**1. Introduction**

As Internet usage has rapidly increased, it has become a part of our lives and is both a medium for providing a wealth of information and an important tool for connecting with others around the globe [1]. The advent of smartphones and easy access to the Internet have brought about many advantages. However, many people have developed concerns about such ease of access and social networking, particularly for adolescents; one concern is Internet addiction. Although not a formal diagnosis, Internet addiction is becoming widely accepted as a problem that may require professional treatment [2]. There is a lack of consensus regarding the definition of Internet addiction, and the term has been used in different studies with different connotations. In general, Internet addiction is defined as excessive use of the Internet accompanied by withdrawal, tolerance, and negative repercussions [3]. For consistency reasons, we utilize the term Internet addiction throughout this paper to represent

a pathological state that occurs due to Internet overuse. A school-based survey for 9th and 10th grade students in seven European countries (Greece, Spain, Poland, Germany, Romania, the Netherlands, and Iceland) revealed that 0.8% to 1.7% of the students were suspected to have Internet addiction using Young's Internet addiction test [4]. A meta-analysis that reviewed studies from 31 nations reported that the global occurrence of Internet addiction was estimated at 6% [5]. However, higher estimates of the problem have been reported in Korea. Reportedly, almost 100% of South Korean adolescents are Internet users [6], and Internet addiction is a serious public health issue among Korean adolescents [7,8]. According to a national survey, adolescents had the highest proportion of high-risk Internet dependence (13.1%), followed by adults (5.8%), and then children (5.0%) [9]. Survey results from different countries also reported that adolescents as a group are the most vulnerable to Internet addiction [10]. Children and adolescents, having grown up in the digital technology era, have been familiar with digital devices and the Internet from an early age. A typical 15-year-old in 2015 would have been using the Internet since age 10 and the hours spent on the Internet is noticeably increasing among Organization for Economic Co-operation and Development countries [11]. Research studies in addiction have indicated that exposure to substance use at an earlier age increases the risk for addiction and risky behaviors later in life [12,13]. This can also be said for Internet use at an early age, and there is a concern that early access to electronic devices could lead to escalated risks for addiction [14]. Also of importance is the contents of the Internet. Different contents of the Internet affect young people to engage in overuse of the Internet. For instance, Greenfield [15] suggests that the most addictive aspects of the Internet are sexual content and computer gaming. Although these content areas are not limited to the Internet, when these are accessed using the Internet, the addictive potential of the contents are known to be amplified. A recent study also reported that people who frequently engage in web surfing experience the same psychological effects as those who are hooked on gambling [16].

Mental health experts believe that Internet addiction may manifest the same troubling effects as substance abuse or gambling disorders [17]. Internet addiction has been associated with various physical and psychological problems, and Internet addiction mostly manifests in adolescents as social withdrawal, loneliness, low motivation, and low educational performance [18].

Many studies have been conducted to identify the factors associated with Internet addiction. Of these, a number of studies have documented personal risk factors, typically psychological factors. These include depression [18–21], anxiety [20,21], aggression [22], impulsivity [23], and low self-esteem [21,24]. Among many personal risk factors of Internet addiction, depression, aggression, and impulsivity are of particular interest due to their association with other clinical subtypes. Depression is related to emotional vulnerability or internalizing problems, and aggression and impulsivity are related to externalizing problems such as attention deficit hyperactivity disorder, which are both known to increase the risk for Internet addiction [25]. There is strong evidence that individuals with higher levels of depression are more vulnerable to developing Internet addiction [19,26–29]. A recent longitudinal study also confirmed that depression is a positive predictor of adolescent Internet addiction [30]. Some studies have identified aggression as a predictor of Internet addiction. In a cross-sectional study of Lebanese adolescents, higher levels of aggression were associated with a higher level of Internet addiction [27]. A similar pattern was witnessed for young people in Taiwan and Korea [22,31]. Impulsivity is another psychological trait that is positively associated with Internet addiction. Several studies have discussed the correlation between high impulsivity and Internet addiction among adolescents [32,33], and a similar finding was reported among young people with internet gaming disorder [34].

Some studies addressed the extent to which interpersonal variables contribute to Internet addiction. They sought to explain Internet addiction in terms of interpersonal difficulties, and confirmed the role of family- and school-related variables in relation to Internet addiction among adolescents. For instance, conflict with parents, family functioning, and family resilience have been identified to influence Internet addiction [35–37]. School-related variables such as teacher's support and attitude toward school life are

also associated with young people's Internet addiction. This is especially true for Korean adolescents, where there is enormous pressure for high academic achievement [38].

Another set of factors that has received relatively less attention is related to the characteristics of the Internet and its influence on addictive behavior. Few studies explicate Internet addiction by understanding what factors of the Internet enhance or reduce a user's excessive form of use. Several researchers have discussed the seductive and gratifying properties specific to the Internet that attract users [39–41]. They claim that certain properties of the Internet are associated with an elevated risk for Internet addiction. For instance, Turkle [42] isolated two dimensions of the Internet that are seductive to the Net generation: the pleasure of control and the perceived fluidity of identity [41]. Several other seductive factors were identified such as self-presentation, diversion, relationship building, and virtual community to name a few [40,41].

Most existing literature on Internet addiction has focused on individual and psychological factors and the proximal environment such as family and peer groups. Extensive meta-analysis or systematic review studies on risk factors of Internet addiction focused mainly on intrapersonal or interpersonal variables [10,31,43], and social environmental factors have received relatively less attention. The studies that have explored environmental factors mostly focused on the proximal environment such as family- and school-related variables [36,38,44]. Only a few studies examined social factors or larger environmental variables (such as exposure to advertisements and accessibility) as they contribute to Internet addiction among young people [45–47]. For instance, in alcohol-related research, factors such as availability of alcohol and exposure to alcohol are known environmental factors that influence young people's drinking behavior [48,49]. It is not difficult to assume that availability and accessibility to the Internet may also affect Internet addiction. However, such factors were infrequently explored in Internet addiction research [46]. Internet addiction is a broader concept that encompasses Internet game addiction. Although this study focuses on Internet addiction in general, we believe that examining the influence of Internet game advertisements as an environmental factor is legitimate given that a large part of Korean adolescents' Internet use is allocated to playing online games [9].

As discussed above, personal variables, family- and school-related variables, characteristics of the Internet, and environmental variables all contribute to the risk of Internet addiction. Effective prevention and intervention of Internet addiction requires a framework that integrates individual- and environmental-level factors. This is well reflected in the public health model or the epidemiologic triangle based on the public health perspective. Originally designed as a model for infectious disease, the public health model considers three areas of causes in understanding and intervening with any health problems [50,51]. One is the external agent that may contain certain destructive potential. Second is the host, who has individual susceptibility. The last factor is the environment that promotes or discourages certain health-related behavior. Applying the model to Internet addiction, the model focuses on the importance of the host (individual characteristics), agent (characteristics of the Internet), and the environment (i.e., social influence and accessibility) in the manifestation of Internet addiction. The public health model posits that any health problem is a result of the interactions among the three factors. Therefore, the development of effective public health measures for prevention and intervention usually requires assessment of all three components and their interactions [51]. To date, few studies have examined Internet addiction from a public health perspective. An awareness of the influence of these three factors may have a significant impact on intervention, prevention, and policy efforts.

The aim of the current study is to fill the gap in addiction literature by examining Internet addiction among adolescents based on a public health framework. Specifically, the study will explore the relationship among individual variables, family/school variables, the Internet characteristics, and environmental factors as they contribute to Internet addiction. The goal is to identify factors that are more influential when all areas are considered within the framework. To achieve this goal, the study utilizes the Internet addiction test (IAT) developed by Young [52,53]. The IAT has been used extensively in many studies to assess Internet addiction in both the clinical and research fields [54]. It has also

been used to identify risk factors [31,43]. However, it should be noted that the IAT is not a diagnostic tool and that categorizing addiction based on the IAT does not define a clinical diagnosis.

The following research questions will be addressed using a representative community sample: (1) Are there any differences between the addict and the non-addict group in regard to psychological factors, family- and school-related factors, perceived Internet characteristics, and environmental factors? (2) Which factors that reflect the agent, host, and environment in a public health model are more important than others in predicting Internet addiction?

#### **2. Materials and Methods**

#### *2.1. Participants*

This study utilized a community-based cross-sectional survey. The survey subjects were junior-high-school students who reported the highest risk of Internet addiction in Korea [9]. A representative sample of 1871 middle (junior-high) school students participated in the study. We randomly selected one school from each of the 56 regions in Seoul and Gyeonggi-do that represented both urban and rural districts. Coeducational schools were selected to ensure equal allocation of genders. All students in one randomly selected classroom from each school were asked to participate in the survey between 1 December and 27 December 2015. Questionnaires were mailed to schools with cooperation from the Ministry of Health and Welfare and the district office of education. The study received approval from the Institutional Review Board. Only students who signed the informed consent participated in the survey. Of the 1871 students who responded to the questionnaire, 1628 were included in the analysis after excluding 243 that had at least one missing value in demographic and Internet use behavior variables. Of the 1628 participants, 52.0% (*n* = 847) were male and 47.0% (*n* = 781) were female, and the mean age was 14.9 (*SD* = 0.34).

#### *2.2. Measures*

#### 2.2.1. Internet Addiction

Young's 20-item Internet addiction test was used to measure Internet addiction [52]. Each item was measured on a five-point Likert scale. Responses for each item were added to obtain a final score. Many years after the initial development of the IAT, Young proposed a new cut-off point [53]. People who score more than 50 on the IAT are thought to be experiencing frequent problems because of Internet use. For those who score above 80, Internet use is thought to be causing significant problems with their lives. Because the purpose of this study is to examine risk factors associated with the higher risk of Internet addiction, we used the addiction category based on the IAT rather than the overall score. Young proposed four groups of Internet addiction: normal range: 0–30; mild: 31–49; moderate: 50–79; and severe: 80–100. In order to focus on more severe cases, the normal range and mild addiction (low risk) group were combined into one group. Therefore, respondents were categorized into three groups: normal use: 0–49; moderate addiction: 50–79; and severe addiction: 80–100. The IAT in this study held a good internal consistency with a Cronbach's alpha (∝) of 0.94.

#### 2.2.2. Psychological Variables

The tool used in the study was developed and validated by the National Information Society Agency to identify Internet addiction and coexisting disorders in a national survey on Internet addiction [55]. Each scale consisted of eight questions measured on a four-point Likert scale ranging from one (not at all) to four (very often). Scores were summed for each of the three features with a higher score reflecting a greater degree of impulsivity (e.g., "act on spur of the moment"), depression (e.g., "feeling hopeless"), and aggressiveness (e.g., "have trouble controlling temper"). All three measures possessed good internal consistency (the Cronbach's alpha was 0.86 for impulsivity, 0.91 for depression, and 0.91 for aggressiveness).

#### 2.2.3. Family and School Factors

The Family Cohesion and Adaptability Evaluation Scales (FACES III) was used to measure family relationship. This scale was originally developed by Olson and colleagues [56], and a Korean version of the scale [57] was utilized in the current study. The family cohesion scale in the FACES III consists of ten questions on a five-point Likert scale that measure emotional ties, family support, time spent together, and interest in leisure time with family. Higher scores reflect higher family cohesion, and the Cronbach's alpha (∝) was 0.96.

A school adaptation scale developed for Korean students [58] was employed to examine school-related factors. The scale evaluates the degree of adaptation to the school environment, which includes questions about the relationship with teachers, the relationship with schoolmates, attitude toward academic activities, and school participation. The tool consists of four subscales and 14 questions on a five-point Likert scale. Higher total scores indicate a positive assessment of school life. The Cronbach's alpha (∝) for the subscales were acceptable: 0.94 for relationship with teachers, 0.90 for relationship with schoolmates, 0.62 for attitude toward academic activities, and 0.61 for school participation.

#### 2.2.4. Perceived Internet Characteristics

The agent characteristics in the public health model refer to the traits of the medium in relation to a given problem. In this study, Internet characteristics are defined as the properties of the Internet perceived by individuals in relation to Internet use. Perceived Internet characteristics were measured by the Internet characteristics scale developed by Chung et al. [59] that was designed to identify the characteristics of the Internet perceived by adolescents. It consists of eight factors with 27 items including entertainment, interpersonal relationship, accessibility, profitability, anonymity, immersion, competition, and escape. Based on a previous study that identified three factors to predict young people's Internet addiction [59], our study included three factors as perceived Internet characteristics: namely, interpersonal relationship (four items; e.g., "the Internet provides opportunities to meet with diverse people"), anonymity (four items; e.g., "People can attract attention without revealing themselves"), and entertainment/pleasure (five items; e.g., "the Internet provides entertainment"). Each item was measured using a five-point Likert scale, from strongly disagree to strongly agree. The interpersonal relationship characteristic of the Internet refers to the perception of the Internet as grounds for building relationships; anonymity refers to the perception of the Internet as a site where people can act freely without identifying themselves; entertainment refers to the perception that the Internet provides fun and excitement. The measures for interpersonal relationships, anonymity, and entertainment showed good internal consistency (∝ = 0.85, 0.86, and 0.87, respectively).

#### 2.2.5. Environmental Factors

Social environmental factors consisted of accessibility to PC cafés and exposure to Internet game advertisements. Previous studies have identified access to PC cafés as a risk factor for Internet addiction among Korean adolescents [47]. PC cafés, a unique feature in Korea, are well-equipped with advanced computers and are optimized for playing Internet games, which attracts many young people. We measured the accessibility to PC cafés using the scale reconstructed by Nam and Lee [60] that consists of a three-item (i.e., There is a nearby PC café for Internet use; I have easy access to PC cafés; there are many PC cafés near my home or school) five-point Likert scale. A higher total score indicated higher accessibility to PC cafés. The Cronbach's alpha (∝) was 0.85.

We also included exposure to Internet game advertisements as a major environmental factor. Exposure to Internet game advertisements consists of exposure and acceptability. To our understanding, there is no known instrument with which to measure exposure to Internet game advertisements. We used the tool that examined exposure to alcohol advertisements [61,62], except we restructured it to fit Internet game advertisement exposure. Two questions about respondent's exposure to Internet

game advertisements (i.e., How often do you see Internet game advertisements? In the past month, how often did you see Internet game advertisements in the medium that you frequently use?) and the frequency of exposure were measured on a four-point Likert scale. The Cronbach's alpha (∝) was 0.88.

The advertisement acceptability questions used to determine alcohol use among youths were adopted for the current study [62,63]. However, for the purposes of our study, the questions were modified to assess the acceptability of Internet game advertisements. Two questions were asked: the number of game ads they had seen in various media over the past month and the number of memorable ads that they remember.

### *2.3. Statistical Analysis*

To verify the differences between sociodemographic characteristics according to Internet use, a chi-square test (χ2) was performed. Between-group comparisons of the variables based on the public health model according to the presence of Internet addiction were analyzed using analysis of variance. A logistic regression analysis was utilized to identify the factors associated with Internet addiction. All statistical analyses were performed using SPSS version 20.0 (IBM Corp. Released, Armonk, NY, USA).

#### **3. Results**

### *3.1. Sociodemographic Characteristics*

Adolescents' demographic profiles were assessed using four survey items, namely gender, residing region, living with parents, and parents' working status (Table 1). Based on the Internet addiction test cut-off scores, 455 (27.9%) were identified as normal Internet users, 1067 (65.5%) belonged to the moderate-addiction or addiction-risk group, and 106 (6.5%) belonged to the severe addiction group. The group differences according to sociodemographic characteristics showed that 6.6% male and 6.4% female students were in the addiction group, and 66.7% male and 64.3% female students were deemed at risk for addiction (moderate addiction). There were no significant differences between three groups in regard to sociodemographic characteristics.


