Prevalence and Risk Factors of Video Game Dependency in Adolescence: Results of a German Nationwide Survey

Abstract

In this article, results of a German nationwide survey (KFN schools survey 2007/2008) are presented. The controlled sample of 44,610 male and female ninth-graders was carried out in 2007 and 2008 by the Criminological Research Institute of Lower Saxony (KFN). According to a newly developed screening instrument (KFN-CSAS-II), which was presented to every third juvenile participant (N = 15,168), 3% of the male and 0.3% of the female students are diagnosed as dependent on video games. The data indicate a clear dividing line between extensive gaming and video game dependency (VGD) as a clinically relevant phenomenon. VGD is accompanied by increased levels of psychological and social stress in the form of lower school achievement, increased truancy, reduced sleep time, limited leisure activities, and increased thoughts of committing suicide. In addition, it becomes evident that personal risk factors are crucial for VGD. The findings indicate the necessity of additional research as well as the respective measures in the field of health care policies.

Introduction

While TV usage time could be observed to be fairly consistent in almost all age groups in the past decade,¹ the relevance of interactive entertainment media as leisure-time activity has grown considerably both for youths and adults. In Germany, for example, computer ownership among adolescents ages 12 to 19 increased from 35% to 71% from 1998 to 2008, and ownership of gaming consoles increased from 23% to 45%. The availability of Internet access in adolescents' bedrooms rose from roughly 10% in 1998 to 51% in 2008.^2,3

Recently, the issue of extensive video game (the term video games is used for both video and computer games in this article) behavior has gained further attention in research and clinical practice.^4–8 Young⁹ specified a wide range of potentially addictive phenomena, such as cybersexual addiction, cyberrelational addiction, Net compulsions, information overload, and computer addiction. Subsequently, frequent attempts to explore these phenomena on an aggregated level called computer-related addictive behavior,¹⁰ Internet addiction, or compulsive Internet use^11,12 are observed. However, due to the richness of different computer- and Internet-related activities, these terms could be criticized as not being precisely operationalized.⁷ To date there has been little empirical insight into these diverse kinds of problematic media use behaviors and a subsumption under one medical category could prove premature.

With video game dependency (VGD), we focused our research efforts on a clinical phenomenon currently reported to be notably widespread among computer-related problematic behaviors.⁸ Video games can satisfying a wide number of individual psychological needs and, because of their interactive structures, give rise to an intense experience of gratification in the user.^13,14 In sophisticated games such as massively multiplayer online role-playing games (MMORPGs), players can regularly acquire or obtain points, placements on ranking lists, pieces of virtual equipment, new abilities, and opportunities to configure their own avatar. The allocation of these rewards frequently follows a pattern of intermittent reinforcement to the effect that the placing and magnitude of rewards are rarely predictable.⁵ According to learning psychology, reinforcement schedules of this kind generate the highest response rates as well as the highest resistance to extinction¹⁵ and therefore particularly promote an excessively rewarding behavior. Hence, video games in their very conception show a structural similarity to gambling and, as a consequence, are suspected of bearing a similar psychotropic dependency risk for some people.¹⁶

The distinctive mechanisms, however, involved in binding certain players to certain virtual worlds are roughly systematized and barely understood in detail to date. Different risk factors of VGD have been discussed, such as playing more complex games such as MMORPGs,^4,17 gaming in terms of dysfunctional coping,^6,7,18 low parental support,¹⁹ school-related behavioral problems,¹⁹ higher aggressiveness, and the acceptance of violence as an inherent part of social life and part of one's own behavioral repertoire,^4,20 as well as in terms of male gender in general.^4,7 Video game–dependent youths have also been assumed to exhibit a lower education level,⁷ whereas other studies have not proven the educational background to be of importance for predicting VGD.¹⁹ VGD is often reported to be accompanied by psychopathological stress indicators, such as attention deficit hyperactivity disorder (ADHD), as well as affective disorders.^8,10 Their relevance for predicting VGD, however, is still a matter under discussion, and large-scale studies considering them altogether are missing.

Like other forms of potentially addictive computer use, VGD has not been clinically accepted so far. In 2007, the application to the Diagnostic and Statistical Manual of Mental Disorders (DSM) of the American Psychiatric Association was still assessed negatively with the note that the database for such a step was not yet sufficiently secured.²¹ Against this backdrop, in the past years, increasing efforts have been made to develop diagnostic instruments to examine VGD and to determine the percentage of persons affected by it.^4,6,7 To allow for clinically significant statements, analysts frequently refer to criteria of substance-related dependency or pathological gambling.⁵ These criteria, however, have been applied and weighted inconsistently so far. Due to this heterogeneous definition and varying sampling approaches in different age groups and populations, prevalence estimates for VGD currently range from 1.5% to 9.3% between different studies.^4,6

The first aim of the present study was to validate the newly developed, short, and easy-to-administer Video Game Dependency Scale (KFN-CSAS-II) in a German nationwide representative sample of ninth-graders. It was intended to get a first nationwide prevalence estimate of VGD in a well-defined age group. A second aim was to identify risk factors predicting VGD in adolescence. We assumed factors of gaming behavior such as playing complex and often online used games like MMORPGs, shooters, and strategy games, as well as gaming in terms of dysfunctional coping and gaming as a source of self-efficacy to contribute to the risk of VGD. We also expected social factors such as a lack of success in leisure-time activities and school related behavioral problems to increase the risk of a dependency. Regarding sociodemographic parameters, we supposed male participants to be at higher risk, but did not expect parental or juvenile educational background to be of any importance for predicting VGD. With regard to psychological constructs, we hypothesized that personal factors such as lower levels of social competence, higher levels of impulsiveness, a higher acceptance of violence, a history of ADHD, a diagnosis of depression or anxiety disorders as well as physical abuse in childhood could contribute to the risk of VGD.

Methods

Procedure

From April 2007 to October 2008, the Criminological Research Institute of Lower Saxony (KFN) conducted a survey with 44,610 ninth-grade students in 61 randomized regions of Germany. Using a proportional sampling of regions from eastern and western Germany as well as independent cities and administrational districts (counties) of different sizes, a representative sample of the population of German ninth-graders at schools offering a general education was achieved.²² A 31-page standardized inventory was administered within the context of the class and completed by the students on their own after instruction by a trained interviewer. The inventory covered a wide range of research interests, such as juvenile delinquency, drug use, victimization, xenophobia, school absenteeism, and parental educational styles. A further inventory of six pages containing more focused questions on the topic of media use and VGD was handed out at random to every third participant. Thus, regarding VGD, 15,168 cases resulted for further analysis. The return quota was 88%.

Measurement of VGD

The Video Game Dependency Scale (KFN-CSAS-II) is based on the Internet Addiction Scale (ISS-20),¹¹ which was extended and adapted to the issue of VGD.⁵ The KFN-CSAS-II consists of 14 items (4-point scale: 1, incorrect, to 4, absolutely true). It follows the classification of dependency of ICD-10 and covers the dimensions preoccupation/salience (4 items), conflict (4 items), loss of control (2 items), withdrawal symptoms (2 items), and tolerance (2 items). Out of the 15,168 adolescents who took part, 10,060 responded to all of the 14 items. Three hundred forty-two persons responded to 13 of the 14 items, whereas the missing item was substituted by the scale mean. Out of the 4,766 participants who could not be diagnosed because of two or more missing values, 2,635 persons were nonplayers or occasional players.

The mean item values reveal high item difficulties for all items (see Table 1). All of the items show good discriminatory power (r_i(t-i) ≥ 0.6). For the diagnostic status, all four-level variables are added together so that the scale assumes a value of between 14 and 56 points (α = 0.92). The mean scale value for our representative sample of 15-year-old adolescents is 19.8 points (SD = 7.42, SE = 0.07). The cutoff values for the KFN-CSAS-II were defined according to the ISS-20.¹¹ A sum score of 35 to 41 indicates that a person is at risk of developing VGD, as the items on an average are not rejected any longer (14 × 2.5 = 35). Persons who reach this value are already about 2 standard deviations above the population mean. With a sum score of 42 and higher, by which on average all items meet with approval, a person is classified as dependent on video games. Those who reach this critical value are about 3 standard deviations above the mean. With these comparatively strict cutoff values, the sensitive subject matter is to be taken into account by accepting a lower diagnostic sensitiveness in favor of a higher diagnostic specificity.

Table 1.

Item Parameters of the Video Game Dependency Scale KFN-CSAS-II (n = 10,402)

	M	SD	r_i(t-i)
Preoccupation/salience
During the time that I don't play video games, in my thoughts I am very much occupied with games.	1.64	0.87	0.61
My thoughts continually circle around playing video games, even when I'm not playing.	1.26	0.62	0.69
At certain times or in certain situations, I actually always play: That has almost become a routine for me.	1.55	0.90	0.63
It happens that I'm virtually doing something completely different and then, without thinking, start a video game.	1.36	0.73	0.60
Conflict
My school achievement suffers under my game habits.	1.31	0.68	0.64
I am so frequently and intensively occupied with video games that sometimes I have problems in school.	1.35	0.70	0.70
People important to me complain that I spend too much time playing.	1.55	0.83	0.64
Because I play too much, I undertake less with others.	1.38	0.72	0.63
Loss of control
I often spend more time playing video games than I planned.	1.50	0.83	0.65
I have a feeling that I cannot control my gaming time.	1.42	0.79	0.64
Withdrawal symptoms
If I can't play, I am irritable and dissatisfied.	1.39	0.71	0.70
If I don't play for quite a while, I become restless and nervous.	1.23	0.59	0.67
Tolerance
I have a feeling that video games are getting more and more important for me.	1.47	0.78	0.73
I feel forced to play longer and longer to be satisfied.	1.30	0.68	0.65
Scale	Cronbach's α = 0.92

Note: Items as translated from German. The statistical values refer to the German version.²⁶ Means are based on a four-stage response format (1, incorrect, 2, hardly correct, 3, rather correct, 4, absolutely true). r_i(t-i) = selectivity according to item-rest correlation. Unidimensionality is safeguarded by factor analysis.

Measurement of validation variables

For a general assessment of VGD, participants were asked on a 6-point scale (not at all to strongly), “To what extent, do you believe, you are dependent on video games?” (Note that all quoted items are translated from German.²⁶) Participants were also asked to estimate the time they spent on different media activities on school days and on weekend days, including online and offline gaming (15-point scale: 0 minutes to 5 hours and more). Based on this data, a daily usage time was calculated: (school day × 5 + weekend day × 2)/7. Indicators of academic performance were assessed as well: participants stated their grades in the last school report (1, very good, to 6, insufficient) and marked whether they skipped single lessons or whole school days in the past half-year. Based on this information, a total number of skipped lessons was calculated (school day × 5 + single lessons). Multiple truancy was defined as missing more than 5 school days in the last half-year. Additionally, participants stated whether video gaming was the reason for their truancy (2-point scale: yes or no). Sleeping time was calculated as the time span between the self-reported time of going to bed the day before and the time of getting up on the day of the interview (33-point time scale: 15-minute breakdown). Participants were further asked whether they had difficulties in falling asleep in the past week (5-point scale: never to always). Sleeping disturbance was coded when participants quoted always for the week before. Using a wide range of items regarding organized youth activities (e.g., sports club, music school, youth association), participants who in the past year had not participated in at least one activity were identified. Finally, for assessing psychological stress indicators, we asked students about suicidal thoughts: “Have you ever thought of committing suicide?” (4-point scale: no, never to yes, often).

Measurement of risk factors

To assess risk factors regarding gaming behavior, students were asked for their usage times of nine different video game genres in the preceding year, including MMORPGs, first-person shooters, and strategy games (7-point scale: never to daily). Participants were identified as using particular genres if they played the respective genre at least monthly. The adolescents were also asked to name their three currently favorite video games. The resulting string variables were recoded to numeric variables by a trained coder, whereas different versions of one game (e.g., Counterstrike, Counterstrike Source, Counterstrike Condition Zero) were merged into one group. The impact of gaming in terms of dysfunctional coping was assessed by the item “I usually play video games when my life is not going well” (4-point scale: incorrect to absolutely true); the role of gaming as a source of experiencing self-efficacy was assessed with the question “How relevant is it for you to feel powerful and in control when playing action games?” (4-point scale: irrelevant to very relevant). We refrained from including gaming time and variables that are clearly confounded with gaming time as predictors for VGD, since they can be expected to be indicators rather than risk factors of VGD.

Regarding social predictors, participants were asked—referring to a wide range of different areas of youth leisure activities and aspects of common life such as sports, music, friends, family, school, and video games—whether they succeeded in these areas within the previous year (2-point scale: yes or no). Subsequently, participants who could name only video games as a domain of success were identified. School-related anxieties were assessed by a five-item scale (4-point scale: incorrect to absolutely correct) mainly regarding embarrassing situations and failures in test situations (α = 0.78). Previous failures in academic achievement were accounted for by the question whether the participants ever had to repeat a school year (2-point scale: yes or no).

For assessing demographic predictors, we used the gender of students as well as juvenile and parental educational background measured by the question “What is the highest qualification of your parents?” A low parental educational background was classified if neither of the parents had at least completed the middle secondary school Realschule (Realschule is usually completed at the end of tenth grade which qualifies a student to go on to upper secondary school), and a low juvenile educational background was classified if the participants themselves attended the lower secondary school Hauptschule (Hauptschule is usually completed at the end of ninth grade which qualifies a student for entering into vocational training or apprenticeship or to switch to Realschule).

With regard to psychological predictors, a five-item scale (2-point scale: no or yes; α = 0.71) mainly indicating social competence and the ability of putting oneself in the place of another interacting person (e.g., “Before I criticize people I try to imagine how I would feel in their place”) was used. Impulsiveness was measured by using two items of the impulsiveness inventory IVE²³ (“I often get in uncomfortable situations because I didn't think it over” and “I often get in trouble because I can't control myself” (2-point scale: no or yes). Additionally, four items were used to assess the acceptance of violence as an inherent part of social life and part of one's own behavioral repertoire (4-point scale: incorrect to absolutely true), in which participants had to appraise their beliefs about the role of violence in everyday life (α = 0.87). Psychological disorders (depressive, anxiety, and attention) in the students' personal history were assessed by the question whether a physician or psychologist had ever diagnosed these disorders (2-point scale: yes or no). Finally, participants were asked if they had been severely physically abused by their mothers or fathers before the age of 12 (6-point scale: never to repeatedly per week) regarding the following incidents: hit with an object, punched or kicked, beaten or bashed. Students were classified as frequently abused if they had experienced such events at least monthly during childhood.

Participants

The average age of the participants was 15.3 years (SD = 0.69), about half of them being male (51.3%), and 27.4% with a migration background. On average, the students use video games 141 minutes a day. Online games made up almost 60% of the time spent with video games (83 min). Boys play 130 minutes on school days (77 min online) and 167 minutes on weekend days (97 min online). Girls play about 53 minutes on school days (33 min online) and 64 minutes on weekend days (38 min online). Altogether, girls with an average daily gaming time of 56 minutes use video games 90 minutes less than boys; 12.6% of the girls and 39% of the boys could be characterized as extensive gamers spending more than 2.5 hours a day with video games.

Results

Prevalence

Of the adolescents surveyed, 2.8% are classified as being at risk and 1.7% as being dependent on video games. Mainly boys are affected: 4.7% of them are at risk and 3% are dependent. In comparison to boys, a small number of girls are affected: 0.5% are at risk and 0.3% are dependent. Hence, roughly 90% of the youths at risk and 91% of the ones dependent are males.

Validation

For a validation of the Video Game Dependency Scale (KFN-CSAS-II), only data of male participants were used because most of the variables under consideration are confounded with gender. As expected, boys classified as video game dependent show increased usage times (p < 0.01). While this group in particular shows a high daily online gaming time of 188 minutes, offline game usage is comparatively modest (see Table 2). The self-evaluation of the participants corresponds with the classification to a considerable extent (r = 0.59, p < 0.01).

Table 2.

Psychosocial Stress Indicators by Usage Group (Boys Only, n = 7,761)

Psychosocial stress indicators	Main group (<2.5 h/day) n = 4,621	Extensive players (>2.5 h/day) n = 2,619	Players at risk (RS ≥35) n = 324	Dependent players (RS ≥42) n = 197
Primary validation
Online gaming time	26 min	159 min^**	172 min^**	188 min^**
Offline gaming time	33 min	100 min^**	74 min^**	78 min^**
Self-assessment VGD (1–6)	M = 1.8	M = 2.4^**	M = 4.1^**	M = 4.8^**
Grades at school
German	3.14	3.28^**	3.32^**	3.46^**
Mathematics	3.05	3.07^ns	3.30^**	3.29^**
History	2.83	2.92^**	3.03^**	3.17^**
Sports	2.05	2.17^**	2.19^ns	2.42^**
School absenteeism
Skipped lessons	10.1 h	10.0 h^ns	16.4 h^**	18.6 h^**
Multiple truancy	12.5%	12.2%^ns	21.6%^**	22.4%^**
Motive: video gaming	8.2%	23.2%^**	44.9%^**	63.8%^**
Health-related factors
Time of sleep	7.5 h	7.3 h^**	7.1 h^**	6.9 h^**
Sleep disturbance	4.1%	6.1%^**	7.6%^ns	15.1%^**
No regular leisure-time activities	30.4%	36.9%^**	39.0%^*	44.6%^**
Frequent thoughts of suicide	2.4%	3.0%^ns	6.3%^**	12.3%^**

Note: To the main group and to the group of extensive players, there were assigned only boys who in accordance with KFN-CSAS-II were not classified as conspicuous either because they did not reach the critical value (RS < 35, n = 6,367) or could not be diagnosed because of two or more missing values (n = 873). Grades at school do correspond to the German grading system: 1, very good, to 6, insufficient. Test of significance by means of analysis of variance (ANOVA). Post hoc testing with Scheffé test. RS = raw score in KFN-CSAS-II. The deviation from the main group is given.

p < 0.05; **p < 0.01; ns, not significant.

Similar patterns are found with regard to school achievement and school absenteeism. While boys who game extensively show slightly worse grades in German, history, and sports (larger numbers indicating lower performance), the school performance clearly deteriorates in the group of video game–dependent boys (p < 0.01). Additionally, extensive gamers play truant similarly to inconspicuous male youths; video game–dependent boys, however, have skipped more lessons and are more often classified as multiple truants with more than 5 days of absence from school (p < 0.01). Moreover, members of this group more often give video gaming as the reason for their truancy (p < 0.01).

Regarding health-related factors, results show successively reduced sleep time through the groups, culminating in video game–dependent boys with the shortest average time of sleep (p < 0.01). Consistent with this observation, especially video game–dependent boys, but not at-risk boys, report intensive problems falling asleep in the past week (p < 0.01). In addition, an increased part of the video game–dependent male adolescents report not participating in any regular, organized leisure-time activities (p < 0.01). The more frequent thoughts of committing suicide among boys at risk (p < 0.01) and especially dependent boys (p < 0.01) indicate an increased mental stress within these two groups.

Predictors of video game dependency

The top 10 games among boys, in order of their circulation, are Counterstrike (27%), FIFA Soccer (16.1%), Need for Speed (11.4%), Grand Theft Auto (10.1%), World of Warcraft (9.8%), Call of Duty (7.8%), Battlefield (5.1%), Warcraft (4.9%), Pro Evolution Soccer (4.8%), and Guild Wars (2.7%). A comparative analysis of these 10 games among male adolescents found that every fifth player of the online role-playing game World of Warcraft shows a risk for or an already existing VGD. Users of Counterstrike, Warcraft, Battlefield, Call of Duty, and Guild Wars, games that are often or preferably played online, show an increased share of dependent persons, too. The data also suggest a higher share of VGD in males using games played exclusively on personal computers. With the three sports games Pro Evolution Soccer, Need for Speed, and FIFA Soccer, a reduced share of persons at risk becomes obvious. The daily video game times follow a similar pattern, with the highest daily gaming times observed for boys using World of Warcraft (see Table 3). An increased risk of developing VGD seems to be inherent in certain video games. One would expect, however, that certain person-immanent factors as well as factors of the person's social environment add to the risk of VGD, too. These factors could similarly increase the likelihood of using certain types of video games.

Table 3.

Imperiled and Dependent Persons among Players of the 10 Most Popular Games of Male Adolescents (Boys Only, n = 6,230, Weighted)

Player of…	Genre	Usage (hardware)	Deviation from scale mean (KFN-CSAS-II)	At risk	Dependent	Deviation from mean gaming time (min/day)
World of Warcraft	MMORPG	Online (a)	+5.1^**	11.6%	8.5%	+88^**
Guild Wars	MMORPG	Online (a)	+2.8^**	5.8%	3.8%	+56^**
Warcraft	Strategy	Online (a)	+2.5^**	8.7%	3.8%	+53^**
Counterstrike	Shooter	Online (a)	+2.3^**	8.2%	4.9%	+49^**
Call of Duty	Shooter	Both (a, b, c)	+1.0^**	6.2%	4.0%	+33^**
Battlefield	Shooter	Online (a, b)	+0.8^ns	6.5%	4.1%	+33^**
Grand Theft Auto	Genre mix	Offline (a, b, c)	−0.8^*	4.5%	2.7%	−28^**
Pro Evol. Soccer	Sports	Offline (a, b, c)	−2.0^**	3.5%	1.9%	−36^**
FIFA (Soccer)	Sports	Offline (a, b, c)	−2.4^**	2.9%	1.2%	−50^**
Need for Speed	Racing	Offline (a, b, c)	−2.7^**	2.8%	1.8%	−35^**

Online, primarily used with active Internet connection; offline, primarily used without active Internet connection; both, similarly used online and offline; a, used on PC systems; b, used on TV video game consoles; c, used on handheld video game consoles.

Test of significance with t test (two-tailed). Difference between mean of participants playing a certain game and participants playing only other games. Included in analysis are only male participants using video games at least occasionally. General mean values for this group: (KFN-CSAS-II scale) mean, 21.8; at risk, 4.9%; dependent, 3.1%; gaming time, 145 min/day.

p < 0.05; **p < 0.01; ns, not significant.

To determine which of the risk factors contributes to the development of VGD in a significant way, we performed a logistic regression analysis, predicting the classification as being dependent (KFN-CSAS-II RS ≥42). Only boys and girls who were gaming regularly were included in the analysis, since we intended to predict VGD and not video gaming in general. Table 4 shows the descriptive measures of the predictors used in this model.

Table 4.

Predictors in the Logistic Regression Analysis Predicting Video Game Dependency in Boys and Girls Gaming Regularly

Predictors	Min	Max	M	SD
Gaming behavior
Gaming in terms of dysfunctional coping	1 (incorrect)	4 (correct)	1.56	0.89
Gaming as source of self-efficacy	1 (irrelevant)	4 (relevant)	2.36	1.06
Usage of MMORPG	0 (no)	1 (yes)	0.24	0.43
Usage of strategy games	0 (no)	1 (yes)	0.30	0.46
Usage of shooter games	0 (no)	1 (yes)	0.24	0.43
Social predictors
No success in leisure-time activities besides gaming	0 (no)	1 (yes)	0.02	0.14
School-related anxieties	1 (no)	4 (many)	1.95	0.66
Previous repetition of a school year	0 (no)	1 (yes)	0.22	0.42
Demographic predictors
Male gender	0 (no)	1 (yes)	0.61	0.49
Low parental education	0 (no)	1 (yes)	0.19	0.39
Low juvenile education	0 (no)	1 (yes)	0.29	0.46
Psychological predictors
Lower social competence	0 (higher)	1 (lower)	0.39	0.32
Impulsiveness	0 (lower)	2 (higher)	0.54	0.74
Acceptance of violence	1 (lower)	4 (higher)	1.57	0.73
ADHD in personal history	0 (no)	1 (yes)	0.08	0.27
Depression disorder in personal history	0 (no)	1 (yes)	0.05	0.21
Anxiety disorder in personal history	0 (no)	1 (yes)	0.04	0.20
Frequent physical abuse in childhood	0 (no)	1 (yes)	0.03	0.17

Sixty-one percent of the regularly playing adolescents are male. Most of the other predictors are moderately distributed, such as the use of the three game genres (24–30%), lower educational background (19–29%), and previous repetition of a school year (22%). As expected, with regard to the predictors no success in leisure-time activities besides gaming (2%), frequent abuse in childhood (3%), and psychological disorders in personal history (4–8%), only a small number of adolescents could be observed, whereas the prevalence for the psychological disorders are fairly consistent with other studies.^24,25 Nine of the 18 factors considered contribute significantly to the prediction of VGD, accounting for 41% of the variance of VGD in adolescents (see Table 5).

Table 5.

Logistic Regression Analyses Predicting Video Game Dependency in Boys and Girls Gaming Regularly (n = 4,727)

				95% confidence interval for Exp(B)
Predictors	B	SE	Exp(B)	Lower bound	Upper bound
Gaming behavior
Gaming in terms of dysfunctional coping	1.22^**	0.10	3.39	2.78	4.13
Gaming as source of self-efficacy	0.45^**	0.11	1.57	1.28	1.93
Usage of MMORPG	0.58^**	0.20	1.78	1.20	2.65
Usage of strategy games	0.17^ns	0.21	1.19	0.79	1.80
Usage of shooter games	0.06^ns	0.23	1.06	0.67	1.67
Social predictors
No success in leisure-time activities besides gaming	1.60^**	0.32	4.94	2.65	9.18
School-related anxieties	0.54^**	0.14	1.72	1.31	2.25
Previous repetition of a school year	0.50^*	0.21	1.65	1.09	2.51
Demographic predictors
Male gender	0.39^ns	0.36	1.48	0.74	2.97
Low parental education	−0.28^ns	0.27	0.76	0.45	1.28
Low juvenile education	0.22^ns	0.22	1.24	0.81	1.92
Psychological predictors
Lower social competence	0.85^**	0.31	2.34	1.29	4.25
Impulsiveness	0.31^**	0.12	1.36	1.08	1.73
Acceptance of violence	0.27^*	0.12	1.31	1.04	1.64
ADHD in personal history	−0.02^ns	0.28	0.98	0.57	1.70
Depression disorder in personal history	−0.04^ns	0.41	0.96	0.43	2.13
Anxiety disorder in personal history	0.43^ns	0.46	1.54	0.63	3.77
Frequent physical abuse in childhood	0.59^ns	0.39	1.81	0.85	3.85

Nagelkerkes R² = 0.41.

p < 0.05; **p < 0.01; ns = not significant. All variables entered simultaneously.

As one would expect, behavioral variables gaming in terms of dysfunctional coping and gaming as a source of self-efficacy both increase the risk of VGD. Our assumption that the use of complex and often online-used games contributes to the risk of VGD is only partially supported, though: while MMORPGs contribute to the risk of developing VGD, strategy or shooter games do not. Our hypothesis that different social factors are of importance for VGD could fully be supported. Adolescents with no success in their leisure-time activities besides gaming, those with higher levels of school-related anxieties, and those who have repeated a school year in the past are significantly at higher risk of developing VGD. The two sociodemographic variables parental and juvenile educational background, on the other hand, provide no substantial contribution to predicting VGD. Hence, our hypothesis of VGD being independent of educational background and therefore not being restricted to specific social environments could be confirmed. Regarding male gender, our assumption could not be sustained. Although male participants are to a far greater extent affected by VGD, when simultaneously considering other predictors, male gender loses its importance in predicting VGD. Finally, our hypothesis of different psychological factors accounting for VGD could partially be supported. In this context, only subclinical predictors, such as low social competence, higher levels of impulsiveness, and the acceptance of violence as an inherent part of social life and part of one's own behavioral repertoire, contribute to VGD. The three considered psychological disorders ADHD, depression, and anxiety disorders do not significantly contribute to the risk of VGD. A trend was observed regarding frequent physical abuse in childhood, which remained statistically significant in a recently published model predicting VGD.²⁶ In the present model, however, considering a larger number of variables, abuse in childhood loses its relevance as a predictor.

Discussion

By means of the KFN schools survey 2007/2008, a first nationwide prevalence value of VGD for 15-year-olds was obtained. The resulting prevalence of 1.7% (boys: 3%; girls: 0.3%) is to be found at the lower end of other estimations, generally showing higher prevalence values in younger⁶ and similar age groups.^7,10 These studies, however, used different diagnostic instruments and operated with a comparatively small number of participants. In either case, the relatively lower prevalence values of this study should not be underestimated, since the observed number of affected adolescents points out an urgent call for public action: in Germany alone, in the age group of 15-year-olds (843,000 in 2007), one must assume about 13,000 video game–dependent boys and 1,300 video game–dependent girls.

With regard to the KFN-CSAS-II, its applicability for drawing a clear dividing line between VGD and a passionate, however comparatively less problematic, gaming behavior could be shown. Video game–dependent adolescents show characteristics of increased psychological and social stress, confirming the clinical relevance of the phenomenon as reported similarly in other studies.^8,10,19 Our study adds to the existing research in showing that school- and health-related stress indicators could be observed both in boys exhibiting conspicuous mean values in the KFN-CSAS-II and, to a minor degree, in boys showing just increased gaming times. This points out the importance of differentiating between extensive gaming without pathological background and VGD. However, further studies are required to validate the KFN-CSAS-II in clinical samples in order to estimate its diagnostic sensitivity and specificity and to give further advice for possible improvements.

Another aim of this study was to identify certain risk factors predicting VGD. Although our cross-sectional study design does not allow causal conclusions between many of the risk factors under consideration and VGD, we found the 10 most common video games among boys in 2007/2008 to differ largely in the percentage of players being at risk and dependent as well as the time spent with these games. Players of MMORPGs, strategy games, and shooters show increased means regarding these aspects. Our regression model, however, illustrates that when considering other important predictors, only the use of MMORPGs remains to add the risk of VGD. The risk of adolescents using these games increases by 78%, whereas the explanatory value of shooter and strategy games turns out to be insignificant in general. These findings are in line with other studies reporting especially MMORPG users being at higher risk of developing VGD.^4,17 In this regard, above all the MMORPG World of Warcraft must be assessed as critical, since in most countries it is classified as suitable for adolescents (European States covered by PEGI and USK rating: 12+; United States ESRB rating: teen), and 8.5% of its German 15-year-old male users are being diagnosed as dependent. More research to identify prominent features embedded in these games with a higher risk potential is needed in order to explain the existing differences and to sensitize the local authorities responsible for the protection of minors.

Additionally, our regression model points out that a multiplicity of risk factors is of importance for the development of VGD. To our knowledge, this is the first study modeling the risk of VGD simultaneously accounting for factors of gaming behavior, social stress, sociodemographic variables, and psychological predictors. Some of the identified predictors support other studies: importance of dysfunctional coping,^6,7,18 higher levels of acceptance of violence as an inherent part of social life and part of one's own behavioral repertoire,^4,20 and school-related behavioral problems.¹⁹ Furthermore, our results indicating a higher level of sleep disturbance and suicidal thoughts in video game–dependent boys are in line with other studies reporting elevated levels of psychological stress in persons with conspicuous video game usage.^8,10 ADHD, depression, and anxiety disorders in personal history, however, do not contribute to the risk of VGD in adolescence. Longitudinal studies are necessary to clarify the relationship between VGD and comorbid disorders, carefully considering cause and effect among developmental changes.

Finally, some findings of this study are contradictory to other studies. Our results prove VGD to be independent of educational background, which is consistent with previous results of a large-scale assessment of 14,301 ninth-graders conducted in Germany in 2005¹⁹ but inconsistent with another German study reporting lower educational background of video game–dependent boys.⁷ One explanation for this discrepancy could be the smaller sample size of 221 children, 14 of whom were identified as dependent,⁷ wherein the importance of educational factors in the population is hardly predictable. Another explanation could be that the authors included video gaming time as a selection criteria of VGD.⁷ Video gaming time is known to be usually higher in adolescents (especially in children) with lower educational background,⁴ which could be expected to lead to a higher chance for this group to be classified as dependent if gaming time is part of the diagnoses. In contrast, in our study, gaming time was not included in the diagnostic assessment of VGD. Moreover, in our regression analysis, we included only students playing at least occasionally in order to predict VGD and not video gaming in general. Therefore, we assume our risk assessment of VGD to be less confounded with usage time and therefore to deliver a better estimate for VGD among different social groups. Further research is needed to clarify these assumptions. In this regard, it seems noteworthy that in our regression model, failures at school, school-related anxieties, and the absence of success in other leisure-time activities besides gaming seem to play a crucial part in developing VGD. Thus, there is growing evidence that not differences in social status per se but the perception of one's own status within the peer group could contribute to an escape into virtual worlds and thereby to the risk of VGD.

We found that male adolescents are affected by VGD to a far greater extent, which again corresponds with other studies.^4,7 Accounting for other risk factors, however, the male gender is no longer of relevance in predicting VGD. Other variables included in the model, such as the more common use of MMORPGs and the greater extent of gaming in terms of dysfunctional coping and as a source of self-efficacy among boys explained gender influences on VGD. Some other factors, however, could account for a gender-specific risk of VGD as well. These have to be dealt with in future studies, especially with regard to other possible forms of media dependency in which girls seem to be far more prominent (e.g., excessive chatting behavior).

Despite a progressive level of knowledge, up until now, VGD has not been clinically acknowledged, so treatment of affected persons is not officially provided by most of the present health care systems. Further research is needed to advance our understanding of VGD and other possible types of media dependencies, especially regarding genesis, long-term consequences, risk and resilience factors, and adequate therapeutic strategies. However, a clarification of these questions can be achieved not only by intensified research efforts but also by a well-documented clinical practice and a growing cross-linking of all protagonists involved.

Footnotes

Acknowledgments

We thank the KFN schools survey team: Susann Rabold, Julia Simonson, Cathleen Kappes, Dirk Baier, and Christian Pfeiffer. We also thank Bert te Wildt (Medizinische Hochschule Hannover) for his support in the development of the KFN-CSAS-II.

Disclosure Statement

No competing financial interests exist.

References

Feierabend

, Klingler

. Was Kinder sehen. Eine Analyse der Fernsehnutzung Drei-bis 13-Jähriger [What Children watch: An Analysis of TV-Usage of 3 to 13 Year Olds] Media Perspektiven, 2008; 4:190–204.

Feierabend

, Rathgeb

. 2008. JIM-Studie 2008. Jugend, Information, (Multi-)Media. Basisuntersuchung zum Medienumgang 12–19 Jähriger [Youth, Information, (Multi-)Media. Survey of Media Usage of 12–19 Year Olds] Stuttgart: Medienpädagogischer Forschungsverbund Südwest.

JIM’98: Jugend, Information, (Multi-)Media: Basisuntersuchung zum Medienumgang 12-bis-19 Jähriger in Deutschland [Youth, Information, (Multi-)Media: Survey of Media Usage of 12–19 Year Olds], 1998 Baden-Baden: Medienpädagogischer Forschungsverbund Südwest.

Mößle

, Kleimann

, Rehbein

. 2007. Bildschirmmedien im Alltag von Kindern und Jugendlichen: Problematische Mediennutzungsmuster und ihr Zusammenhang mit Schulleistungen und Aggressivität [Screen Media in Everyday Life of Children and Adolescents: Problematic Usage Patterns and Their Relation to School Success and Aggressive Behavior] Baden-Baden: Nomos.

Rehbein

, Borchers

. Süchtig nach virtuellen Welten? Exzessives computerspielen und computerspielabhängigkeit in der jugend [Addicted to Virtual Worlds? Excessive Video Gaming and Video Game Addiction in Adolescents] Kinderärztliche Praxis, 2009; 80:42–9.

Grüsser

, Thalemann

, Albrecht

et al.

Exzessive computernutzung im kindesalter: ergebnisse einer psychometrischen Erhebung [Excessive Computer Usage in Adolescents–a Psychometric Evaluation]

Wiener Klinische Wochenschrift, 2005; 117:188–95.

Wölfling

, Thalemann

, Grüsser

. Computerspielsucht: Ein psychopathologischer Symptomkomplex im Jugendalter [Computer Game Addiction: A Psychopathological Complex of Symptoms in Adolescence] Psychiatrische Praxis, 2008; 35:226–32.

te Wildt

, Putzig

, Zedler

et al.

Internetabhängigkeit als ein symptom depressiver störungen [Internet Addiction As a Symptom of Depression Disorders]

Psychiatrische Praxis, 2007; 34:S318–22.

Young

, Pistner

, O'Mara

et al. The mental health concern for the new millennium. CyberPsychology & Behavior, 2000; 3:475–9.

10.

Yang

. Sociopsychiatric characteristics of adolescents who use computers to excess. Acta Psychiatrica Scandinavica, 2001; 104:217–22.

11.

Hahn

, Jerusalem

. 2001. Internetsucht: Validierung eines Instruments und explorative Hinweise auf personale Bedingungen [Internet Addiction: Validation of an Instrument and Explorative Evidence on Personal Causes] Theobald

, Dreyer

, Starsetzki

. Handbuch zur Online-Marktforschung. Beiträge aus Wissenschaft und Praxis. Wiesbaden: Gabler, 213–33.

12.

Meerkerk

, Van Den Eijnden

RJJM

, Vermulst

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

13.

Kunczik

, Zipfel

. 2004. Medien und Gewalt [Media and Violence] Osnabrück: Bundesministerium für Familie, Senioren, Frauen, und Jugend.

14.

Stierle

. Computerspiele und Gewalt: Eine Sekundäranalyse zum Zusammenhang zwischen violenten Computerspielen und Gewaltverhalten [Video Games and Violence: A Reanalysis of Violent Video Game Usage and Aggressive Behavior] Forum Kriminalprävention, 2007; 2:5–7.

15.

Zimbardo

, Gerrig

. 2004. Psychologie [Psychology] München: Pearson.

16.

Griffiths

. Relationship between gambling and video-game playing: a response to Johansson and Gotestam. Psychological Reports, 2005; 96:644–6.

17.

Smyth

. Beyond self-selection in video game play: an experimental examination of the consequences of massively multiplayer online role-playing game play. CyberPsychology & Behavior, 2007; 10:717–21.

18.

Wan

, Chiou

. Why are adolescents addicted to online gaming? An interview study in Taiwan. CyberPsychology & Behavior, 2006; 9:762–6.

19.

Baier

, Rehbein

. 2009. Computerspielabhängigkeit im Jugendalter [Video Game Dependency in Adolescence] Tuly

. Multilokalität und Vernetzung: Beiträge zur technikbasierten Gestaltung jugendlicher Sozialräume. München: Weinheim, 139–55.

20.

Grüsser

, Thalemann

, Griffiths

. Excessive computer game playing: evidence for addiction and aggression? CyberPsychology & Behavior, 2007; 10:290–2.

21.

American Medical Association. CSAPH Report 12-A-07: Emotional and behavioral effects of video games and Internet overuse. Presented at the AMA annual meeting 2007.

22.

Baier

, Pfeiffer

, Simonson

et al. 2009. Jugendliche in Deutschland als Opfer und Täter von Gewalt. Erster Forschungsbericht zum gemeinsamen Forschungsprojekt des Bundesministerium des Innern und des KFN [German Juveniles as Crime Victims and Offenders. First Research Report on a Joint Research Project of German Department of the Interior and Criminological Research Institute of Lower Saxony], 107. Hannover: KFN.

23.

Stadler

, Janke

, Schmeck

. 2004. IVE: Inventar zur Erfassung von Impulsivität, Risikoverhalten und Empathie bei 9-bis 14-Jährigen Kindern. Göttingen: Hogrefe.

24.

Comer

, Sartory

. 2001. Klinische Psychologie [Clinical Psychology] Heidelberg, Berlin: Spektrum.

25.

Schlack

, Hölling

, Huss

. Die Prävalenz der Aufmerksamkeitsdefizit-Hyperaktivitätsstörung (ADHS) bei Kindern und Jugendlichen in Deutschland [Prevalence of Attention Deficit Hyperactivity Disorder in German Children and Adolescents] Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz, 2007; 5/6:827–35.

26.

Rehbein

, Kleimann

, Mößle

. 2009. Computerspielabhängigkeit im Kindes- und Jugendalter. Empirische Befunde zu Ursachen, Diagnostik und Komorbiditäten unter besonderer Berücksichtigung spielimmanenter Abhängigkeitsmerkmale [Video Game Dependency in Childhood and Adolescence. Empirical Results on Causes, Diagnostic, and Comobid Conditions with Special Regard on Video Game Characteristics], 108. Hannover: KFN.