The Interdependence of Family Functioning and Problematic Internet Use in a Representative Quota Sample of Adolescents

Abstract

Few studies have been carried out investigating the interdependence of family structures or interactions and excessive adolescent Internet use. In this study, we surveyed a representative German quota sample of 1,744 adolescents aged between 14 and 17 years with standardized questionnaires. Adolescents assessed their perceived own functioning in the family with the Self-Rating Scale (FB-S) of the German version of the Family Assessment Measure III, and reported on problematic Internet use with the Compulsive Internet Use Scale (CIUS). To predict problematic Internet use (CIUS summary score), we conducted a multiple stepwise linear regression analysis with the seven FB-S scales, the FB-S overall index, and gender and age as explanatory variables. For the full sample, a model with only one predictor (FB-S overall index) that summarizes the quality of family functioning produced a corrected coefficient of determination of 0.239 and explained variance of nearly 24%. t Test results for unpaired samples showed significant differences in the mean values of the FB-S scales and the FB-S overall index for comparisons of both sexes, as well as of a lower age group and higher age group. The prediction of problematic Internet use between both sexes and both age groups showed comparable findings (males: corrected coefficient of determination=0.288; females: corrected coefficient of determination=0.183; lower age group: corrected coefficient of determination=0.231; higher age group: corrected coefficient of determination=0.251), each with a single predictor (FB-S overall index). The results emphasize the importance of family functioning for the occurrence of problematic Internet use in adolescents.

Introduction

Prevalence and patterns of Internet use among adolescents in Germany

Using the Internet for various activities is an integral part of life for German adolescents. According to recent surveys, 100% of 14 to 19 year olds spend time on the Internet at least occasionally.^1,2 In 1997, by contrast, the prevalence of Internet use by adolescents was only 6.3%.² On average, German adolescents aged between 12 and 19 years spend 138 minutes per day online.³ Internet use is the second most important type of media use for adolescents in Germany (86% consider it as “important” or “very important”) following “listening to music” (91%), but it is more popular than “use of cell phones” (80%) or “watching television” (56%).³

Empirical findings on family structures and Internet use

Parental behavior and family structures have a major impact on the psychosocial development of children and adolescents. Van den Eijnden et al.⁴^(p78) emphasize the important role of the parents as “important and influential agents,” and state that “their parenting practices may promote or prevent the development of Internet-related problems.” However, few studies thus far have investigated the role of family structures or interactions (e.g., intra-familial communication or conflicts, parental attitudes, and upbringing behavior) in the context of adolescent excessive Internet use.

Yen et al.⁵ examined 3,662 adolescents (37% females; average age 15.47 years) and discovered that higher parent–adolescent conflict, lower family function, and perceived parents' positive attitude to adolescent substance use were predictive for adolescent Internet addiction. Park et al.⁶ conducted a study with 903 adolescents (30% females, no information given on the average age of the sample). The results of Park et al.⁶^(p895) showed that “parenting attitudes, family communication, family cohesion, and family violence exposure (e.g. conjugal violence and parent-to-child violence) were associated with Internet addiction.” In their sample of 1,289 adolescents (48% females, average age 17.46 years), Lin et al.⁷^(p993) described “participative and supportive parental monitoring” as a “major inhibitor of Internet addiction.” Further protective factors were family and outdoor activities.⁷

Huang et al.⁸^(p401) investigated 304 adolescents (6% females, average age 18.0 years) and reported that “adolescents with IAD [Internet addiction disorder] consistently rated parental rearing behaviors as being over-intrusive, punitive, and lacking in responsiveness” compared to controls. Siomos et al.⁹^(p211) described “Father's care” as well as “Mother's care and overprotection” to be “the best predictor variables for Internet and computer addiction.” The other examined factors (parental knowledge about Internet use, parental security practices against Internet use, age of parents or children, ethnical background, etc.) had only minor significance.⁹

Family function and adolescent Internet use

Family functioning seems to play an important role in youth problematic Internet use. Siomos et al.⁹^(p212) describe a lack of “normal family function in general” as “an important contributing factor to the development of Internet addiction.” Ko et al.¹⁰^(p549) report “poor family function” to be a risk factor of Internet addiction in adolescents, which was also confirmed by the results of Yen et al.⁵

Studies investigating family functioning and problematic Internet use are extremely rare. Exceptions are the studies just mentioned by Yen et al.⁵ and Ko et al.¹⁰ who explored this research question with a 5-item screening questionnaire (Family AGPAR). In our survey, we considered family functioning in a representative German quota sample of adolescents in order for the results to be transferable to the general population (to our knowledge, ours is the first study of this kind). By using a multidimensional questionnaire as a diagnostic tool, we investigated whether specific aspects or family functioning as a whole were of particular importance for adolescent Internet use. Furthermore, we examined whether the findings of Yen et al.⁵ also apply in Germany. In addition, we take a new perspective on family influences regarding problematic Internet use by exploring how the adolescent assesses his or her own role within the family (“self-functioning”) compared to a more general assessment of family functioning as a whole.

In several studies investigating the frequency of problematic to pathological adolescent Internet use, researchers have detected gender differences.^11–13 Furthermore, some empirical findings have indicated a change in computer/Internet use in the course of adolescence.^14,15 For German youth, an increase in average daily Internet use time for different age groups was detected, ranging from 80 minutes in 12–13 year olds to 168 minutes in 18–19 year olds.¹⁶ Therefore, we decided to examine age and gender differences in the our sample.

Aim and research questions of the study

The aim of this explorative study was to investigate the interdependence of perceived adolescent self-functioning in the family and the occurrence of problematic or excessive Internet use by the youth, considering age and gender. We explored the following research questions:

RQ1: Are there differences between males and females in their perceived self-functioning in the family?

RQ2: Are there differences between younger and older youths in their perceived self-functioning in the family?

RQ3: To what extent does perceived self-functioning in the family determine the variance of occurrence of problematic Internet use (a) for the whole sample, (b) between both sexes, (c) between the lower and higher age groups?

Methods

Data collection

We investigated a representative German quota sample of 1,744 adolescents aged between 14 and 17 years in each case together with one parent. Data collection was carried out by an experienced German market research institute. In advance, our research group defined fixed targets concerning age and gender of the youth. Only adolescents aged between 14 and 17 years were surveyed. For the sample, 50% of males and females and 25% of adolescents in each age group (14, 15, 16, and 17 years) were examined. The market research institute determined target values for all other sample quotation features (type of school, single parents, number of inhabitants in a town, and federal state) to assure that the sample was representative of the wider German population. In the data collection, the target achievement was very close to the defined target values. Data were collected by interviewers in all 16 German federal states. They conducted face-to-face interviews with each adolescent and a parent in the family's home.

Measures

We used the Self-Rating Scale (FB-S) of the German version of the Family Assessment Measure III (Familienbögen)¹⁷ to assess family functioning. The self-rating scale “focuses on the individual's perception of his/her own functioning in the family.”¹⁸ The interviewee assesses his/her situation within the family, and the results provide an indication for potential family resources and problems. The FB-S consists of 28 items with a 4-level response format (0=“completely right”; 1=“rather correct”; 2=“rather incorrect”; 3=“completely wrong”). Seven scales are computed, each consisting of four items: “task accomplishment” (e.g., “If problems arise in my family, I usually let others solve them”); “role performance” (e.g., “I have a different opinion on who should do what things in our family”); “communication” (e.g., “When I say something, my family understands what I mean”); “affective expression” (e.g., “If I get mad with my family, this feeling is quite persistent”); “involvement” (e.g., “I care about my family”); “control” (e.g., “Sometimes I demand from other family members to do what I want”); and “values and norms” (e.g., “My family and I agree on what is right and wrong”). Low scores indicate good functioning. By summing up all 28 items or the seven scale values, an overall index¹⁸ (Cronbach's α=0.91 in our sample) was determined.¹⁷

Characteristics of adolescent Internet use were assessed with the Compulsive Internet Use Scale (CIUS).¹⁹ The CIUS questionnaire consists of 14 items with a 5-level response format (0=“never,” 1=“seldom,” 2=“sometimes,” 3=“often,” 4=“very often”) and “provides a dimensional score (i.e., severity) of problematic Internet use.”²⁰ Higher scores indicate problematic or pathological Internet use. By comprising all 14 items, a CIUS summary score (Cronbach's α=0.93 in our sample) was determined.

Sample

The study included 1,744 adolescents (age range: 14–17 years), of which 50% were male. In each age group (14, 15, 16, and 17 years), we investigated 436 adolescents (25%). The average age of the sample was M=15.50 (SD=1.12) years. In our sample, 39% of the adolescents attended a grammar school, 23% attended the junior high school, and 19% a secondary modern school. Furthermore, 14% went to a comprehensive school, and 3% to a special school. Almost 2% were receiving vocational training. Twenty-seven percent of the sample lived in large cities (with more than 100,000 inhabitants), 29% lived in medium-sized towns with between 20,000 and 100,000 inhabitants, and 44% lived in small towns (no more than 20,000 inhabitants).

Statistical analyses

All statistical analyses were performed with the Predictive Analysis SoftWare v18.0 (PASW, SPSS Inc., Chicago, IL). In our cross-sectional study, we abstained from alpha-level adjusting following Bender and Lange²¹ who state that data of exploratory studies could “be analyzed without multiplicity adjustment.” Two-sided error probabilities were used. The Student t test was applied to compare means between two groups (gender or age). We calculated multiple stepwise linear regression analysis to predict problematic Internet use. In addition, we controlled for multicollinearity between the independent variables of a model with the condition index and by calculating Pearson's product–moment correlations.

Results

Results for the full sample

We computed multiple linear regression analysis with the CIUS summary score as response variable using the seven FB-S scale values, the FB-S overall index, and with gender and age as explanatory variables. The first model with the FB-S overall index as the only predictor produced a corrected R² of 0.239 and explained variance of nearly 24%. In four other models, four additional predictors were included: “involvement” (alteration in R²=0.005), “age” (alteration in R²=0.002), “role performance” (alteration in R²=0.002), and “gender” (alteration in R²=0.002). The final model produced a corrected R² of 0.248, explaining variance of nearly 25%, but with clear indication of multicollinearity of the explanatory variables. As Fromm²²^(p100) states, a condition index (CI) “of >15 points towards a distinct problem and a CI of>30 points towards a serious problem of multicollinearity.” Condition indices for the last three models showed values of 38.60, 43.07, and 47.25, leading to rejection of these solutions. The CI for the first model (only FB-S overall index as predictor) was 5.35, and the CI for the second model (FB-S overall index plus affective involvement) increased to a value of 9.56. On the one hand, the correlation coefficient between the FB-S overall index and the scale affective involvement was quite high (r=0.798); on the other hand, the explained variance only marginally increases in the second model (corrected R² of 0.243). Therefore, it was decided that the first model with the FB-S overall index as sole predictor—B (nonstandardized regression coefficient)=0.392, Beta (standardized beta coefficient)=0.490, corrected R²=0.239, p<0.001—was the best statistical solution for the full sample.

Results for male and female youth

t Tests for unpaired samples demonstrated significant differences between male and female youth in the mean values of all seven FB-S scales and the FB-S overall index (see Table 1). Table 1 illustrates that male adolescents (MA) reported lower family self-functioning expressed in higher mean values for all scales and the FB-S overall index than female adolescents (FA) (small effect sizes, Cohen's d between 0.11 and 0.28).

Table 1.

FB-S Profiles of Male and Female Adolescents (n=1,744) and Results of the Student t Tests

	Male adolescents (MA, n=872)		Female adolescents (FA, n=872)		Student t test results
FB-S	M	SD	M	SD	t	df	p	Effect size (d)
Task accomplishment	5.40	2.18	4.93	2.11	4.55	1,742	0.000	MA>FA^*** (d=0.22)
Role performance	5.40	2.41	5.13	2.44	2.36	1,742	0.018	MA>FA^* (d=0.11)
Communication	4.24	2.32	3.76	2.30	4.37	1,742	0.000	MA>FA^*** (d=0.21)
Affective expression	4.02	2.06	3.78	2.12	2.37	1,742	0.018	MA>FA^* (d=0.11)
Involvement	4.25	2.23	3.62	2.21	5.95	1,742	0.000	MA>FA^*** (d=0.28)
Control	5.84	2.00	5.60	1.91	2.58	1,742	0.010	MA>FA^* (d=0.12)
Values and norms	4.02	2.40	3.66	2.52	3.02	1,742	0.003	MA>FA^** (d=0.15)
Overall index	33.17	12.23	30.47	12.21	4.60	1,742	0.000	MA>FA^*** (d=0.22)

Note. ^***p<.001; ^**p<.01; ^*p<.05.

Following the same procedure as for the full sample, we conducted separate multiple linear regression analyses for male and female youth with the CIUS summary score as response variable, using the seven FB-S scale values, the FB-S overall index, and age as explanatory variables. The analysis revealed two models for the adolescent males, with one or two significant predictors (FB-S overall index and role performance) respectively. However, the correlation coefficient between the two explanatory variables (as an indicator for multicollinearity) again was high (r=0.821), and we only found a minimal increase of explained variance in the second model compared to the first one (from a corrected R² of 0.288 to a corrected R² of 0.297). Again, for the male adolescents, the first model with the FB-S overall index as sole predictor—B=0.439, Beta=0.538, corrected R²=0.288, p<0.001—was seen as the best statistical solution.

For the analysis of the female adolescents, four models with up to four significant predictors (FB-S overall index, involvement, task accomplishment, and affective expression) were identified. Once again, the correlation coefficients between the FB-S overall index and the three other explanatory variables were quite high (r between 0.790 and 0.801), and only a minimal increase of explained variance was achieved from the first to the fourth model (from a corrected R² of 0.183 to a corrected R² of 0.192). Thus, for the adolescent females, the first model with the FB-S overall index as unique predictor—B=0.337, Beta=0.429, corrected R²=0.183, p<0.001—also appeared to be the best statistical solution.

Results for lower and higher age groups

We divided the full sample into two age groups. The lower age group (LA) consisted of all 14 to 15 year old adolescents; the higher age group (HA) comprised all 16 to 17 year olds. t Tests for unpaired samples showed significant differences between the LA and HA in the mean values of four FB-S scales (role performance, communication, involvement, and values and norms) and the FB-S overall index (see Table 2). Table 2 illustrates that younger adolescents reported lower family functioning—that is, higher mean values in all these scales and the FB-S overall index—than the HA (very small effect sizes, Cohen's d between 0.10 and 0.12).

Table 2.

FB-S Profiles of the Lower and Higher Age Group (n=1,744) and Results of the Student t Tests

	Lower age group (LA, n=872)		Higher age group (HA, n=872)		Student t test results
FB-S	M	SD	M	SD	t	df	p	Effect size (d)
Task accomplishment	5.26	2.20	5.07	2.11	1.89	1,742	0.058	ns
Role performance	5.38	2.48	5.14	2.37	2.06	1,742	0.040	LA>HA^* (d=0.10)
Communication	4.13	2.37	3.87	2.27	2.33	1,742	0.020	LA>HA^* (d=0.11)
Affective expression	3.94	2.15	3.86	2.04	0.78	1,742	0.437	ns
Involvement	4.06	2.32	3.81	2.16	2.32	1,732.16	0.020	LA>HA^* (d=0.11)
Control	5.78	1.99	5.66	1.93	1.27	1,742	0.204	ns
Values and norms	3.99	2.53	3.69	2.40	2.46	1,742	0.014	LA>HA^* (d=0.12)
Overall index	32.53	12.62	31.10	11.92	2.44	1,736.46	0.015	LA>HA^* (d=0.12)

p<0.05; ns=not significant.

As before, we conducted linear regression analyses with the CIUS summary score as response variable for each age group, using all seven FB-S scale values, the FB-S overall index, and gender as explanatory variables. The analysis revealed two models with one or two significant predictors (FB-S overall index and role performance) for the LA. Again, the correlation coefficient between the two explanatory variables (as an indicator for multicollinearity) was very high (r=0.839), and we detected only a very small increase of explained variance in the second model (from a corrected R² of 0.231 to a corrected R² of 0.234). Therefore, for the LA, the first model with the FB-S overall index as the only predictor—B=0.374, Beta=0.482, corrected R²=0.231, p<0.001—appeared to be the best statistical solution.

Finally, for the HA, analysis showed two models with up to two significant predictors (FB-S overall index and involvement). Once again, the correlation coefficient between the FB-S overall index and the second predictor was high (r=0.786) and from explained variance only increased slightly in the second model (from a corrected R² of 0.251 to a corrected R² of 0.255). Thus, for the HA, the first model with the FB-S overall index as unique predictor—B=0.416, Beta=0.502, corrected R²=0.251, p<0.001]—was chosen as best statistical solution.

Discussion

The results of this study point to the important role family functioning may play for youth problematic Internet use. In a representative German quota sample of adolescents, approximately one quarter of the variance of problematic Internet use was explained by perceived self-functioning in the family. The positive correlation between the CIUS summary score (response variable) and the most important explanatory variable (FB-S overall index) indicates that, in our sample, lower perceived self-functioning in the family is associated with a higher level of Internet use problems. Due to systematic sampling, our results are transferable to the population of all adolescents in Germany. The high level of relevance family functioning has for problematic Internet use in this study is in accordance with the result of another published study. As in our survey, Yen et al.⁵ reported “lower family function” (measured with a brief screening questionnaire) to be predictive for problematic or even pathological Internet use by adolescents.

Other research studies have often examined partial aspects of parental behavior such as “parental monitoring,”⁷ “parental rearing behaviors,”⁸ or “parental bonding variables.”⁹ In contrast, the youth surveyed here were asked to appraise their own functioning in the family, rendering our results difficult to compare. Instead, the findings of these different approaches are complementary: children's views on parental behavior and on their own behavior within the family both seem relevant for problems within the family such as excessive Internet use.

Almost all published surveys use self-report measures to assess family processes. Future studies should combine different research perspectives, for instance, those proposed in the Process Model of Family Functioning by Skinner et al.¹⁸ For example, parent and youth assessments of familial processes and Internet use could be compared and combined in the data analysis. Even more differentiated assessments, for instance of various profiles of excessive Internet use (e.g., online gaming, use of social networks) depending on gender (e.g., male and female youth, mothers' vs. fathers' view) could further deepen the understanding of the role family functioning plays in this context.

Now that the important role of family functioning for problematic adolescent Internet use has been demonstrated, it would be useful to combine this aspect in prospective study designs with a broader focus, including the assessment of parental attitudes and behavioral strategies. In the “cognitive-behavioral model of pathological Internet use,”²³ family is only mentioned once in the context of “lack of social support from family or friends.” This seems a very narrow definition of the role family factors play. In future studies, a new multifactorial model for problematic or pathological Internet use should be shaped that takes the role of the family more into account, especially for adolescents. A better understanding of family influences will help improve prevention and therapeutic treatment approaches for pathological Internet use.

Footnotes

Acknowledgment

The EXIF-study (directed by Professor Rudolf Kammerl) was funded by the Federal Ministry of Family Affairs, Senior Citizens, Women and Youth (BMFSFJ).

Author Disclosure Statement

No competing financial interests exist.

References

van Eimeren

, Frees

. Ergebnisse der ARD/ZDF-Onlinestudie 2010—Fast 50 Millionen Deutsche online—Multimedia für alle? [Results of the ARD/ZDF-Online study 2010—Almost 50 million Germans online—Multimedia for everyone?]. Media Perspektiven, 2010; 41:334–349.

van Eimeren

, Frees

. Ergebnisse der ARD/ZDF-Onlinestudie 2011—Drei von vier Deutschen im Netz—ein Ende des digitalen Grabens in Sicht? [Results of the ARD/ZDF-Online study 2011—Three out of four Germans on the Internet—An end of the digital divide in sight?]. Media Perspektiven, 2011; 42:334–349.

Feierabend

, Rathgeb

. Ergebnisse der JIM-Studie 2010—Medienumgang Jugendlicher in Deutschland [Results of the JIM-study 2010—Media usage behavior of adolescents in Germany]. Media Perspektiven, 2011; 42:299–310.

van den Eijnden

, Spijkerman

, Vermulst

, et al. Compulsive Internet use among adolescents: bidirectional parent–child relationships. Journal of Abnormal Child Psychology, 2010; 38:77–89.

Yen

, Yen

, Chen

, et al. Family factors of Internet addiction and substance use experience in Taiwanese adolescents. CyberPsychology & Behavior, 2007; 10:323–329.

Park

, Kim

, Cho

. Prevalence of Internet addictions and correlations with family factors among South Korean adolescents. Adolescence, 2008; 43:895–909.

Lin

, Lin

, Wu

. The effects of parental monitoring and leisure boredom on adolescents' Internet addiction. Adolescence, 2009; 44:993–1004.

Huang

, Zhang

, Li

, et al. Mental health, personality, and parental rearing styles of adolescents with Internet addiction disorder. CyberPsychology, Behavior, & Social Networking, 2010; 13:401–406.

Siomos

, Floros

, Fisoun

, et al. Evolution of Internet addiction in Greek adolescent students over a two-year period: the impact of parental bonding. European Child & Adolescent Psychiatry, 2012; 21:211–219.

10.

, Yen

, et al. Factors predictive for incidence and remission of Internet addiction in young adolescents: a prospective study. CyberPsychology & Behavior, 2007; 10:545–551.

11.

Durkee

, Kaess

, Carli

, et al. Prevalence of pathological Internet use among adolescents in Europe: demographic and social factors. Addiction, 2012; 107:2210–2222.

12.

, Yen

, Chen

, et al. Gender differences and related factors affecting online gaming addiction among Taiwanese adolescents. The Journal of Nervous & Mental Disease, 2005; 193:273–277.

13.

Kormas

, Critselis

, Janikian

, et al. Risk factors and psychosocial characteristics of potential problematic and problematic Internet use among adolescents: a cross-sectional study. BMC Public Health, 2011; 11:595.

14.

Dumith

, Garcia

, da Silva

, et al. Predictors and health consequences of screen-time change during adolescence—1993 Pelotas (Brazil) birth cohort study. Journal of Adolescent Health, 2012; 51:S16–21.

15.

Leskauskas

, Gudaitytė

, Kiudulaitė

, et al. Attitudes of Lithuanian secondary school children toward addictive behaviors, their promoting and preventive factors with regard to the age. Medicina (Kaunas), 2011; 47:113–119.

16.

Feierabend

, Rathgeb

. Ergebnisse der JIM-Studie 2011—Medienumgang Jugendlicher in Deutschland [Results of the JIM-study 2011—Media usage behavior of adolescents in Germany]. Media Perspektiven, 2012; 43:339–352.

17.

Cierpka

, Frevert

. (1994) Die Familienbögen—Ein Inventar zur Einschätzung von Familienfunktionen [Family questionnaire—an inventory for the assessment of family functions]. Göttingen: Hogrefe.

18.

Skinner

, Steinhauer

, Sitarenios

. Family Assessment Measure (FAM) and process model of family functioning. Journal of Family Therapy, 2000; 22:190–210.

19.

Meerkerk

, Van Den Eijnden

RJJM

, Vermulst

, et al. The Compulsive Internet Use Scale (CIUS): some psychometric properties. CyberPsychology & Behavior, 2009; 12:1–6.

20.

Khazaal

, Chatton

, Atwi

, et al. Arabic validation of the Compulsive Internet Use Scale (CIUS). Substance Abuse Treatment, Prevention, & Policy, 2011; 6:32.

21.

Bender

, Lange

. Adjusting for multiple testing—when and how?. Journal of Clinical Epidemiology, 2001; 54:343–349.

22.

Fromm

. (2012) Multiple lineare Regressionsanalyse [Multiple linear regression analysis]. In: Fromm

, ed. Datenanalyse mit SPSS für Fortgeschrittene 2: Multivariate Verfahren für Querschnittsdaten [Data analysis with SPSS for advanced users 2: Multivariate methods for cross-sectional data]. Wiesbaden: Springer VS, pp. 83–106.

23.

Davis

. A cognitive-behavioral model of pathological Internet use. Computers in Human Behavior, 2001; 17:187–195.