Summary. —The Bergen Facebook Addiction Scale (BFAS), initially a pool of 18 items, three reflecting each of the six core elements of addiction (salience, mood modification, tolerance, withdrawal, conflict, and relapse), was constructed and administered to 423 students together with several other standardized self-report scales (Addictive Tendencies Scale, Online Sociability Scale, Facebook Attitude Scale, NEO–FFI, BIS/BAS scales, and Sleep questions).

That item within each of the six addiction elements with the highest corrected item-total correlation was retained in the final scale. The factor structure of the scale was good (RMSEA? =?. 046, CFI? =?. 99) and coefficient alpha was . 83. The 3-week test-retest reliability coefficient was . 82. The scores converged with scores for other scales of Facebook activity. Also, they were positively related to Neuroticism and Extraversion, and negatively related to Conscientiousness. High scores on the new scale were associated with delayed bedtimes and rising times.

Although pathological gambling is the only behavioral addiction, so far, to be assigned status as a formal psychiatric disorder, increasing research has been conducted on other potential behavioral addictions, such as video-game addiction (Fisher, 1994), exercise addiction (Adams & Kirkby, 2002), mobile-phone addiction (Choliz, 2010), online sex addiction (Griffiths, 2012), shopping addiction (Clark & Calleja, 2008), workaholism (Andreassen, Hetland, & Pallesen, 2010), and Internet addiction (Young, 1996; Beard, 2005).

With regard to Internet addiction, it has been questioned whether people become addicted to the platform or to the content of the Internet (Griffiths, 1999). Young (2009) argued that Internet addicts become addicted to different aspects of online use. She differentiates between three subtypes of Internet addicts: excessive gaming, online sexual pre-occupation, and e-mailing/texting (Young, 2009). Social networks are one type of online activity in which e-mailing/texting has been predominant.

Among social networks, Facebook is by far the most popular, with Address correspondence to Cecilie Schou Andreassen, Department of Psychosocial Science, University of Bergen, Christiesgt. 12, 5015 Bergen, Norway or e-mail (cecilie. [email protected] psych. uib. no). 2 The authors thank Annika Hessen and Mathilde Doving for their help with data collection. 1 DOI 10. 2466/02. 09. 18. PR0. 110. 2. 501-517 ISSN 0033-2941 502 C. S. andreassen, et al. over 600 million users worldwide (Carlson, 2011).

In one study, students classified as Internet-addicted used the Internet more for social functions than students considered non-addicted (Kesici & Sahin, 2009). A recently published review article on social networking and addiction suggests that social network sites are predominantly used for maintenance of established offline networks which, for many, are important in terms of academic and professional opportunities. The maintenance of such networks and staying connected are assumed to function as an attraction factor, which might explain why some individuals use social network sites excessively (Kuss & Griffiths, 2011).

Researchers have linked Facebook use to specific individual characteristics. People scoring high on narcissism tend to be more active on social network sites, as social network sites provide an opportunity to present oneself in a favorable way in line with one’s ideal self (Buffardi & Campbell, 2008; Mehdizadeh, 2010). Other studies have focused on the five-factor model of personality, in which personality assessment is based on five main dimensions of Extraversion (e. g. , being outgoing, talkative), Agreeableness (e. g. , being sympathetic and warm), Conscientiousness (e. g. , being organized and prompt), Neuroticism (e.g. , being nervous and moody), and Openness to experience (e. g. , being creative and intellectually oriented) (Wiggins, 1996).

Some previous researchers have reported extraversion as positively related to Internet use in general (Yang & Lester, 2003). In addiction to social media, addictive tendencies have been reported to be positively related to Extraversion and negatively related to Conscientiousness (Wilson, Fornasier, & White, 2010). Also, Correa, Hinsley, and de Zuniga (2010) reported that Extraversion, Neuroticism, and Openness to experience were all positively associated with frequency of social media use.

It has been suggested that extroverts use social media for social enhancement, whereas introverts use it for social compensation, each of which appears to be associated with elevated use (Kuss & Griffiths, 2011). People who score low on Conscientiousness are assumed to use social media as a way of procrastinating, hence, Conscientiousness is assumed to be negatively associated with social media use (Wilson, et al. , 2010). Neuroticism is assumed to be positively related to use of social media as it may be a way of seeking support.

In addition, social media gives people with high scores on Neuroticism more time for contemplation before acting compared to face-to-face interactions (Ehrenberg, Juckes, White, & Walsh, 2008; Ross, Orr, Sisic, Arseneault, Simmering, & Orr, 2009; Correa, et al. , 2010). Addictive behaviors may also be related to individual differences in sensitivity to reward and punishment. According to Gray (1982), one system, the behavioral inhibition system (BIS), is associated with sensitivity to conditioned punishment, whereas another system, the behavioral

Bergen Facebook Addiction Scale 503 approach system (BAS), is associated with sensitivity to conditioned reward. These two systems can be measured using self-report scales, one scale for BIS, and three subscales for BAS: Reward Responsiveness, Drive, and Fun-seeking (Carver & White, 1994). It has been suggested that high behavioral approach system (BAS) sensitivity predisposes to conditions that are characterized by high engagement in approach behaviors, such as alcohol and drug abuse (Franken, Muris, & Georgieva, 2006).

In one study, Internet addiction was positively related to scores on the BIS scale and the BAS Fun-seeking subscale (Yen, Ko, Yen, Chen, & Chen, 2009). Poor and short sleep has, in several studies, been linked to impaired academic performance (Dewald, Meijer, Oort, Kerkhof, & Bogels, 2010). Recently, studies have shown that excessive use of electronic media may delay bedtimes and rising times (Suganuma, Kikuchi, Yanagi, Yamamura, Morishima, Adachi, et al. , 2007; Brunborg, Mentzoni, Molde, Myrseth, Skouveroe, Bjorvatn, et al. , 2011).

These researchers, however, did not consider the content of computer and mobile-phone use. Since Facebook has become one of the most used sites on the Internet, and since poor sleep may be detrimental to the academic performance of students, investigation of whether Facebook addiction, in particular, may be directly associated with sleep habits would be of interest. In relation to assessing Facebook addiction, Wilson, et al. (2010) previously developed the Addictive Tendencies Scale, which has three items reflecting salience, loss of control, and withdrawal.

Although these three aspects have been central in thinking about addictions, in the literature, addiction has involved six core components: (1) salience—the activity dominates thinking and behavior; (2) mood modification—the activity modifies/improves mood; (3) tolerance—increasing amounts of the activity are required to achieve previous effects; (4) withdrawal—the occurrence of unpleasant feelings when the activity is discontinued or suddenly reduced; (5) conflict—the activity causes conflicts in relationships, in work/education, and other activities; and (6) relapse—a tendency to revert to earlier patterns of the activity after abstinence or control (Brown, 1993; Griffiths, 1996, 2005).

In line with this, studies have shown that social-network site use can lead to a variety of negative consequences such as decrease in real-life communities, worsening of academic performance, and relationship problems (Kuss & Griffiths, 2011). As addiction to Facebook may be a specific form of Internet addiction, and since the use of Facebook is increasing very rapidly, there is a need for a psychometrically sound procedure for assessing a possible addiction. Against this background, a Facebook addiction scale (the Bergen Facebook Addiction Scale) with as few items as possible (one reflecting each of the six above-mentioned elements of addiction, ensuring its content va- 504 C. S.

andreassen, et al. lidity) was constructed. A new Facebook addiction scale should correlate highly with measures of similar constructs (convergent validity) and less with measures of more divergent or unrelated constructs (discriminant validity) (Cozby, 2009). The following hypotheses were tested: (1) the Bergen Facebook Addiction Scale (BFAS) will have a unidimensional factor structure with high factor loadings for all items, fit indexes [root mean square error of approximation (RMSEA) and comparative fit index (CFI)] showing good fit with the data and factor loading invariance across sexes; (2) the 3-week test-retest reliability will be high (r? >?.

75); (3) ratings on the BFAS will correlate positively and significantly with scores on other scales of Facebook use (the Addictive Tendencies Scale, as well as scales measuring Facebook attitudes and use, respectively); (4) ratings on the scale will be positively related to those on Neuroticism and Extraversion and negatively related to those on Conscientiousness; (5) ratings on the scale will be positively associated with ratings on the BIS scale and with those on the BAS Funseeking subscale; and (6) the scores on the BFAS will correlate positively and significantly with bedtimes and rising times. Method Participants The sample comprised a total of 423 college students (227 women). Their mean age was 22. 0 yr. (SD? =? 4. 0).

A subsample (n? =? 153, 118 women, 35 men) of these were present at a later lecture and were used for test-retest of the BFAS. The mean age of the retest sample was 21. 3 yr. (SD? =? 4. 1). Procedure Potential items to be included in the Facebook addiction scale were constructed for the six basic components of addiction proposed by Brown (1993) and Griffiths (1996).

Three items for each component were chosen. Wording was similar to that used in the diagnostic criteria for pathological gambling (American Psychiatric Association, 2000) and in the Game Addiction Scale (Lemmens, Valkenburg, & Peter, 2009). These items were included in a self-report questionnaire with additional questions about demography, Facebook activity, personality, and sleep habits. The questionnaire was distributed at undergraduate lectures in psychology at the University of Bergen, Norway, to engineering students at Bergen College, and students at the Royal Norwegian Naval Academy during the spring of 2011. Questionnaire completion took approximately 20 minutes.

No monetary or other material incentives were offered in return for participation. Response rate was 95%. Questionnaires were coded with unique numbers that students were asked to note and keep for later re-administration of some of the questions. They were not informed which questions Bergen Facebook Addiction Scale 505 would be re-administered. Three weeks after the first questionnaire was administered, the 18 items were re-administered to 36. 2% of these undergraduates. Participants were asked to write the unique number code on the questionnaire for administrative use in identifying which students answered questions twice. Measures The Bergen Facebook Addiction Scale (BFAS).

—This scale comprised 18 items, three for each of the six core features of addiction: salience, mood modification, tolerance, withdrawal, conflict, and relapse. Each item is scored on a 5-point scale using anchors of 1: Very rarely and 5: Very often. Higher scores indicate greater Facebook addiction. All 18 original items are listed in Appendix A (p. 516). Cronbach alpha was . 83 in this sample. The Facebook Attitude Scale. —This scale has six items for assessing attitudes toward Facebook. Each item is rated on a 5-point scale with anchors of 1: Strongly disagree and 5: Strongly agree. Higher scores then reflect positive attitudes toward Facebook (Ellison, Steinfield, & Lampe, 2007). Internal consistency (Cronbach alpha) was . 82 in the present study. The Online Sociability Scale.

—This scale comprises five items, each pertaining to frequencies of different uses of Facebook (e. g. , comment on other photographs, sending private messages). Scores are ratings on a 9-point scale using anchors of 1: Less than once per year and 9: More than once daily (Ross, et al. , 2009). High ratings reflect high frequency of Facebook use. Cronbach alpha of this scale was . 63 in the present study. The Addictive Tendencies Scale. —The scale (Wilson, et al. , 2010) has three items representing salience to, loss of control of, and withdrawal from Facebook use. Each item is rated on a 7-point scale, with anchors of 1: Strongly disagree and 7: Strongly agree. High ratings indicate high addictive tendencies.

These items were from previous scales assessing addictive tendencies in use of text messages and instant messaging services (Ehrenberg, et al. , 2008). Cronbach alpha of this scale was . 72 in the present study. The NEO–Five Factor Inventory (NEO–FFI). —This is a short 60-item version of the NEO Personality Inventory–Revised, which provides a brief, comprehensive measure of the domains of the five-factor model of personality: Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness. Each subscale has 12 items rated on a 5-point scale (Costa & McCrae, 1992). Values of Cronbach alpha for the scales in the present study were . 89 (Neuroticism), . 80 (Extraversion), . 74 (Openness), . 71 (Agreeableness), and . 82 (Conscientiousness). The BIS/BAS scales.

—The BIS scale assesses behavioral inhibition using seven items. Focus is on measuring predisposition to avoid threatening or punishing stimuli. The BAS scale of 13 items assesses predisposition to approach appetitive stimuli. There are three subscales: Reward 506 C. S. andreassen, et al. responsiveness (BAS–RR), Drive (BAS–D), and Fun-seeking (BAS–FS). Participants indicate how much they agree with statements on a 4-point scale using anchors of 1: Very false for me and 4: Very true for me (Carver & White, 1994). Internal consistencies (Cronbach alpha) of the scales in the present study were for BIS . 79, BAS–RR . 58, BAS–D . 78, and BAS–FS . 58.

Sleep questions. —Content concerned habitual bedtimes and rising times on weekdays and weekends, respectively. These questions have been used in previous research (Pallesen, Saxvig, Molde, Sorensen, Wilhelmsen-Langeland, & Bjorvatn, 2011) and seem to reflect the circadian rhythm of the participant (Bjorvatn & Pallesen, 2009). High numbers/ scores indicate late bedtimes and rising times. Analysis Item selection. —Of the three items within each of the six core addiction elements, the one with the highest item-total correlation with the sum of ratings for all the other 17 items was retained. These analyses were conducted with PASW statistics, Version 18. 0.

Factor analysis. —A one-factor solution was expected and investigated. The error term of each indicator was assumed to be uncorrelated with each of the others. The CFI and the RMSEA were used as fit indexes. As a rule of thumb, for a model with acceptable fit to the data, these indexes should be ?. 90, respectively, whereas the corresponding values for a good fit would be ?. 95, respectively (Hu & Bentler, 1999). Missing data were excluded pairwise. Pearson correlations among all items are shown in Appendix B (p. 517). Correlations and regression analysis. —All other analyses were conducted using PASW, Version 18. 0, unless explicitly stated otherwise.

To investigate the test-retest reliability of responses to the BFAS, the Pearson product-moment correlation coefficient between ratings from the first administration and the re-administration of the scale was calculated. Score­ Rel CI software was used to calculate the 95%CI for the test-retest correlation (Barnette, 2005). Pearson product-moment correlation coefficients were calculated to investigate the convergent validity between scores on the BFAS and on the Facebook Attitude Scale, the Online Sociability Scale, and the Addictive Tendencies Scale.

A hierarchical multiple regression analysis was conducted to assess how ratings on the BFAS were related to the five-factor model of personality as well as to measures of the behavioral inhibition system and of the behavioral activation system. Participants’ age and sex were entered in the first step.

In the second step, the ratings for the five subscales (Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness) of the NEO Five-Factor Inventory were entered, as well as ratings from the four subscales of the BIS/BAS scales (the Behavioral Inhibition Scale and the Behavioral Approach Scales: Reward Bergen Facebook Addiction Scale 507 Responsiveness, Drive, and Fun-seeking).

Preliminary analyses were conducted to ensure there was no violation of the assumption of normality, linearity, multicollinearity, and homoscedasticity. Pearson product-moment correlations were calculated for the relations of the scores on the BFAS with responses to the sleep questions. Results Factor Structure The corrected item-total correlation coefficients for all initial 18 items are presented in Appendix A. The corrected item-total correlation coefficient of each of the six core addiction elements retained ranged from . 60 to . 73 (see Appendix A, p.

516). The confirmatory factor analysis showed that all standardized loadings of the six indicators on the one-factor solution (? 2/df? =? 1. 84, p? >?. 05) were above . 50 (range? =?. 59 to . 80; see Fig. 1). The RMSEA of the model was 0. 05 (90%CI? =? 0. 00, 0. 08) and the CFI was . 99. Cronbach alpha for the BFAS was . 83 for the whole sample and . 83 for the retest subsample. Comparing a model with no constraints to a model with constraints on the factor loadings across sexes indicated factor loading invariance (?? 2? =? 8. 86, df? =? 5, p? >?. 05). Test-retest Reliability The 3-week test-retest correlation coefficient (n? =? 153) was . 82