011; Moreno, Christakis, Egan, Brockman, & Becker, 2012; Moreno et al., 2011). In the present study, the definition of a displayed alcohol reference was expanded so that an alcohol-related attitude or intention was also included as a displayed alcohol reference.

This codebook was applied systematically to Facebook profiles to assess the presence of displayed alcohol references. First, coders reviewed the About section, in which alcohol references appeared, for example, as favorite quotes. The second section evaluated was the Likes area; an example liked page was “Beer Pong!” Third, coders examined the Timeline, including status updates and wall posts by friends. Wall posts by friends were only included in analysis if they had been liked by the participant. The fourth section reviewed was the Photos section; this area included the profile picture, cover photo, participants’ personal uploads, and tagged photos appearing on the participant’s timeline. Coders reviewed these sections to identity photos that included images of alcohol. For example, if a participant uploaded a photo of herself holding a can visibly labeled as beer, this would be included as an image of alcohol. Ambiguous photos, like the participant holding a red Solo cup with unknown contents, were not coded as an image of alcohol. Thus, data consisted of either typewritten descriptions of image-based alcohol references or verbatim text from profiles.

Measures

Interview

Alcohol Use.

Phone interviews after the first coded alcohol related Facebook post were used to collect alcohol experience information. Interviewers first asked, “Have you had a drink of alcohol in your lifetime?” and then, “In the past 28 days, have you had any drinks of alcohol?” Among those who reported last 28-day alcohol use, the interviewer used the Time Line Follow Back (TLFB) method to explore current drinking habits, which has been validated as an accurate measure of drinking (Sobell, Brown, Leo, & Sobell, 1996; Sobell & Sobell, 1992). Following this technique, the interviewer talked the participant through the day by day for the previous four weeks, asking on each day if alcohol was consumed, and if so, how many drinks. Thus, outcome variables pertaining to the TLFB include total drinks in the past 28 days and total binge drinks (5+ drinks for a male, 4+ drinks for a female) from that period (Alcoholism, 2018).

Social media

Content Analysis.

Displayed alcohol references were totaled for each year, for each participant. Thus, descriptive statistics associated with content analysis and content used in language measures included yearly total displayed alcohol references.

Language measures.

This study used the software program Linguistic Inquiry and Word Count (LIWC) to conduct natural language processing. Decades of research to understand narrative voice in health support this approach (Berry & Pennebaker, 1993; Francis & Pennebaker, 1992; Pennebaker, 1993; Pennebaker & King, 1999; Pennebaker, Mayne, & Francis, 1997; Rosenberg et al., 2002). Developed through a rigorous process in which groups of judges reviewed 2,000 words or word stems and determined their relation to specific categories, LIWC uses validated internal dictionaries. LIWC analyzes a document by comparing every word in the transcript with “dictionaries” of up to 74 dimensions across these categories, and then calculates a proportion of words falling into different categories, ranging from emotion words to words about social context (LIWC, 2016). LIWC has been validated for content and construct validity (Francis & Pennebaker, 1992; Stirman & Pennebaker, 2001), and interrater reliability discrimination of categories has been found to range from 86% to 100% depending on the dimension being assessed, supporting content validity.

All of a participant’s Facebook posts for a given academic year that were coded as referencing alcohol were copied into a transcript and edited for analysis in LIWC. Namely, LIWC dictionaries were used to identify negative emotion words (e.g., hurt, ugly, nasty, worthless) and swear words (e.g, fuck, damn, shit) in displayed alcohol references. See Table 1 for examples.

Table 1.

Examples of Facebook posts for each language category analyzed

LIWC variable Example status update
Negative emotion words *”Is freshman year really over soon? Before I know it, I’ll be graduating and looking for a job. This is scary… I love beer.”
Swear words *”i feel like shit… I’m never drinking again until 4”
*”I got jäger bomb. I think we could have ourselves one hell of a night”
Open in a new tab

Analyses

Demographic variables, alcohol use, displayed alcohol references on Facebook, and language measures were evaluated with descriptive statistics.

Results

Demographics

Interview data of problematic drinking was only collected after the first alcohol-related social media post. Thus, if the first alcohol-related post occurred during freshman year, then problematic drinking was assessed during this year only. Due to the difficultly of obtaining participation, participants were only interviewed after their first alcohol-related social media post and not after subsequent posts. Thus, data is cross-sectional, and participants’ drinking was only analyzed for the year where they initiated alcohol-related posts on Facebook. Data was not analyzed for senior year because initial references to alcohol mostly emerge during the first three years of college and by senior year there were few participants who had not already displayed initial alcohol references on Facebook. Year in school is included as a covariate. For freshman year, there were 95 participants who initiated at least one alcohol related post on Facebook. A total of 61 participants had a Facebook display change and completed an interview to determine their number of binge drinking episodes (M = 4.76, SD = 3.98, min = 0, max = 16) and number of drinks (M = 40.09, SD = 34.65, min = 1, max = 157) over a 30 day period. Among these 61 freshman participants whose data were analyzed, 31 were female and 30 male; 10 attended the Northwestern university and 51 attended the Midwestern university. Three participants (4.9%) reported as Asian, 50 (82%) as white, 3 (4.9%) as Hispanic, 4 (6.6%) as multi-racial, and one participant did not respond. Further, they posted an average of 2.48 alcohol related posts on Facebook (SD = 2.88, min = 1, max = 16).

For sophomore year, there were 88 participants who initiated at least one alcohol related post on Facebook. A total of 39 participants had a Facebook display change and completed an alcohol change interview to determine their number of binge drinking episodes (M = 3.73, SD = 2.57, min = 0, max = 10) and number of drinks (M = 31.31, SD = 25.20, min = 9, max = 137) over a 30 day period. Among these 39 sophomore participants whose data were analyzed, 24 were female and 15 male, 12 attended the Northwestern university and 27 attended the Midwestern university. Six participants (15.4%) reported as Asian, 31 (79.5%) as white, and 2 (5.1%) as multi-racial. Further, they posted an average of 2.69 alcohol related posts on Facebook (SD = 2.94, min = 1, max = 18).

For junior year, there were 95 participants who had at least one alcohol related post on Facebook. A total of 27 participants had a Facebook display change and completed an interview to determine their number of binge drinking episodes (M = 3.56, SD = 3.14, min = 0, max = 10) and number of drinks (M = 30.48, SD = 20.12, min = 3, max = 83) over a 30 day period. Among these 27 junior participants whose data will be analyzed, 20 are female and 7 male, 6 attended the Northwestern university and 21 attended the Midwestern university. Three participants (11.1%) reported as Asian, 21 (77.8%) as white, and 3 (11.1%) as multi-racial. Further, they posted an average of 2.15 alcohol related posts on Facebook (SD = 1.75, min = 1, max = 8).

Control Variables

Women engaged in fewer binge drinking episodes (Women = 3.71, SD = 2.95; Men = 5.04, SD = 3.93) and had fewer total alcoholic drinks (Women = 25.92, SD = 17.91; Men = 48.96, SD = 36.86) than men, t(125) =−2.18, p = .03; t(125) =−4.68, p < .001, respectively. Therefore, gender is included in regression analysis as a control. Students at the Northwestern university engaged in fewer binge drinking episodes (Northwestern = 2.86, SD = 2.43; Midwestern = 4.65, SD = 3.58) and had fewer alcoholic drinks (Northwestern = 23.38, SD = 18.91; Midwestern = 38.74, SD = 31.06) than students in the Midwestern university, t(125) = −2.49, p = .01; t(125) =−2.49, p = .01, respectively. Therefore, university is included in the regression. Because there were fewer students from the Northwetsern university, we examined the range for outliers. For total drinks, the range was 3 to 70 (M = 24.02, SD = 18.89), and for binge drinking episodes (M = 2.90, SD = 2.40), the range was 0 to 9. For students at the Midwestern university, the range for total drinks was 1 to 158 (M = 38.62, SD = 30.62), and the range for binge drinking episodes was 0 to 16 (M = 4.66, SD = 3.53). There was not a large outlier among students from the Northwestern university.

Negative emotion words

A regression with the predictors of gender (0 = female, 1 = male), university (0 = Northwestern, 1 = Midwestern), year in school (1 = freshman, 2 = sophomore, 3 = junior), and negative emotion language and the dependent variable of binge drinking episodes was significant. Negative emotion words significantly predicted episodes of binge drinking. The same regression was conducted but with the dependent variable of number of alcoholic drinks reported in a 30 day period. The regression was significant. Negative emotion words significantly predicted number of alcoholic drinks. See Table 2 for regression table.

Table 2.

Regression analyses on negative emotion language

t p β Unstandardized B SE
Binge drinkinga
Negative emotion language 4.35 <.001 .36 .46 .11
Year in school −1.87 0.06 −.16 −.68 .37
Gender 2.11 0.04 .18 1.21 .57
University 2.36 0.02 .20 1.58 .67
Number of alcoholic drinksb
Negative emotion language 3.80 <.001 .30 3.43 .90
Year in school −1.07 .29 −.09 −3.30 3.09
Gender 4.54 <.001 .37 22.07 4.86
University 2.22 .03 .18 12.61 5.68
Open in a new tab

Note.

a

R = .49, R2 = .24, SE = 3.01, F(4, 118) = 9.16, p < .001;

b

R = .55, R2 = .30, SE = 25.56, F(4, 118) = 12.03, p < .001.

Swear words

A regression with gender, university, year in school, and swear words significantly predicted episodes of binge drinking, and percentage of swear words was a significant predictor. A similar regression predicting number of alcoholic drinks was significant, and swear words was a significant predictor. See Table 3 for regression.

Table 3.

Regression analyses on swear words

t p β Unstandardized B SE
Binge drinkinga
Swear words 3.51 <.001 .30 .54 .15
Year in school −1.39 .17 −.12 −.52 .38
Gender 2.14 .03 .18 1.26 .59
University 2.06 .04 .18 1.42 .69
Number of alcoholic drinksb
Swear words 3.11 .002 .25 4.00 1.29
Year in school −.67 .50 −.06 −2.12 3.15
Gender 4.53 <.001 .37 22.43 4.95
University 1.96 .05 .16 11.37 5.80
Open in a new tab

Note.

a

R = .45, R2 = .20, SE = 3.10, F(4, 118) = 7.29, p < .001;

b

R = .52, R2 = .27, SE = 26.05, F(4, 118) = 10.52, p < .001.

Discussion

This study examined if language used in Facebook posts referencing alcohol was related to self-reported problematic drinking that was assessed in interviews after an initial alcohol related Facebook post. The study analyzed freshman, sophomores, and juniors and included participants with variety in their drinking, as not all participants had problematic drinking behavior. In addition, behavior was measured within four weeks of the observed social media post so that links between the language and behavior could be established for recent behavior. Hypotheses that use of negative emotion language and swear words would be related to problematic drinking were supported. This study adds to the previous literature in several ways. First, language use is related to a specific self-reported behavioral measure, rather than intention or self-reported well-being. Second, the study is the first to focus in on language use on social media and problematic drinking behavior. Previous research has examined use of negative emotion words and swear words in an online Alcohol Use Disorder recovery chat group and found a relationship between this language use and relapse rate (Kornfield et al., 2017), so this current study helps build on this previous research and extend it to a different context with a different population.

Research on language use has demonstrated that use of negative emotion words is related to negative emotions and well-being (Gill et al., 2008; Johnsen et al., 2014; Settanni & Marengo, 2015; Tov et al., 2013). Further, research has specifically examined how language use is related to health behaviors (Pennebaker & King, 1999; Rosenberg et al., 2002), like how language use relates to physical health during bereavement (Pennebaker et al., 1997) or relapse rates of problematic drinking of individuals suffering from AUD (Kornfield et al., 2017). These results add to previous research that has found a link between problematic drinking and use of negative emotions in online forums for sufferers of AUD (Kornfield et al., 2017). However, this study cannot speak to any causality in the relationship between use of negative emotion words and problematic drinking.

In complication/scar models, depression and negative emotions are a consequence of problematic drinking (Mushquash et al., 2013). These models suggest that negative emotions are induced by problematic drinking through physiological or negative psychological effects associated with problematic drinking (Swendsen & Merikangas, 2000). For example, Hussong, Hicks, Levy, and Curran (2001) found that problematic drinking among undergraduates predicted negative affect. However, vulnerability models of drinking posit that negative emotions create a vulnerability for problematic drinking. The “self-medication hypothesis” fits with vulnerability models and suggests that stress-driven drinking to reduce negative emotional states may be a motivator for problematic drinking outcomes (Swendsen at al., 2000). College students often have high levels of stress, which is related to impulse control disorders and poor health (Leppink, Odlaug, Lust, Christenson, & Grant, 2016), and one coping mechanism for stress and anxiety can be problematic drinking (Anker et al., 2016; Falk et al., 2008). However, other earlier research (Egan & Moreno, 2011) found that stress references were common on Facebook for college students, but not related to alcohol references. The difference between that earlier study and this study could reflect the measurement of negative emotion words rather than specific human coding of stress-related content in the Facebook post or could reflect a cultural shift in the past few years. Further, negative emotion words encompass a larger variety of emotions than just stress. This study could not offer support for either the complication/scar models or vulnerability models because the results are correlational. Most likely alcohol and negative emotions influence each other in a dynamic and reciprocal cycle with other additional factors, like impulse control or negative life circumstances, affecting both the emotional content of Facebook posts and problematic drinking.

Use of swear words are related to negative emotions and also to immoderation (Mehl et al., 2006; Santos & Viera, 2017; Yarkoni, 2010). Use of swear words was related to problematic drinking in this study. Research on the correlation between the 5-factor model of personality traits and word use indicated that low conscientiousness and low agreeableness are both associated with higher use of swear words (Qiu, Lin, Ramsay, & Yang, 2012). People who are more disorderly and/or with a less cooperative nature are more likely to use swear words. Berkowitz and Perkins (1986) note students who are impulsive and predisposed to deviant behaviors often have more problematic drinking, possibly for escapist reasons. Research has also found that alcohol involvement was associated with low conscientiousness and low agreeableness (Malouff, Thorsteinsson, Rooke, & Schutte, 2007) and that heavy drinking in college is associated with personality factors such as impulsivity and nonconformity (Baer, Kivlahan, Blume, McKnight, & Marlatt, 2001). People who score low on conscientiousness and agreeableness are also more likely to engage in antisocial (Ozer & Benet-Martinez, 2006) and dark triad behavior (Paulhus & Williams, 2002). Thus, put together, college students who use more swear words to express themselves on social media may be more likely to engage in solitary or escapist heavy episodic drinking, which research findings have suggested is associated with having greater alcohol-related problems than heavy drinking in only social contexts (Gonzalez & Skewes, 2013). For future research, examining the link between swear words on social media and problematic drinking will allow a better understanding of how these variables are related.

Limitations and Future Research

While adding to previous research on language use, this study had several limitations. First, we only examined Facebook posts that referenced alcohol. It is possible that negative emotion language or swear words in general, not just in alcohol-related posts, could be related to problematic drinking, but it was beyond the scope of this study to collect this data. Second, we only examined Facebook and not other social media that have gained in prominence, like Snapchat or Instagram. Given the longitudinal nature of this study, when we began this research Facebook was more popular for college students, but it has lost popularity to Snapchat and Instagram. Therefore, while Facebook was not competing with Snapchat and Instagram when we began our study, this is a limitation of our results. Boyle, LaBrie, Froidevaux, and Witkovic (2016) found that college students are increasingly using these social media platforms more than Facebook to post alcohol-related content and that content on these platforms is more strongly related to alcohol consumption. In addition, content on these sites is more video and image based, which could limit generalizing results from this study. Third, participants knew they were being monitored by the research team, and this could have resulted in a social desirability bias in posted content. For example, features like Snapchats’ disappearing posts could encourage participants to post more deviant content like referencing drinking to their social media than they would post to Facebook. Fourth, we could only code what participants posted publicly, but participants could be disclosing information about their drinking in more private forums to friends. This also highlights the need to balance privacy when examining social media. Fifth, people have a positivity norm in content posted to Facebook (Waterloo, Baumgartner, Peter, & Valkenburg, 2018), so students may be hesitant to post negative content. This could limit practical implications of this study.

Sixth, while computer-aided analysis of Facebook posts has the benefits of efficiency and unobtrusiveness, there are limitations to this research. One is that LIWC only examines single words and is unable to account for the meaning in context (e.g., “you’re killing me” to express positive emotions). Further, while we found a relationship between negative emotion words and problematic drinking, LIWC is not validated to measure depression. Therefore, caution should be exercised to make a connection between use of negative emotion words, depression, and problematic drinking from this study, but future research could try to more explicitly measure depression to tests these relationships. Further, some studies have had mixed results for validation of LIWC (McDonnell, 2015).

Future research could address some of these limitations. For example, both alcohol-related posts and other posts on social media could be analyzed for use of language and its relationship to problematic drinking. College students have high rates of depression and anxiety (i.e., Mahmoud, Staten, Hall, Lennie 2012) and often use a lot of profanity, especially online (Turel and Qahir-Saremi 2017). Therefore, use of negative emotion language and swearing in general could be tested for their relationship with problematic drinking. In addition, future research could examine how social media content changes after participants start to initiate alcohol-related social media posts.

In conclusion, this study sought to determine if linguistic correlates of problematic drinking could be found in social media posts referencing alcohol. Use of negative emotion language and swearing was strongly related to problematic drinking. A broad literature has indicated that brief interventions for college students are effective in reducing heavy drinking (Baer et al., 2001). In addition to traditional intervention such as providing individualized feedback and advice using motivational interviewing, another way is to work with social media platforms to take advantage of algorithms. For example, it may be beneficial to design an algorithm that detects word usage and identify problem drinkers, allowing more personalized health messages that encourage college students to seek help for drinking problems or emotional distress.

References

  1. Alcoholism, N. I. o. A. A. a. (2018). Drinking Levels Defined Alcohol Facts and Statistics. and Practice (pp. 67–98). Philadelphia, PA: John Wiley and Sons. [Google Scholar]
  2. Anker JJ, Kushner MG, Thuras P, Menk J, & Unruh AS (2016). Drinking to cope with negative emotions moderates alcohol use disorder treatment response in patients with co-occurring anxiety disorder. Drug and Alcohol Dependence, 159, 93–100. doi: 10.1016/j.drugalcdep.2015.11.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baer JS, Kivlahan DR, Blume AW, McKnight P, & Marlatt GA (2001). Brief intervention for heavy-drinking college students: 4-year follow-up and natural history. American Journal of Public Health, 91(8), 1310–1316. doi: 10.2105/AJPH.91.8.1310 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Berry DS, & Pennebaker JW (1993). Nonverbal and verbal emotional expression and health. Psychotherapy and Psychosomatics, 59, 11–19. doi: 10.1159/000288640 [DOI] [PubMed] [Google Scholar]
  5. Bolton JM, Robinson J, & Sareen J (2009). Self-medication of mood disorders with alcohol and drugs in the National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Affective Disorders, 115, 367–375. doi: 10.1016/j.jad.2008.10.003 [DOI] [PubMed] [Google Scholar]
  6. Boyle SC, LaBrie JW, Froidevaux NM, & Witkovic YD (2016). Different digital paths to the key? How exposure to peers’ alcohol-related social media content influences drinking among male and female first year college students. Addictive Behaviors, 57, 21–29. doi: 10.1016/j.addbeh.2016.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chan J, & Schunn CD (2017) A computational linguistic approach to modeling the dynamics of design processes In Christensen BT, Ball LJ, & Halskov K (Eds.), Analysing Design Thinking: Studies of Cross-Cultural Co-Creation (pp. 521–540). Leiden: CRC Press/Taylor & Francis. [Google Scholar]
  8. Cohen-Gilbert JE, Nickerson LD, Sneider JT, Oot EN, Seraikas AM, Rohan ML, & Silveri MM (2017). College binge drinking associated with decreased frontal activation to negative emotional distractors during inhibitory control. Frontiers in Psychology: Psychopathology, 8, 1–12. doi: 10.3389/fpsyg.2017.01650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Deas D, & Brown ES (2006). Adolescent substance abuse and psychiatric comorbidities. Journal of Clinical Psychiatry, 67, e02. doi: 10.4088/JCP.0706e02 [DOI] [PubMed] [Google Scholar]
  10. Derman KH, Cooper ML, & Agocha VB (1998). Sex-related alcohol expectancies as moderators of the relationship between alcohol use and risky sex in adolescents. Journal of Studies on Alcohol, 59, 71–77. doi: 10.15288/jsa.1998.59.71 [DOI] [PubMed] [Google Scholar]
  11. Duggan M, Ellison NB, Lampe C, Lenhart A, & Madden M, (2014). Social media update 2014. Pew Research Center. Retrieved April 24, 2015, from https://www.pewinternet.org/2015/01/09/social-media-update-2014/ [Google Scholar]
  12. Eaton DK, Kann L, Kinchen S, Shanklin S, Ross, Hawkins J, … & Lim C (2008). Youth risk behavior surveillance-United States, 2007. MMWR Surveillance Summary, 57, 1–131. Retrieved from http://worldcat.org/issn/00910031 [PubMed] [Google Scholar]
  13. Egan KG, & Moreno MA (2011). Prevalence of Stress References on College Freshmen Facebook Profiles. Cin-Computers Informatics Nursing, 29(10), 586–592. doi: 10.1097/NCN.0b013e3182160663 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Facebook Newsroom. (2019). Retrieved from https://newsroom.fb.com/company-info/
  15. Falk DE, Yi HY, & Hilton ME (2008). Age of onset and temporal sequencing of lifetime DSM-IV alcohol use disorders relative to comorbid mood and anxiety disorders. Drug and Alcohol Dependence, 94, 234–245. doi: 10.1016/j.drugalcdep.2007.11.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fast E, & Horvitz E (2016). Identifying dogmatism in social media: Signals and models Proceedings from Conference on Empirical Methods in Natural Language Processing (pp. 690–699). Austin, Texas, November, 2016. Association for Computational Linguistics. [Google Scholar]
  17. Foote J, Wilkens C, & Vavagiakis P (2004). A national survey of alcohol screening and referral in college health centers. Journal American College Health, 52, 149–157. Retrieved from https://europepmc.org/abstract/med/15018426 [PubMed] [Google Scholar]
  18. Fortney J, Mukherjee S, Curran G, Fortney S, Han XT, & Booth BM (2004). Factors associated with perceived stigma for alcohol use and treatment among at-risk drinkers. Journal of Behavioral Health Services & Research, 31, 418–429. doi: 10.1007/BF02287693 [DOI] [PubMed] [Google Scholar]
  19. Francis ME & Pennebaker JW (1992). Talking and writing as illness prevention. Medicine, Exercise, Nutrition and Health, 1, 27–33. [Google Scholar]
  20. Frohlick JR, Rapinda KK, O’ Connor RM, & Keough MT (2018). Examining co-patterns of depression and alcohol misuse in emerging adults following university graduation. Addictive Behavior Reports, 8, 40–45. doi: 10.1016/j.abrep.2018.06.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gill AJ, French RM, Gergle D, Oberlander J (2008). The language of emotion in short blog texts. Proceedings of CSCR, 299–302. doi: 10.1145/1460563.1460612 [DOI] [Google Scholar]
  22. Gilles DM, Turk CL, & Fresco DM (2006). Social anxiety, alcohol expectancies, and self-efficacy as predictors of heavy drinking in college students. Addictive Behaviors, 31, 388–398. doi: 10.1016/j.addbeh.2005.05.020 [DOI] [PubMed] [Google Scholar]
  23. Gonzalez VM, & Skewes MC (2013). Solitary heavy drinking, social relationships, and negative mood regulation in college drinkers. Addiction Research & Theory, 21(4), 285–294. doi: 10.3109/16066359.2012.714429 [DOI] [Google Scholar]
  24. Ham LS, & Hope DA (2006). Incorporating social anxiety into a model of college problem drinking: Replication and extension. Psychology of Addictive Behaviors, 20, 348–355. doi: 10.1037/0893-164X.20.3.348 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hingson RW, Heeren T, & Winter MR (2006). Age at drinking onset and alcohol dependence: age at onset, duration, and severity. Archives of Pediatrics & Adolescent medicine, 160(7), 739–746. doi: 10.1001/archpedi.160.7.739 [DOI] [PubMed] [Google Scholar]
  26. Hingson RW, Zha W, & Weitzman ER (2009). Magnitude of and trends in alcohol related mortality and morbidity among U.S. college students ages 18–24, 1998–2005. Journal of Studies on Alcohol and Drugs Supplement, 16, 12–20.doi: 10.15288/jsads.2009.s16.12 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Hingson R, Heeren T, Winter M, & Wechsler H (2005). Magnitude of alcohol-related mortality and morbidity among U.S. college students ages 18–24: Changes from 1998 to 2001. Annual Review of Public Health, 26, 259–279. doi: 10.1146/annurev.publhealth.26.021304.144652 [DOI] [PubMed] [Google Scholar]
  28. Holtgraves TM, & Kashima Y (2008). Language, meaning, and social cognition. Personality and Social Psychology Review, 12, 73–94. doi: 10.1177/1088868307309605 [DOI] [PubMed] [Google Scholar]
  29. Hussong AM, Hicks RE, Levy SA, & Curran PJ (2001). Specifying the relations between affect and heavy alcohol use among young adults. Journal of Abnormal Psychology, 110, 449–461. 10.1037/0021-843X.110.3.449. [DOI] [PubMed] [Google Scholar]
  30. Johnsen JK, Vambheim SM, Wynn R, & Wangberg SC (2014). Language motivation and emotion in an internet support group for smoking cessation: Explorative use of automated content analysis to measure regulatory focus. Psychology Research Behavior Management, 7, 19–29. doi: 10.2147/PRBM.S54947 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Khantzian EJ (1997). The self-medication hypothesis of substance use disorders: A reconsideration and recent applications. Harvard Review of Psychiatry, 4, 231–244. doi: 10.3109/10673229709030550 [DOI] [PubMed] [Google Scholar]
  32. Kimberlin CL, & Winterstein AG (2008). Validity and reliability of measurement instruments used in research. American Journal of Health System Pharmacy, 65, 2276–84. doi: 10.2146/ajhp070364. [DOI] [PubMed] [Google Scholar]
  33. Kornfield R, Toma CL, Shah DV, Moon TJ, & Gustafson DH (2017). What do you say before you relapse? How language use in a peer-to-peer online discussion forum predicts risky drinking among those in recovery. Health Communication, 33, 1184–1193. doi: 10.1080/10410236.2017.1350906 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lenhart A, Purcell K, Smith A, & Zickuhr K (2010). Social media and young adults. Pew Internet and American Life Research Center Project: Washington, DC: Retrieved from http://www.pewinternet.org/2010/02/03/social-media-and-young-adults/ [Google Scholar]
  35. Leppink EW, Odlaug BL, Lust K, Christenson G, & Grant JE (2016). The young and the stressed: Stress, impulse control, and health in college students. Journal of Nervous Mental Disease, 12, 931–938. doi: 10.1097/NMD.0000000000000586 [DOI] [PubMed] [Google Scholar]
  36. Mahmoud JSR, Staten RT, Hall LA, & Lennie TA (2012) The Relationship among young adult college students’ depression, anxiety, stress, demographics, life satisfaction, and coping styles. Issues in Mental Health Nursing, 33:3, 149–156. doi: 10.3109/01612840.2011.632708 [DOI] [PubMed] [Google Scholar]
  37. Malouff JM, Thorsteinsson EB, Rooke SE, & Schutte NS (2007). Alcohol involvement and the Five-Factor Model of personality: A meta-analysis. Journal of Drug Education, 37, 277–294. doi: 10.2190/DE.37.3.d [DOI] [PubMed] [Google Scholar]
  38. McDonnell M (2015). Comparison of text analysis programs for identification of emotional expression. Loma Linda University Electronic Theses, Dissertations & Projects. Retrieved from https://scholarsrepository.llu.edu/etd/308 [Google Scholar]
  39. Mehl MR, Gosling SD, & Pennebaker JW (2006). Personality in its natural habitat: Manifestations and implicit folk theories of personality in daily life. Journal of Personality and Social Psychology, 90, 862–877. doi: 10.1037/0022-3514.90.5.862 [DOI] [PubMed] [Google Scholar]
  40. Meyer IH, Rossano L, Ellis JM, & Bradford J (2002). A brief telephone interview to identify lesbian and bisexual women in random digit dialing sampling. Journal of Sex Research, 39, 139–144. doi: 10.1080/00224490209552133 [DOI] [PubMed] [Google Scholar]
  41. Moreno MA, Christakis DA, Egan KG, Brockman LN, & Becker T (2012). Associations between displayed alcohol references on Facebook and problem drinking among college students. Archives of Pediatric Adolescent Medicine, 166, 157–163. doi: 10.1001/archpediatrics.2011.180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Moreno MA, Briner LR, Williams A, Brockman L, Walker L, & Christakis DA (2010). A content analysis of displayed alcohol references on a social networking web site. Journal of Adolescent Health, 47, 168–175. doi: 10.1016/j.jadohealth.2010.01.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Moreno MA, Cox ED, Young HN, & Haaland W (2015). Underage college students’ alcohol displays on Facebook and real-time alcohol behavior. Journal of Adolescent Health, 56, 646–651. doi: 10.1016/j.jadohealth.2015.02.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Moreno MA, D’Angelo J, Kacvinsky LE, Kerr B, Zhang C, & Eickhoff J (2014). Emergence and predictors of alcohol reference displays on Facebook during the first year of college. Computers and Human Behavior, 30, 87–94. doi: 10.1016/j.chb.2013.07.060 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Moreno MA, Egan KG, & Brockman L (2011). Development of a Researcher Codebook for Use in Evaluating Social Networking Site Profiles. Journal of Adolescent Health, 49(1), 29–35. doi: 10.1016/j.jadohealth.2011.04.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Moreno MA, Parks MR, Zimmerman FJ, Brito TE, & Christakis DA (2009). Display of health risk behaviors on MySpace by adolescents: Prevalence and associations. Archives of Pediatrics and Adolescent Medicine, 163, 27–34. doi: 10.1001/archpediatrics.2008.528 [DOI] [PubMed] [Google Scholar]
  47. Moreno MA, Parks M, Richardson LP (2007). What are adolescents showing the world about their health risk behaviors on MySpace? Medscape General Medicine, 9, 9 Retrieved from https://europepmc.org/abstract/med/18311359 [PMC free article] [PubMed] [Google Scholar]
  48. Mushquash AR, Stewart SH, Sherry SB, Sherry DL, Mushquash CJ, & Mackinnon AL (2013). Depressive symptoms are a vulnerability factor for heavy episodic drinking: A short-term, four-wave longitudinal study of undergraduate women. Addictive Behaviors, 38, 2180–2186. doi: 10.1016/j.addbeh.2012.11.008 [DOI] [PubMed] [Google Scholar]
  49. Ozer DJ, & Benet-Martinez V (2006). Personality and the prediction of consequential outcomes. Annual Review of Psychology, 57, 401–421. doi: 10.1146/annurev.psych.57.102904.190127 [DOI] [PubMed] [Google Scholar]
  50. Paulhus DL, & Williams KM (2002). The dark triad of personality: Narcissism, Machiavellianism, and psychopathy. Journal of Research in Personality, 36(6), 556–563. doi: 10.1016/S0092-6566(02)00505-6 [DOI] [Google Scholar]
  51. Pennebaker JW (1993). Putting stress into words: Health, linguistic, and therapeutic implications. Behaviour Research and Therapy, 31, 539–548. doi: 10.1016/0005-7967(93)90105-4 [DOI] [PubMed] [Google Scholar]
  52. Pennebaker JW, Booth RJ, Boyd RL, & Francis ME (2015). Linguistic Inquiry and Word Count: LIWC2015. Retrieved from: https://s3-us-west-2.amazonaws.com/downloads.liwc.net/LIWC2015_OperatorManual.pdf
  53. Pennebaker JW & King LA (1999). Linguistic styles: Language use as an individual difference. Journal of Personality and Social Psychology, 77, 1296–1312. doi: 10.1037//0022-3514.77.6.1296 [DOI] [PubMed] [Google Scholar]
  54. Pennebaker JW, Mayne TJ, & Francis ME (1997). Linguistic predictors of adaptive bereavement. Journal of Personality and Social Psychology, 72, 863–871. doi: 10.1037/0022-3514.72.4.863 [DOI] [PubMed] [Google Scholar]
  55. Pennebaker JW, Mehl MR, & Niederhoffer KG (2003). Psychological aspects of natural language use: Our words, our selves. Annual Review of Psychology, 54, 547–577. doi: 10.1146/annurev.psych.54.101601.145041 [DOI] [PubMed] [Google Scholar]
  56. Perkins HW, & Berkowitz AD (1986). Perceiving the community norms of alcohol use among students: Some research implications for campus alcohol education programming. International Journal of the Addictions, 21, 961–976. 10.3109/10826088609077249 [DOI] [PubMed] [Google Scholar]
  57. Qiu L, Lin H, Ramsay J, & Yang F (2012). You are what you tweet: Personality expression and perception on Twitter. Journal of Research in Personality, 46(6), 710–718. doi: 10.1016/j.jrp.2012.08.008 [DOI] [Google Scholar]
  58. Robbins ML, Mehl MR, Smith HL, & Weihs KL (2013). Linguistic indicators of patient, couple, and family adjustment following breast cancer. Psycho-Oncology, 22, 1501–1508. doi: 10.1002/pon.3161 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Rodriguez AJ, Holleran SE, & Mehl MR (2010). Reading between the lines: The lay assessment of subclinical depression from written self-descriptions. Journal of Personality, 78, 575–598. doi: 10.1111/j.1467-6494.2010.00627.x [DOI] [PubMed] [Google Scholar]
  60. Rosenberg HJ, Rosenberg SD, Ernstoff MS, Wolford GL, Amdur RJ, Elshamy M,… Pennebaker JW (2002). Expressive disclosure and health outcomes in a prostate cancer population. International Journal of Psychiatry in Medicine, 32, 37–53. doi: 10.2190/AGPF-VB1G-U82E-AE8C [DOI] [PubMed] [Google Scholar]
  61. Santos HDP, & Viera R (2017). Psycholinguistic features for emotion intensity prediction in tweets. Proceedings from the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, (pp. 189–192), Copenhagen, Denmark. [Google Scholar]
  62. Settanni M, & Marengo D (2015) Sharing feelings online: studying emotional well-being via automated text analysis of Facebook posts. Frontiers in Psychology, 6, 1045. doi: 10.3389/fpsyg.2015.01045 [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Sobell LC, Brown J, Leo GI, & Sobell MB (1996). The reliability of the Alcohol Timeline Followback when administered by telephone and by computer. Drug and Alcohol Dependence, 42(1), 49–54. doi: 10.1016/0376-8716(96)01263-x [DOI] [PubMed] [Google Scholar]
  64. Sobell LC, & Sobell MB (1992). Timeline follow-back: A technique for assessing self-reported alcohol consumption. Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, 41–72. doi: 10.1007/978-1-4612-0357-5_3 [DOI] [Google Scholar]
  65. Stewart SH, Morris E, Mellings T, & Komar J (2006). Relations of social anxiety variables to drinking motives, drinking quantity and frequency, and alcohol-related problems in undergraduates. Journal of Mental Health, 15, 671–682. doi: 10.1080/09638230600998904 [DOI] [Google Scholar]
  66. Stirman SW, & Pennebaker JW (2001). Word use in the poetry of suicidal and non-suicidal poets. Psychosomatic Medicine, 63, 517–522. doi: 10.1097/00006842-200107000-00001 [DOI] [PubMed] [Google Scholar]
  67. Swendsen JD, & Merikangas KR (2000). The comorbidity of depression and substance use disorders. Clinical Psychology Review, 20, 173–189. 10.1016/S0272-7358(99)00026-4. [DOI] [PubMed] [Google Scholar]
  68. Swendsen JD, Tennen H, Carney MA, Affleck G Willard A, & Hromi A (2000). Mood and alcohol consumption: An experience sampling test of the self-medication hypothesis. Journal of Abnormal Psychology, 109, 198–204. doi: 10.1037/0021-843X.109.2.198 [DOI] [PubMed] [Google Scholar]
  69. Tov W, Ng NL, Lin H, & Qiu L (2013). Detecting well-being via computerized content analysis of brief diary entries. Psychological Assessment, 25, 1069–1078. doi: 10.1037/a0033007 [DOI] [PubMed] [Google Scholar]
  70. Turel O, & Qahri-Saremi H (2016) Problematic use of social networking sites: Antecedents and consequence from a dual-system theory perspective. Journal of Management Information Systems, 33:4, 1087–1116. doi: 10.1080/07421222.2016.1267529 [DOI] [Google Scholar]
  71. Turner S, Mota N, Bolton J, & Sareen J (2018). Self-medication with alcohol or drugs for mood and anxiety disorders: A narrative review of the epidemiological literature. Depression and Anxiety, 35, 851–860. doi: 10.1002/da.22771 [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. van Hoof JJ, Bekkers J, & van Vuuren HA (2014). Son, you’re smoking on Facebook! College students’ disclosures on social networking sites as indicators of real-life risk behaviors. Computers in Human Behavior, 34, 249–257. doi: 10.1016/j.chb.2014.02.008 [DOI] [Google Scholar]
  73. van Swol LM, Kane A (2019). Language and group processes: An integrative, interdisciplinary review. Small Group Processes, 50, 3–38. doi: 10.1177/1046496418785019 [DOI] [Google Scholar]
  74. Waterloo SF, Baumgartner SE, Peter J, & Valkenburg PM (2018). Norms of online expressions of emotion: Comparing Facebook, Twitter, Instagram, and WhatsApp. New Media & Society, 20, 1813–1831. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Wechsler H, Lee JE, Kuo M, Seibring M, Nelson TF, & Lee H (2002). Trends in college binge drinking during a period of increased prevention efforts. Findings from 4 Harvard School of Public Health College Alcohol Study surveys: 1993–2001. Journal American College Health, 50, 203–217. doi: 10.1080/07448480209595713 [DOI] [PubMed] [Google Scholar]
  76. Witkiewitz K (2011). Predictors of heavy drinking during and following treatment. Psychology of Addictive Behavior, 25, 426–438. doi: 10.1037/a0022889 [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Yarkoni T (2010). Personality in 100,000 words: A large-scale analysis of personality and word use among bloggers. Journal of Research in Personality, 44, 363–373. doi: 10.1016/j.jrp.2010.04.001 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES