Since Apple launched the iPhone in 2007, smartphones have become ubiquitous and pervasive in people’s daily lives. The number of smartphone users worldwide already surpassed three billion in 2019 and is forecasted to reach 3.8 billion in 2021, which means 48.46% of the world’s population own smartphones (Statista, 2021). The rapid diffusion of smartphones has been attributed to their portability and myriad of novel functions to enhance productivity, entertainment, and social relationships (Fischer-Grote et al., 2019; Tangmunkongvorakul et al., 2020). This development has also resulted in people’s excessive smartphone use that may interfere with their daily lives as well as physical and mental health (Herrero et al., 2019; Yu & Sussman, 2020).

One of the most concerning consequences of excessive use of smartphones is the possibility of developing so-called smartphone addiction in the form of frequent and habitual use (S.-H. Jeong et al., 2016; Oulasvirta et al., 2012). The concept of addiction includes such symptoms as tolerance, withdrawal, dependence, social problems, and loss of control over not only substances but also behaviors like gambling and online gaming (Yu & Sussman, 2020). From the late 1990s, applying the concept of addiction to the use of media such as the Internet, researchers have argued that media activities may lead to problematic symptoms of addiction such as obsession, positive perception, tolerance, withdrawal, and disturbance of daily lives (Griffths, 1999; Shaw & Black, 2008; Young, 1998).

The negative consequences of smartphone addiction, which have been found in previous research, suggest that addiction to smartphones may be a pathological disorder like gaming disorder, which the World Health Organization (WHO) decided to categorize as a disease (6C51) in May 2019. The symptoms of smartphone addiction have been listed as loss of control, compulsive behavior, functional impairment, withdrawal, and tolerance (Kwon et al., 2013; Lin et al., 2017; Lopez-Fernandez et al., 2014). According to the prior reports, furthermore, smartphone addiction-oriented consequences are observed across physical, psychological, and social dimensions. For example, physical problems include altered craniocervical posture and mobility (Kee et al., 2016), insomnia or lower sleep quality (Bruni et al., 2015; Demirci et al., 2015; J. E. Lee et al., 2017; Soni et al., 2017), pain in neck (AlAbdulwahab et al., 2017), eye soreness or dry eye syndrome (Moon et al., 2014), and lower levels of physical activity (Haug et al., 2015). Psychological threats include anxiety and depression (Demirci et al., 2015; Elhai et al., 2017), stress (Chiu, 2014; Samaha & Hawi, 2016), lower levels of emotional intelligence (Cho & Lee, 2017), and loneliness (Bian & Leung, 2015; Enez Darcin et al., 2016). Social concerns are related to disturbance of work/study habits (Samaha & Hawi, 2016) and weakening of social relationships (Bian & Leung, 2015).

Moreover, a noteworthy point is that these psychological, physical, and social consequences generated by smartphone addiction are particularly discovered in younger populations, especially teenage groups (Tangmunkongvorakul et al., 2020; Yu & Sussman, 2020). The prevalence of smartphone addiction among children and young people ranged from 10% to 30% across studies and countries with a median of 23.3% (Sohn et al., 2019). In Korea, the recent prevalence has been reported between 13.5% and 36% (Cha & Seo, 2018; Jo et al., 2018; C. Lee & Lee, 2017; H. J. Lee et al., 2017; J. E. Lee et al., 2017). These increasing rates of smartphone addiction drive scholars to understand the types of risks and protective factors on adolescents’ smartphone addiction.

With respect to the smartphone addition of younger generations, most research has focused on (a) uncovering the symptoms of smartphone addiction, (b) resulting negative consequences, and (c) risk factors of smartphone addiction—78% of recent studies on smartphone addiction focus on elementary to college students (Yu & Sussman, 2020). However, the literature has not clearly explained how to prevent smartphone addiction among adolescents due to the lack of understanding of protective factors. Furthermore, previous research has failed to specifically identify discriminatory factors among different age groups in adolescence or to consider various developmental stages in adolescence and subsequent environmental differentials.

Given this background, this study investigates prominent factors that could treat and/or prevent youth smartphone addiction by school levels: (1) elementary, (2) middle, and (3) high school. Considering the distinctive characteristics of adolescence in which adolescents are substantially influenced by adults’ parenting, protection, and education, this study investigates which protective factors are associated with adolescents’ smartphone addiction. The findings of the current study will provide meaningful assets for school administrators and government officials to create an effective prevention system or tools for the mitigation of younger generations’ smartphone addition.

# Literature Review

## Environmental and Developmental Characteristics of Adolescence in Korea

Adolescents are more adaptable to new information or technology than adults, while they are vulnerable to negative media influences due to lack of social skills or self-control and are especially at high risk of media addiction (Griffiths, 2000; Rheingold, 1993). In Korea, the spatial infrastructure of leisure activities is not sufficiently provided for adolescents (Yoon et al., 2014). Korean adolescents have received a lot of stress from their academic tasks due to college entrance exams (H. K. Shin, 2002). Thus, psychological stress and poor leisure environments in Korea drive adolescents, who are too immature to cope with stress, to frequently rely on their smartphones (Ebata & Moos, 1991; J. S. Jeong & Lee, 2020).

Previous studies have consistently suggested that psychological pressures on children and adolescents’ academic performances are closely related to their tendency to escape reality and rely on virtual space on the Internet (Young, 1998), and the higher the psychological burden from studying, the higher the level of smartphone addiction. For example, Lee & Lee (2012) found that elementary school students who feel stressed are vulnerable to smartphone addiction. For Korean middle and high school students, too, a positive relationship was found between academic stress and smartphone addiction (Seo, 2018). These findings suggest that if the excessive academic stress experienced by teenagers cannot be properly resolved in reality, teenagers may use smartphones as a stress reliever, which could eventually lead to addiction by being used as an alternative means of leisure to escape reality.

A noteworthy point is that adolescents, who are conceptually and legally unclearly defined, embrace a wide range of age groups—elementary, middle, and high school students. This study notes that adolescents’ smartphone addiction is influenced by psychological and environmental factors and is varied depending on different stages of adolescents’ development. For example, Oh and Lee (2014) found that the factors/effects of smart media addiction are different across developmental stages in adolescence, which were divided into the school levels. The literature highlights the meaningfulness of exploring discriminative factors on smartphone addition, regardless of risk factors or protective factors, in consideration of different developmental stages.

## Adolescent’s Demographics and Smartphone-Usage Factors

Among demographic factors affecting adolescents’ smartphone addiction, gender seems to be the most prominent. Specifically, female adolescents are more prone to smartphone addiction than male adolescents. For example, Spanish female teenagers are more prone to smartphone addiction compared to male teenagers (Sánchez-Martínez & Otero, 2009); this was also found in a study on teenagers in Hong Kong (Leung, 2008). This is the case for the Korean context, given that females are more vulnerable to smartphone addiction (S.-H. Jeong et al., 2016; S.-G. Kim et al., 2019; C. Lee & Lee, 2017).

Prior studies found that younger smartphone users experience more negative problems related to smartphone addiction than older generations (Jo et al., 2018; S.-G. Kim et al., 2019; C. Lee & Lee, 2017; J. E. Lee et al., 2017). Researchers note that smartphone addiction might be attributed to minors’ difficulties in self-regulation (van Deursen et al., 2015) and immature control competencies (Chambers et al., 2003). However, scholarly findings are mixed: smartphone addiction was found to be more prevalent among younger students with a Swiss sample (Haug et al., 2015), but the association between age and smartphone addiction was not found in United States and German adolescents (Barnes et al., 2019; Randler et al., 2016). Some found older adolescents to be at a higher risk (Fırat et al., 2018; E. J. Lee & Ogbolu, 2018) and others found no predictive value of age (Ayar et al., 2017; H. Lee et al., 2016).

According to the Korean literature, the cases of adolescents’ smartphone addiction have been frequently discovered in elementary, middle, and high school. A recent national survey on smartphone overdependence in Korea revealed that the prevalence of high risk and latent risk was the highest in middle school students (34.7%), followed by high school students (29.4%), and elementary school students (24.4%), although these teenagers are more overdependent on smartphones compared to the other age categories (Ministry of Science and ICT & National Information Society Agency, 2019). Since students in upper elementary school gradually begin to become addicted to using smartphones, the level of smartphone addiction increases as they move to higher levels of education (H. Shin & Jeong, 2018). This notion suggests that the school level, which reflects the psychological and environmental characteristics of the developmental stage, may be a noteworthy variable in explaining smartphone addiction among adolescents.

With respect to the effects of economic status on adolescents’ smartphone addiction, some of the previous studies have revealed a positive relationship between household income and excessive use of smartphone, which boosts monthly phone bills (Nikken & Schols, 2015; Sánchez-Martínez & Otero, 2009). However, scholarly findings are mixed: some studies suggest that teenagers with low socio-economic status are more likely to rely on smartphones in their everyday lives (Chang et al., 2019; J. Lee & Kim, 2020). By contrast, other studies indicate that there is no significant relationship between a family’s economic status and smartphone addiction among children and adolescents (Cho & Lee, 2017; Suk & Ku, 2016).

The present study also notes that the economic situation and cultural characteristics are different across types of local communities, classified as (a) metropolitan (urban), (b) suburban, and (c) rural areas. In addition, extent to which adolescents use smartphones in daily life could be different across types of local communities. The literature has not clarified the relationship between types of local community and adolescents’ smartphone use/addiction, unlike prior research that has focused on game addiction in Korean adolescents. According to the literature, the percentage of adolescents’ risky game addiction is the highest in rural areas, such as small towns and villages, followed by suburban areas like small and medium-sized cities and metropolitan communities such as larger cities (Korea Creative Content Agency, 2019a, 2019b). The finding supports the notion that adolescents’ smartphone addiction could be higher in rural communities.

Among various factors that influence smartphone usage, time spent using smartphones is positively associated with addiction (Cha & Seo, 2018; Haug et al., 2015; H. Lee et al., 2016). In particular, addiction was found to be related to time spent using social media applications (Cha & Seo, 2018; S.-H. Jeong et al., 2016; H. Lee et al., 2016; S.-J. Lee et al., 2016), online chatting services (H. J. Lee et al., 2017), and game applications (Cha & Seo, 2018) on smartphones.

## Adolescent’s Interpersonal Relationships with Friends, Parents, and Teachers

Adolescents’ interpersonal relationships have been considered important factors that can influence their smartphone addiction. The problem-behavior theory highlights that those who do not receive social support and encounter conflicts with their parents/friends tend to experience emotional and behavioral difficulties (Karaman, 2013; Rotter, 1954). In particular, perceived social support is a positive resource gained by establishing mutual relationships with others (Cohen & Hoberman, 1983), and adolescents have an internal tendency to seek social support from others due to the nature of their social growth environment. Therefore, adolescents who lack social support are likely to use smartphones to replace real interpersonal relationships with those in the virtual world and compensate for their psychological losses due to lack of social support (Billieux et al., 2015; J.-H. Kim, 2017). Researchers argue that if adolescents are alienated from their interpersonal relationships or have weak ties, the likelihood of smartphone addiction increases, and the vicious cycle of worsening interpersonal relationships can be repeated (Billieux et al., 2015; Herrero et al., 2019; Ihm, 2018; J.-H. Kim, 2017).

Based on this background, this study seeks to examine the effects of adolescents’ relationships with friends, parents, and school/academic institute teachers as the protective factors on their smartphone addiction. For adolescents, their relationships with friends, parents, and teachers are important factors that directly influence their emotional and social development, so the effects of these relationships in which adolescents spend most of their daily lives at home, school, and academic institutes are bound to be significant. According to the results of a four-year panel study of 3,449 adolescents from the second year of middle school to the second year of high school in Korea, for example, adolescents’ perceived peer attachment, parent attachment, and teacher attachment were revealed to positively affect their self-control, the ability to properly manage and control oneself in a given situation (Y. M. Kim & Lim, 2014).

The findings of previous studies that examined the effects of adolescents’ interpersonal relationships on smartphone addiction are as follows: First, a larger network of friends positively affects the possibility of smartphone addiction (Gallimberti et al., 2016). On the other hand, positive peer relationships (Woo, 2013), friendship satisfaction (Bae, 2015), supportive social networks (Ihm, 2018), and friendship quality (H.-J. Kim et al., 2018) have been regarded as predictors that prevent smartphone addiction. In fact, Kim, Lee, and Moon (2016) conducted a study with middle and high school students in Korea and found that peer relationships had a greater impact on smartphone addiction than the characteristics of the smartphone or personal characteristics. Choi (2020) suggests that close and good relationships with friends, rather than having a direct impact on smartphone addiction, have indirect effects through increased self-esteem and decreased social isolation.

The results of the previous research reviewed above suggest that adolescents’ mutual relationships with their friends, parents, and school teachers could play key roles for reducing their smartphone addiction. Based on the literature, this study proposes the following research questions to scrutinize the effects of adolescents’ interpersonal relationships, especially with adults (parents and school/private institution teachers), on smartphone addiction by levels of school in Korea.

RQ1a: Which factors influence adolescents’ smartphone addiction by their school levels (elementary/middle/high school)?

RQ2: What are the effects of adolescents’ relationships with adults (parents and school/institute teachers) on their smartphone addiction?

RQ2a: How do adolescents’ relationships with their parents and school/private institution teachers influence their smartphone addiction by their school levels (elementary/middle/high school)?

# Method

## Research Design and Participants

To explore the research questions suggested above, this study utilized the data of the 2019 Korean Children & Adolescents’ Happiness Index Research gathered by the Korean Foundation of Bang Jung-Hwan. The Korean Children & Adolescents’ Happiness Index Research, following the model of the United Nation’s UNICEF’s research of children and adolescents’ well-being worldwide, has been conducted every year since 2009 in order to compare Korean children and adolescents’ happiness with those of the European Union’s OECD nations in the realm of material well-being, health and welfare, education, relationships with family and friends, behaviors and security, and subjective well-being.

The data were gathered from a total of 7,454 adolescents, including 2,360 elementary school students (4th - 6th grade), 2,296 middle school students, and 2,798 high school students across the entire country, except Jeju Island. These participants were selected by class unit following the PPS (Probability Proportional to Size) method depending on the participant’s gender, residential district, types of local community, and level of school on the basis of the 2018 basic education statistics provided by the municipal and provincial education offices. The data were gathered from March 7 to April 6 in 2019 through a self-administered survey mode. The demographic composition of the participants is summarized in Table 1.

Table 1. Demographic Composition of the Participants (%)
 Demographics Elementary Middle High Total Gender Male 1,186(50.3) 1,138(49.6) 1,413(50.5) 3,737(50.1) Female 1,174(49.8) 1,158(50.4) 1,385(49.5) 3,717(49.9) Residential District Seoul/Gyeonggi 1,073(45.5) 1,065(46.4) 1,262(45.1) 3,400(45.6) Gangwon 69(2.9) 83(3.6) 114(4.1) 266(3.6) Chungchung 312(13.1) 301(13.1) 374(13.4) 987(13.2) Gyungsang 650(25.6) 587(25.6) 718(25.7) 1,955(26.2) Jeolla 256(10.9) 260(11.3) 330(11.8) 846(11.4) Types of local community Metropolitan 954(40.4) 1,011(44.0) 1,217(43.5) 3,182(42.7) Urban 1,073(43.9) 993(43.3) 1,301(46.5) 3,331(44.7) Rural 369(15.6) 292(12.7) 280(10.1) 941(12.6) School Level Elementary School 4th 757(32.1) 2,360(31.7) 5th 870(36.9) 6th 733(31.1) Middle School 1st 678(29.5) 2,296(30.8) 2nd 770(33.5) 3rd 848(36.9) High School 1st 830(29.7) 2,798(37.5) 2nd 915(32.7) 3rd 1,053(37.6) Total 2,360(31.7) 2,296(30.8) 2,798(37.5) 7,454(100)

## Measures

### Dependent Variable

Smartphone addiction. The level of smartphone addiction, the dependent variable of this study, was measured by utilizing six items related to the use of smartphones. Participants were asked to indicate the degree to which they agree with each item on a four-point Likert scale ranging from 1 = strongly disagree to 4 = strongly agree. Responses to the six items were averaged to generate each participant’s level of smartphone addiction, according to which a higher score refers to a higher level of smartphone addiction.

Table 2 shows measurement items as well as mean values, standard deviations (SD), and inter-item reliabilities of the scale by school levels. Comparing the average level of smartphone addiction by school level, high school students showed the highest level of smartphone addiction (M = 1.92, SD = .55), followed by middle school students (M = 1.82, SD = .53) and elementary school students (M = 1.63, SD = .54), and the mean differences were statistically significant, F(2, 6,944) = 175.30, p < .001. Scheffe’s post-hoc test indicated that the mean differences between elementary school and middle school (M = -.19, SE = .02, p < .001) as well as between middle school and high school (M = -.10, SE = .02, p < .001) were statistically significant. The inter-item reliabilities of the scale for the school levels were all acceptable above .70.

Table 2. Mean (SD) and Reliability of the Level of Smartphone Addiction Scale by School Level
 Items Mean (SD) / Cronbach’s $\alpha$ Elementary Middle High “My school grades have fallen because of excessive use of a smartphone.” 1.63 (.54)/ .79 1.82 (.53)/ .77 1.92 (.55)/ .79 “It is more fun to use a smartphone than to be with family or friends.” “If smartphones become unusable, it will be hard to bear.” “It's hard to do the planned work (study, homework, attending classes, etc.) because of the use of a smartphone.” “I feel restless and nervous without a smartphone.”

### Independent Variables

Demographic factors. The present study classified the factors that affect adolescents’ smartphone addiction into four dimensions including (a) demographic factors, (b) smartphone use, (c) friendship, and (d) relationships with adults. First, the demographic factors include four variables: (a) participant’s gender, (b) school year, (c) type of local community, and (d) economic status of household. Gender was dummy coded ‘0’ for male and ‘1’ for female. Participant’s school year was utilized for the purpose of identifying the age effects within each of the school levels. Types of local community were categorized by following the administrative district in which participants currently reside into ‘1’ for rural, ‘2’ for urban, and ‘3’ for metropolitan. The economic status of the household was measured by participants’ self-reports on a six-point Likert scale (1 = extremely low, 6 = extremely high).

Smartphone use. Daily time (minutes) spent using a smartphone was measured for smartphone use with the question of “How much time (minutes) do you usually spend using your smartphone per day?” The mean differences among three school levels were found to be statistically significant, F(2, 6,880) = 36.46, p < .001, in order of high school (M = 208.75, SD = 143.63), middle school (M = 186.80, SD = 131.81), and elementary school (M = 174.87, SD = 132.34). Scheffe’s post hoc test revealed that the mean differences between elementary school and middle school (M = -11.93, SE = 4.10, p = .015) as well as between middle school and high school (M = -21.95, SE = 3.94, p < .001) were statistically significant.

Friendship. Friendship factors include number of close friends as a quantitative aspect of friendship and the level of friendship as a qualitative aspect of friendship. The number of close friends, which was measured with an open-ended question asking “How many close friends do you have?” was the most for elementary school students (M = 9.41, SD = 17.26), followed by middle school students (M = 8.68, SD = 13.79) and high school students (M = 7.33, SD = 18.32) in order. The mean differences were statistically significant, F(2, 7,382) = 10.25, p < .001. The results of Scheffe’s post hoc test indicated that the mean differences in the number of close friends were statistically significant between elementary school and high school (M = 2.08, SE =.47, p < .001) as well as middle school and high school (M = 1.35, SE = .47, p = .017).

In the case of the level of friendship, the participants were asked about the quality of their mutual friendships using six items on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). The responses were averaged after reverse coding (see Table 3).

Table 3. Mean (SD) and Reliability of the Level of Friendship Scale by School Level
 Items Mean (SD) / Cronbach’s $\alpha$ Elementary Middle High “I feel ashamed or foolish when I talk to my friends about my problems.” (R) 4.06 (.68)/ .70 4.06 (.66)/ .74 4.11 (.66)/ .75 “I wish my friends are not the current ones but others.” (R) “I feel lonely even when I am with my friends.” (R) “I sometimes get angry because of my friends.” (R) “When I'm upset, my friends tend to not take it seriously.” (R) “My friends understand me.”

Note. (R) Reverse coded

Table 3 presents mean scores with standard deviations and inter-item reliabilities of the scale by school levels. As descriptive statistics, high school students’ level of friendship (M = 4.11, SD =.66) was higher than those of elementary (M = 4.06, SD =.68) and middle school students (M = 4.06, SD =.66), and the mean differences were statistically significant, F(2, 7,136) = 3.10, p = .045. The inter-item reliabilities of the scale for the school levels were all acceptable (above .70).

Relationships with adults. Finally, the main predictors of this study, the levels of relationships with adults, were measured through participants’ subjective assessment of the quality of their relationships with adults including their father, mother, and school/private institution teachers. Participants responded to two items for each adult such as “I get along with my father well,” and “I can talk to my father about my troubles on any matter,” on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).

Table 4 shows the mean scores for each adult by school level. The levels of relationships with parents (father and mother) were the highest for elementary school students followed by middle and high school students, showing a tendency by which the higher level of school (or the older adolescents get), the lower level of relationships with parents. On the contrary, there was a tendency that at higher levels of school (as adolescents get older), the levels of relationships with school and institute teachers go up. On all the school levels, the level of relationship with mother was the highest followed by the level of relationship with father for elementary and middle school students. However, among high school students, the level of relationship with father was the lowest. In terms of the relationships with teachers, the levels of relationship with private institution teachers were higher than those with school teachers on all of the school levels.

Table 4. Mean Scores (SD) for the Levels of Relationship with Adults by School Level
 Adults Elementary Middle High ANOVA Results Father 4.16 (.94) 3.93 (1.01) 3.80 (1.06) F(2, 7,050) = 76.57, p < .001 Mother 4.50 (.75) 4.29 (.86) 4.25 (.86) F(2, 7,172) = 60.39, p < .001 School Teacher 3.62 (1.03) 3.70 (1.02) 3.83 (.98) F(2, 7,440) = 26.26, p < .001 Institute Teacher 3.66 (1.01) 3.77 (.07) 3.93 (.94) F(2, 5,298) = 34.32, p < .001

# Results

To provide answers to the research questions presented above, a series of hierarchical regression analyses were conducted for each of the school levels. In the first and second models of hierarchical regression, four demographic factors (gender, school year, type of local community, and economic status of the household) and the smartphone usage factor (time spent in using smartphone per day) were included as control variables. Interpersonal relationship factors, which were divided into friendship factors (number of friends, quality of friendship) and adult-relationship factors (quality of relationship with father, mother, school/private institution teachers) were added in turn to the total four-step model of hierarchical regression analysis.

First, the results of regression analysis for the factors of elementary school students’ smartphone addiction are shown in Table 5. The first model with four demographic factors accounted for 4.9% of the total variance of elementary school students’ smartphone addiction (F(4, 1,503) = 20.32, p < .001). For elementary school students, gender, type of local community, and economic status of household were identified as significant factors affecting the level of smartphone addiction. Male students (β = -.137, t = -5.43, p < .001), students in less populated local communities (β = -.096, t = -3.79, p < .001), and those with lower economic status of household (β = -.159, t = -6.30, p < .001) had higher levels of smartphone addiction.

Table 5. Factors Affecting Elementary School Students’ Smartphone Addiction
 Factors Model 1 Model 2 Model 3 Model 4 β t β t β t β t Demographic Gender(M=0) School Year Local Community Economic Status -.137  .004 -.096 -.159 -5.43***      .15 -3.79*** -6.30*** -.106 -.044 -.066 -.133 -4.58*** -1.90 -2.86** -5.73*** -.111 -.040 -.051 -.090 -4.96*** -1.79 -2.30* -3.97** -.107 -.037 -.039 -.071 -4.84*** -1.70 -1.77 -3.15** Usage Time .395 16.87*** .363 16.01*** .345 15.28*** Friends Number Quality -.026 -.264 -1.17 -11.65*** -.010 -.223 -0.46 -9.56*** Adults Father Mother S. Teacher I. Teacher -.010 -.107 -.075 -.019 -0.36 -4.15*** -3.09** -0.78 R2(adjusted R2) .051(.049) .202(.200) .270(.267) .291(.286) $\Delta$R2 .051*** .151*** .068*** .021***

Note. *p < .05, **p < .01, ***p < .001.

After controlling for demographic variables, the second model with the average amount of time using smartphone per day explained a total of 20.0% (F(5, 1,502) = 76.26, p <001) by increasing 15.1% of variance in elementary school students’ smartphone overuse (△F(1, 1,502) = 284.69, p < .001), which indicates that the more time spent using smartphones, the higher level of smartphone addiction.

The third model, adding the number of friends and the level of friendship quality, accounted for 26.7% of the total variance of dependent variable at the elementary school level (F(7, 1,500) = 79.33, p < .001), and the increased explained variance of 6.4% over Model 2 was statistically significant (△F(2, 1,500) = 69.59 p <.001). However, while the number of friends in the quantitative dimension of friendship did not significantly affect smartphone addiction (β = -.026, t = -1.17, n.s.), the level of friendship quality was found to function as a negative factor for smartphone addiction (β = -.264, t = -11.65, p < .001). In other words, in the case of elementary school students, regardless of the number of friends they interacted with, their good relationships with friends tended to decrease the level of smartphone addiction.

The final Model 4, including additionally four types of adult relationship factors (father, mother, school teacher, and institute teacher) accounted for 28.6% of the total variance of smartphone addiction (F(11, 1,496) = 55.87, p < .001). The adult relationship variables increased 2.0% of the variance explained compared to Model 3, which was statistically significant (△F(4, 1,496) = 11.08, p < .001). Among the adult relationship factors, relationships with father (β = -.010, t = -0.36, n.s.) and institute teachers (β = -.019, t = -0.78, n.s.) had no significant impact on smartphone addiction in elementary school students, while relationships with their mothers (β = -.107, t = -4.15, p < .001) and school teachers (β = -.075, t = -3.09, p = .002) were found to have negative effects on smartphone addiction as statistically significant factors. In other words, elementary school students’ risk of smartphone addiction tended to be reduced if they had a better relationship with their mothers and schoolteachers.

Next, the effects of predictors on smartphone addiction among middle school students were analyzed in the same way as the elementary school group. The results are summarized in Table 6. The first-phase model with demographic and sociological factors, which were control variables, accounted for 5.9% of the game addiction variance in middle school students (F(4, 1,406) = 23.30, p < .001). For middle school students, gender, school year, and economic status of household were identified as significant factors for smartphone addiction. Specifically, in female students rather than male students (β = .153, t = 5.91, p < .001), those in higher the school years (β = .129, t = 4.98, p < .001), and those with lower household incomes (β = -.112, t = -4.28 p < .001), the level of smartphone addiction tended to be higher. Average smartphone usage time per day included in the second model was found to be a significant factor affecting positively middle school students’ smartphone addiction (β = .326, t = 12.78, p < .001). The second model added 15.9% of variance to the first model (△F(1, 1,405) = 163.31, p < .001), explaining a total of 15.7% of variance in middle school students’ smartphone addiction (F(5, 1,405) = 53.45, p < .001).

Table 6. Factors Affecting Middle School Students’ Smartphone Addiction
 Factors Model 1 Model 2 Model 3 Model 4 β t β t β t β t Demographic Gender(M=0) School Year Local Community Economic Status .153  .129 -.035 -.112 5.91***  4.98*** -1.33 -4.28*** .085  .080 -.029 -.084 3.37***    3.23**  -1.17 - 3.37*** .101  .088 -.011 -.059 4.25***     3.78***   -0.45   -2.47* .082  .087 -.017 -.037 3.43***    3.72***  -0.75  -1.55 Usage Time .326 12.78*** .304 12.61*** .297 12.43*** Friends Numbers Quality .057 -.319 2.43* -13.72*** .058 -.285 2.52* -11.79*** Adults Father Mother S. Teacher I. Teacher -.128  .036 -.028 -.037 -4.46***      1.29    -1.10    -1.48 R2(adjusted R2) .062(.059) .160(.157) .260(.257) .277(.271) $\Delta$R2 .062*** .098*** .101*** .017***

Note. *p < .05, **p < .01, ***p < .001.

The friendship factors added to the Model 3 were found to increase 10.1% of explained variance of the dependent variable (△F(2, 1,403) = 95.42, p < .001). While the number of friends as the quantitative dimension of friendship was found to have positive effects (β = .057, t = 2.43, p = .015), the quality of relationship with friends as the qualitative dimension of friendship was found to have negative effects on middle school students’ smartphone addiction (β = -.319, t = -13.72, p < .001). In other words, in the case of middle school students, the more friends they had, the their level of smartphone addiction tended to increase, but the higher the quality of their relationships with friends, the lower their level of smartphone addiction.

The four factors of adult relationships added to the final model increased 1.7% of explained variance (△F(4, 1,399) = 8.09, p < .001), which resulted in a total of 27.1% of explained variance in middle school students’ smartphone addiction(F (11, 1,399) = 48.76, p < .001). For middle school students, the quality of their relationship with their father was found to be the only significant factor that affected their smartphone addiction in a negative way (β = -.128, t = -4.46, p < .001). That is, middle school students tended to have a lower levels of smartphone addiction if they had a better relationship with their fathers.

Finally, the results of regression analyses of smartphone addiction factors in high school students are summarized in Table 7. The first model, in which demographic and sociological factors were included, was found to account for 4.6% of the total variance of smartphone addiction among high school students (F(4, 1,266) = 16.16, p < .001). For high school students, gender and economic status of household were identified as the significant factors affecting smartphone addiction. In female students (β = .188, t = 6.85, p < .001) and in those with a lower household economic status (β = -.105, t = -3.80, p < .001) smartphone addiction levels were higher. After controlling for these demographic and sociological variables, the second model with average smartphone usage time per day, increased the explained variance by 3.5%(△F(1, 1,265) = 48.42, p < .001), explaining a total of 8.0% of the variance in smartphone addiction for high school students (F(5, 1,265) = 23.09, p < .001). Average smartphone usage time per day was found to positively affect high school students’ smartphone addiction.

Table 7. Factors Affecting High School Students’ Smartphone Addiction
 Factors Model 1 Model 2 Model 3 Model 4 β t β t β t β t Demographic Gender(M=0) School Year Local Community Economic Status .188  .011 -.041 -.105 6.85***   0.42 -1.49 -3.80*** .167  .024 -.029 -.094 6.12***   0.88 -1.07 -3.49*** .160  .013 -.024 -.050 6.16***     0.51   -0.92   -1.91 .138  .011 -.021 -.028 5.27***   0.45 -0.81 -1.09 Usage Time .190 6.96*** .177 6.84*** .165 6.45*** Friends Numbers Quality -.001 -.306 -0.04 -11.77*** .010 -.249 0.38 -9.07*** Adults Father Mother S. Teacher I. Teacher -.138  .017 -.048 -.084 -4.36***   0.53 -1.71 -3.13** R2(adjusted R2) .049(.046) .084(.080) .175(.170) .205(.198) $\Delta$R2 .049*** .035*** .091*** .030***

Note. *p < .05, **p < .01, ***p < .001.

Model 3 with the number of friends and the level of relationship with friends accounted for 17.0% of variance in the dependent variable (F(7, 1,263) = 38.24, p < .001) by explaining 9.1% more than Model 2 (△F(2, 1,263) = 69.82, p <.001). However, the number of friends did not significantly affect high school students’ smartphone addiction (β = -.001, t = -.04, n.s.), while the level of relationship with friends was revealed to act as a factor that prevented high school students’ smartphone addiction (β = -.306, t = -11.77, p < .001). In the case of high school students, regardless of the number of friends they had, the higher the quality of their relationship with friends, the lower the level of smartphone addiction.

The four types of adult relationships included in the last model explained 3.0% more of the variance, which was statistically significant (△F(4,1,259) = 12.02, p < .001), and a total of 19.8% of variance in smartphone addiction among high school students were explained by Model 4 (F(11, 1,259 = 29.56, p < .001). For high school students, the levels of relationships with father (β = -.138, t = -4.36, p < .001) and institute teachers (β = -.084, t = -3.13, p = .002) were found to be statistically significant factors that decreased the level of smartphone addiction. That is, high school students tended to have a lower level of smartphone addiction when they had better relationships with their fathers and private institution teachers.

Table 8. Comparative Summary of the Significant Factors Affecting Adolescents’ Smartphone Addiction
 Model Factors Elementary Middle High β t β t β t 1 Demographic Gender(M=0) School Year Local Community Economic Status -.137 -.096 -.159 -5.43***    -3.79***    -6.30*** .153  .129 -.112 5.91***     4.98***    -4.28*** .188 -.105 6.85***   -3.80*** 2 Usage Time .395 16.87*** .326 12.78*** .190 6.96*** 3 Friends Number Quality -.264 -11.65*** .057 -.319 2.43* -13.72*** -.306 -11.77*** 4 Adults Father Mother S. Teacher I. Teacher -.107 -.075 -4.15***    -3.09** -.128 -4.46*** -.138 -.084 -4.36***   -3.13**

Note. *p < .05, **p < .01, ***p < .001.

In Model 3, the number of friends, the quantitative aspect of friendship, affected only middle school students’ smartphone addiction; the more friends they had, the higher levels of their smartphone addiction (β =.057, t = 2.43, p = .015). Meanwhile, the quality of friendships was found to be a protective factor that negatively affected the levels of smartphone addiction in all of the school levels (elementary school: β = -.264, t = -11.65, p < .001; middle school: β = -.319, t = -13.72, p < .001; high school: β = -.306, t = -11.77, p < .001).

Finally, it was confirmed that adult relationship factors included in the final Model 4 were statistically significant in explaining adolescents’ smartphone addiction. However, the adult-relationship factors influencing adolescents’ smartphone addiction varied by their school levels. For elementary school students, the relationships with their mothers (β = -.107, t = -4.15, p < .001) and school teachers (β = -.075, t = -3.09, p = .002) among the four types of adult relationships were found to be protective factors for their smartphone addiction. For middle school students, on the other hand, their relationship with their fathers was the only significant protective factor lowering the level of smartphone addiction (β = -.128, t = -4.46, p < .001). Lastly, high school students’ smartphone addiction was found to be negatively affected by their relationships with father (β = -.138, t = -4.36, p < .001) and their institute teacher (β = -.084, t = -3.13, p = .002).

# Discussion

To achieve these research aims, this study analyzed the data of the 2019 Korean Children & Adolescents’ Happiness Index Research. First, time spent in using a smartphone was found to increase as the school level rises in order of elementary, middle, and high school. Subsequently, the level of smartphone addiction increased as the school level goes up from elementary through middle and high school. Consistent with previous research suggesting that older adolescents are at a higher risk of smartphone addiction (Fırat et al., 2018; E. J. Lee & Ogbolu, 2018), this finding suggests that adolescents’ smartphone addiction can be explained linearly based on the age variable. Nevertheless, it is suggested for future research to identify factors that increase risk and are preventable for adolescents’ smartphone use, specifying age group.

The fact that not only smartphone usage but also smartphone addiction increases along with adolescent’s school level means that risk factors will have more effect on adolescents than factors that protect them in daily life. Given that excessive amounts of smartphone use in early adolescence could lead them to be vulnerable to smartphone addiction later, we may need to ponder what factors prevent adolescents’ smartphone use and/or addiction at younger ages.

In spite of the noteworthy findings, there are several limitations. Although a variety of internal and external factors can be associated with smartphone use/addiction, this study did not comprehensively test all possible variables. Based on the findings of this study, other types of hazardous factors could be proposed in future studies. In addition, this study only focused on time spent in using smartphone per day in the data analysis. In this sense, future researchers could propose other types of smartphone-use factors. Not only the amount of time using smartphones, but also numerous usage-related factors such as purposes of using smartphones, smartphone features, newness of smartphone purchase, and various kinds of applications drive users to frequent use in daily life. These factors could have different effects on adolescents’ smartphone addiction in their growth environments.

After analyzing the effects of demographic factors on smartphone addiction by school level, the economic status of the household was revealed as the only common demographic factor that affects smartphone addiction in a negative way throughout adolescence. This finding is consistent with the recent studies (Chang et al., 2019; J. Lee & Kim, 2020), suggesting that the prevalence of smartphone addiction is higher among teenagers from households of low socioeconomic status. It is suggested that even within the younger group, in which there is little access gap, there may be a usage gap depending on the socioeconomic status of parents or families. Therefore, academic, social, and policy efforts to bridge the digital divide in the future need to consider the problem of smartphone addiction among adolescents.

As expected, the amount of time using the smartphone was revealed to be positively associated with the level of smartphone addiction across all school levels. However, due to the limitations of the data, this study failed to explore more specific factors related to addiction, such as functions or applications mainly used through smartphones. It should also be suggested for future research to specify the amount of time spent using various kinds of functions or applications on smartphones.

While the number of friends, the quantitative aspect of friendship, was found to have a positive effect on smartphone addiction only in middle school students, the higher the quality of friendship, the lower the risk of smartphone addiction in all adolescents. These findings are consistent with those of previous studies suggesting that factors related to the quality of friendship such as friendship satisfaction (Bae, 2015), supportive social networks (Ihm, 2018), and friendship quality (H.-J. Kim et al., 2018) are protective factors for teenagers’ addiction to smartphones. On the other hand, this study confirmed that the number of friends, the quantitative dimension of friendship that has been overlooked in previous studies, has no significant relationship with smartphone addiction in adolescents or may rather act as a risk factor, especially for middle school students. Given these findings, specific programs to help adolescents form and maintain healthy and good friendships need to be considered as a way to prevent or intervene in smartphone addiction.