Multi-dimensional discriminative factors for Internet addiction among adolescents regarding gender and age

Authors

  • Cheng-Fang Yen md, phd,

    1. Department of Psychiatry, Faculty of Medicine and
    2. Department of Psychiatry, Kaohsiung Medical University Hospital, Kaohsiung Medical University,
    Search for more papers by this author
  • Chih-Hung Ko md ,

    1. Department of Psychiatry, Faculty of Medicine and
    2. Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University,
    3. Department of Psychiatry, Kaohsiung Medical University Hospital, Kaohsiung Medical University,
    Search for more papers by this author
  • Ju-Yu Yen md ,

    1. Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University,
    2. Department of Psychiatry, Kaohsiung Medical University Hospital, Kaohsiung Medical University,
    3. Department of Psychiatry, Kaohsiung Municipal Hsiao-Kang Hospital, Kaohsiung Medical University, Kaohsiung,
    Search for more papers by this author
  • Yu-Ping Chang rn, phd,

    1. School of Nursing, State University of New York at Buffalo, New York, USA
    Search for more papers by this author
  • Chung-Ping Cheng phd

    Corresponding author
    1. Department of Psychology and Research Center for Mind, Brain, and Learning, National Chengchi University, Taipei, Taiwan and
    • *Chung-Ping Cheng, PhD, Department of Psychology, National Chengchi University, No.64, Section 2, ZhiNan Road, WenShan District, Taipei City 11605, Taiwan. Email: cpcheng@nccu.edu.tw

    Search for more papers by this author

Abstract

Aims:  The aim of the present study was to examine the discriminative effects of sociodemographic, individual, family, peers, and school life factors on Internet addiction in Taiwanese adolescents.

Methods:  A total of 8941 adolescents were recruited and completed the questionnaires. Multi-dimensional discriminative factors for Internet addiction were examined using chi-squared automatic interaction detection for gender and sex.

Results:  Depression and low family monitoring were the discriminative factors for Internet addiction in all four gender- and age-specified groups of adolescents. Low connectedness to school, high family conflict, having friends with habitual alcohol drinking, and living in rural areas also had discriminative effects on adolescent Internet addiction in adolescents of different gender and age.

Conclusions:  Multi-dimensional factors were able to discriminate between those adolescents with and without Internet addiction. It is suggested that parents and health and educational professionals monitor the Internet-using behaviors of adolescents who have the factors discriminating for Internet addiction identified in the present study.

ADOLESCENTS ARE MORE vulnerable to Internet addiction as they have less ability to control their enthusiasm for Internet activities.1 Adolescents with Internet addiction may be trapped in their own cyber world, neglect other creative activities, and finally destroy their real-life relationships.2,3 Internet addiction is associated with poor mental health status1,4,5 and low self-esteem6 in adolescents. All the research findings indicate that it is necessary to identify adolescents with Internet addiction as early as possible. Thus, identifying the discriminative factors for adolescents with Internet addiction is a fundamental step in intervention programs.

Sociodemographic correlates of adolescent Internet addiction are the first ones needing to be examined. The prevalence rates and correlates of adolescent Internet addiction have been found to differ according to gender and age.6 Relationships between adolescent Internet addiction and residential background and socioeconomic status (SES), however, need further study. Depression and low self-esteem have been found to be associated with adolescent Internet addiction.1,4,6 Further studies, however, are needed to examine whether the associations persist when a series of individual and social factors are taken into consideration at the same time.

Social context is an important influence on adolescent health,7 the family, peer and school contexts being among the most critical. Previous studies have found that high family conflict and low family function increased the risk of Internet addiction in adolescents.5 Meanwhile, given that common Internet activities for adolescents are characterized as being without predefined stopping points,8 one might have predicted that low family monitoring increases the risk of adolescent Internet addiction. To our knowledge, however, no study has examined the relationship between Internet addiction and the level of family monitoring in adolescents. Adverse family conditions, including broken marriage and family members with substance use, may decrease family monitoring and increase family conflict. Their associations with adolescent Internet addiction also need further study.

The results of previous studies on the association between adolescent Internet addiction and peer interaction are mixed.2,9 Meanwhile, the association between adolescent Internet addiction and the level of connectedness to the peer group needs further study. Previous studies have found that a high proportion of adolescents with Internet addiction have comorbid substance use experience.10,11 Because peer substance-using behavior is a risk factor of adolescent substance use disorder,12 it is worth examining whether peer substance-using behavior also increases the risk of Internet addiction in adolescents. Poor academic achievement has also been found to be associated with adolescent Internet addiction.2 No study, however, has examined whether low connectedness to school and the experience of suspension of schooling increase the risk of adolescent Internet addiction.

Until now, no study has taken the multi-dimensional factors into consideration when examining the discriminative factors for adolescent Internet addiction. The aim of the present study was therefore to examine the discriminative effects of sociodemographic, individual, family, peer, and school life characteristics on Internet addiction in a large-scale, representative adolescent population in Taiwan.

METHODS

Subjects

The current study was based on data from the Project for the Health of Adolescents in Southern Taiwan in 2004, which has been described in detail elsewhere.13 A total of 12 210 adolescent students from 207 classes were randomly selected based on the stratified random sampling strategy. The protocol was approved by the Institutional Review Board of Kaohsiung Medical University. A total of 11 111 adolescent students (91.0%) returned written informed consent. They were invited to anonymously complete the questionnaire. A total of 8941 participants (80.5%; 4646 girls, 4295 boys) completed research questionnaires without omission. Their mean age was 14.7 ± 1.7 years. Because the Internet-using behavior of adolescents differs with age and gender,6 we analyzed the discriminative factors for Internet addiction in adolescents regarding age and gender. Those who were <15 years old were classified as young adolescents, and those who were ≥15 years old were classified as old adolescents. We divided the participants into four groups according to gender and age: old girls (n = 2412), young girls (n = 2234), old boys (n = 2176), and young boys (n = 2119).

Assessment

Chen Internet Addiction Scale

We used the Chen Internet Addiction Scale (CIAS) to assess the level of addiction to Internet use. The 26-item CIAS assesses five dimensions of Internet-related problems.14 The internal reliability of the scale and the subscales ranged from 0.79 to 0.93 in the original study.15 According to the diagnostic criteria of Internet addiction,16 the 63/64 cutoff point has the highest diagnostic accuracy, sensitivity and specificity for Internet addiction.17

Center for Epidemiological Studies' Depression Scale

The 20-item Chinese version of the Center for Epidemiological Studies' Depression Scale (CES-D)18 was used to assess the frequency of depressive symptoms in the preceding week.19 The Cronbach's alpha was 0.93 and 2-week test–retest reliability was 0.78. Those participants whose total score of the CES-D Chinese version was >28 were defined as having significant depression.20

Adolescent Family and Social Life Questionnaire

The subscale on the Adolescent Family and Social Life Questionnaire (AFSLQ) was adapted to assess the levels of family conflict (three items), family monitoring (four items), connectedness to peer group (four items), and connectedness to school (four items),12,16 with Cronbach's alpha ranging from 0.68 to 0.74 and 2-week test–retest reliability from 0.64 to 0.71. The participants whose total scores on the subscales were higher than the median were classified as having high family conflict, low family monitoring, low connectedness to peer group, and low connectedness to school, respectively. The AFSLQ also assessed the habitual alcohol consumption (alcohol consumption three times per week) and illicit drug use among family members and peers.

Family APGAR index

The 5-item Chinese-version of the Family APGAR Index21 was used to measure participants' perceived family support.22 The Cronbach's alpha was 0.91 and 2-week test–retest reliability was 0.68. The participants whose total APGAR score was lower than the median were classified to have low family support.

Rosenberg Self-Esteem Scale

The Rosenberg Self-Esteem Scale (RSES) contains 10 4-point items that assess current self-esteem.23 High scores indicate high levels of self-esteem. The Cronbach's alpha was 0.86 and 2-week test–retest reliability was 0.70. We classified those adolescents whose total RSES score was lower than the 15th percentile of population as having low self-esteem.

Participant sex, age (<15 years old vs ≥ 15 years old), residential background (urban vs rural), and parental education level (>9 years vs ≤9 years) were collected. Participants' experience of suspension of schooling, academic achievement (first two-thirds vs last one-third ranked in their classmates), and parental marital status were also collected.

Procedure

Research assistants explained the purpose and procedure of this study to the students in class. Written, informed consent was obtained from the adolescents prior to participation. All students received a gift that was worth NT$33 at the end of the assessment.

Statistical analysis

Chi-squared automatic interaction detection (CHAID) in the Answer Tree 3.1 software24 was used to detect mutually exclusive subgroups of the sample that differed markedly in regard to Internet addiction in four age- and gender-specified groups. The analysis selected the best predictors of the outcome and divided the sample into subgroups based on that variable while merging non-significant categories. Compared with logistic regression, CHAID directly considers the interaction between variables. Because this is an exploratory procedure, we investigated the replicability of the resulting subgroup categories by analyzing one-half of the sample (calibration sample) and by examining the replication with the remaining half of the sample (cross-validation sample). Two-tailed P < 0.05 was considered statistically significant.

RESULTS

Internet addiction and sociodemographic, individual, peer, family and school characteristics in four gender- and age-specific groups of adolescents are listed in Table 1. Old boys had the highest rate of Internet addiction among the four groups, followed by young boys. No difference in the rate of Internet addiction was found between old and young girls.

Table 1. Multi-dimensional characteristics
 Old girls
(≥15 years)
n (%)
Young girls
(<15 years)
n (%)
Old boys
(≥15 years)
n (%)
Young boys
(<15 years)
n (%)
Internet addiction334 (13.8)273 (12.2)579 (26.6)477 (22.5)
Sociodemographic factors    
 Living in rural areas852 (35.3)1111 (49.7)743 (34.1)929 (43.8)
 Low paternal education864 (35.8)740 (33.1)650 (29.9)725 (34.2)
 Low maternal education1033 (42.8)843 (37.7)837 (38.5)812 (38.3)
Individual factors    
 Low self-esteem365 (15.1)277 (12.4)261 (12.0)175 (8.3)
 Depression362 (15.0)272 (12.2)271 (12.5)171 (8.1)
Family factors    
 High family conflict577 (23.9)481 (21.5)457 (21.0)387 (18.3)
 Low family monitoring707 (29.3)568 (25.4)957 (44.0)792 (37.4)
 Low family support on the APGAR966 (40.0)898 (40.2)1137 (52.3)913 (43.1)
 Family members with habitual alcohol drinking603 (25.0)556 (24.9)537 (24.7)520 (24.5)
 Family members using illicit drugs43 (1.8)20 (0.9)33 (1.5)29 (1.4)
 Broken parental marriage360 (14.9)307 (13.7)280 (12.9)295 (13.9)
Peer factors    
 Low connectedness to peers885 (36.7)880 (39.4)889 (40.9)969 (45.7)
 Friends with habitual alcohol drinking400 (16.6)248 (11.1)479 (22.0)240 (11.3)
 Friends using illicit drugs312 (12.9)139 (6.2)229 (10.5)63 (3.0)
School factors    
 Low connectedness to school865 (35.9)611 (27.4)941 (43.2)837 (39.5)
 Suspension of schooling68 (2.8)18 (0.8)76 (3.5)32 (1.5)
 Low academic achievement495 (20.5)433 (19.4)654 (30.1)601 (28.4)

Results of stepwise CHAID for the discriminative factors for Internet addiction in four gender- and age-specified groups are shown in Figs 1–4. In the calibration sample of all four gender- and age-specified groups, depression was the first and the most significant variable discriminating between the adolescents with and without Internet addiction. In all four gender- and age-specified groups of adolescents without depression, the level of family monitoring was another discriminating factor for Internet addiction. Among the non-depressed old girls perceiving high family monitoring, the level of connectedness to school could further discriminate between the adolescents with and without Internet addiction.

Figure 1.

Discriminating factors for Internet addiction in girls who were ≥15 years old. CA, calibration sample; CR, cross-validation sample.

Figure 2.

Discriminating factors for Internet addiction in girls who were <15 years old. CA, calibration sample; CR, cross-validation sample.

Figure 3.

Discriminating factors for Internet addiction in boys who were ≥15 years old. CA, calibration sample; CR, cross-validation sample.

Figure 4.

Discriminating factors for Internet addiction in boys who were <15 years old. CA, calibration sample; CR, cross-validation sample.

Among the non-depressed young girls perceiving high family monitoring, the level of family conflict was another discriminative factor for Internet addiction. Among the non-depressed young girls perceiving low family monitoring, having friends with habitual alcohol drinking was another discriminative factor for Internet addiction. Among the non-depressed old boys perceiving high family monitoring, residential background could further discriminate between the adolescents with and without Internet addiction. Among the non-depressed young boys perceiving high family monitoring, the level of family conflict was another discriminative factor for Internet addiction.

Results in the cross-validation sample of four gender- and age-specified groups are also shown in Figs 1–4, which indicated that the results of the CHAID analysis were replicable.

DISCUSSION

This study is the first one to examine the discriminative effects of sociodemographic, individual, family, peer, and school life factors on adolescents with Internet addiction in the community regarding gender and age. In the present study, depression had the most powerfully discriminative effect on adolescent Internet addiction in all gender- and age-specified groups. Two models may account for the association between depression and adolescent Internet addiction.4 First, Kraut et al. reported that Internet use results in negative effect on psychological well-being.25 The negative impact decreased, however, in the follow-up study.26 Second, the Internet could facilitate development of a ‘virtual self’.27 Adolescents with depression may experience the pleasure of control and respect from others on the Internet,28 which may compensate for imperfection in real life. If depression was not well treated, however, they would spend more and more time on the Internet and progress to addiction.

In the present study, low family monitoring was another discriminative factor for adolescent Internet addiction in all gender- and age-specified groups of non-depressed adolescents. Internet activities usually provide the pleasure of control, innominate interaction with others, and perceived fluidity of identity for adolescents.28 Meanwhile, online games and online chatting usually have no predefined stopping points.8 Without effective supervision and discipline in the family, the nature of the Internet activities described here will attract adolescents' excessive engagement and increase the risk of development of Internet addiction.

High family conflict was another discriminative factor for Internet addiction in non-depressed young girls and boys who perceived high family monitoring. Ary et al. have reported that families with higher conflict have lower levels of parent–child involvement,29 which would result in inadequate parental monitoring, which would predict, in turn, adolescents being predisposed to Internet addiction. Furthermore, social control theory suggests that, when adolescents are close to their parents, they feel obligated to act in non-deviant ways to please their parents.30 Thus, adolescents with higher conflict with parents would refuse to conform to the supervision of parents, including rules set for Internet use. Also, perceiving high conflict in the family, adolescents may seek social support from interactions on the Internet.31 Unfortunately, heavy Internet use by adolescents usually results in further conflict with their parents, which may make the problem of adolescent Internet addiction more difficult to resolve.5 It is noteworthy that the discriminative effect of high family conflict on Internet addiction was found only in young adolescents, which indicates that the stage of chronological development may play a role in the association between high family conflict and adolescent Internet addiction.

The present study found that low connectedness to school could discriminate between those with and without Internet addiction among the non-depressed old girls perceiving high family monitoring. There were several possible explanations for this. First, Internet addiction might destroy adolescents' daily schedules and reduce total sleep time by substituting for it, which might reduce the responsiveness to school affairs. Second, adolescents with lower connectedness to school might receive less support and admiration from the real world, which might make them look for support and admiration in Internet activities. Third, both Internet addiction and low connectedness to school might be the results of a chaotic lifestyle of adolescents.

Among the non-depressed young girls perceiving low family monitoring, having friends with habitual alcohol drinking was another discriminative factor for Internet addiction. Adolescent Internet addiction and substance use have been found to occur together frequently. Internet addiction was proposed to be one kind of problem behavior among adolescents.10 Peers' habitual alcohol drinking has been found to be a risk factor for adolescent alcohol use disorder,11 and the association between Internet addiction and peers' habitual alcohol drinking further supported that Internet addiction might be one kind of problem behaviors at least among a portion of adolescents.

We found that among the non-depressed old boys perceiving high family monitoring, residential background could discriminate between the adolescents with and without Internet addiction. Compared with those living in urban areas, adolescents and parents living in rural areas might have less access to information on the adverse impact of excessive Internet use and the necessity to control Internet-using behaviors, and adolescents would have fewer kinds of entertainment in their leisure time, apart from Internet activities.

Some other limitations of this study should be addressed. First, the cross-sectional research design limited the possibility of drawing conclusions regarding causal relationships. In order to examine the causal relationships between Internet addiction and related factors, we are conducting a 2-year follow-up study to investigate this. Second, the data were provided by the adolescents themselves, and the validity of some data cannot be easily quantified. Third, the present study recruited adolescent students as the research population, but adolescents who had dropped out of school and were attending night school were not included. They may have different patterns of Internet addiction and correlates compared with the adolescents recruited in the present study. Fourth, although we examined the relationships between many factors in Internet addiction, there are still other individual and social factors that were not included. Further study is needed to examine their associations with Internet addiction. Fifth, some of these discriminative factors are approximate and may have limited specificity to identify adolescents with Internet addiction for clinical use. We will examine which subtypes of family and peer interactions increase the risk of Internet addiction, for example, the types of parenting and the rank in peer group. In future we will also organize parenting skills training groups for the parents of adolescents with Internet addiction to improve the quality of family monitoring and reduce family conflict. Severity of adolescents' Internet addiction will be followed up to examine the effects of parenting skills training.

CONCLUSION

This study found that demographic, individual, family, peer, and school life factors had discriminative effects on Internet addiction in adolescents. Given the possible adverse effects of Internet addiction on adolescent development, we suggest that adolescents who have the discriminative factors for Internet addiction as identified in the present study should be monitored for the risk of Internet addiction. Because Internet addiction and depression may have a reciprocal relationship with each other, monitoring the existence of and understanding interventions for depression are important in interventions for Internet addiction in adolescents. Family-based prevention for Internet addiction should include skills training for parents to improve monitoring and discipline with regard to Internet addiction, fostering skills for healthy family interactions, and monitoring for the alcohol-using behaviors of peers. School staff should also watch for the possibility of Internet addiction in adolescent students who have low connectedness to school. It is also necessary to develop effective strategies for screening and intervening for adolescents living in rural areas who have Internet addiction, by more efficiently applying the limited health and educational professional resources in rural areas.

ACKNOWLEDGMENTS

This study was supported by grant NSC 93-2413-H- 037-005-SSS awarded by the National Science Council, Taiwan (ROC).

Ancillary