Violent Online Games Exposure and Cyberbullying/Victimization Among Adolescents

Abstract

This population-based cross-sectional survey examined the association between exposure to violent online games and cyberbullying and victimization in adolescents recruited from two large cities utilizing a stratified two-stage random cluster sampling technique. Cyberbullying and victimization were assessed by the E-victimization and E-bullying scales validated in a previous study. Exposure to violent online games was measured by self-nomination of the degree of violent content in the games played. Results indicated that the majority (74.3 percent) of respondents did not experience any cyberbullying or victimization in the last 7 days before the survey, 14.4 percent reported to be victimized via cyberspace, 2.9 percent admitted that they had bullied others, and 8.4 percent reported to be both perpetrators- and- victims. One hundred and eighty seven (15.3 percent) considered games they were playing were of moderate to severe violence. Students who had been involved in cyberbullying as well as being victimized were two times as likely to have been exposed to violent online games, and nearly four times as likely for those involved in bullying others. Exposure to violent online games was associated with being a perpetrator as well as a perpetrator-and-victim of cyberbullying. Parents and clinicians need to be aware of the potential harm of these exposures. The policy implications of results were also discussed.

Introduction

The relationship between exposure to violence and aggression has long been noted since the 60s.¹ In recent years, due to the rapid development of media technologies, particularly video games, the association between violent video games and aggression became a focus of research in areas of personality and social psychology as well as health. Theories and models about the relationship between violent video games and aggression have been put forward and studied.^2–4 In the last decade, many studies, including systematic reviews and meta-analytical studies, have been conducted to investigate the possible link between violent video games and aggression and were reported in the literature.^5–9 Of these, two reviews have found insufficient evidence for the association between violent video games playing and aggressive behavior in children and adults with one only focused on children with behavioral and emotional difficulties.^7,8 On the other hand, other reviews have concluded that there is sufficient evidence to support the notion of an association between playing violent video games and aggressive behavior.^5,6,9

Among adolescents, bullying, harassment, and victimization are terms used to describe aggressive behavior that can be characterized by an intention to harm, repetitiveness, and occur in an imbalanced power relationship.¹⁰ With the recent rapid development of cybertechnologies, aggressive behavior can also be expressed and manifested online resulting in the term “cyberbullying and victimization” being suggested as an equivalent phenomenon of that observed offline.¹¹ Since then, there has been a growing volume of work in the area of cyberbullying and victimization among adolescents. Unlike the well-established traditional bullying and victimization, information on the population-based prevalence of cyberbullying and victimization has not been forthcoming. Adopting the concept and definition of tradition bullying and translating online, the recent national population Health Behavior in School-Age Children survey conducted in the United States, suggested that 13.6 percent of young people were involved in or affected by cyberbullying. Of these, 3.8 percent were perpetrators only, 5.3 percent victims only, and 4.5 percent were both perpetrators as well as victims.¹² In another U.S. study on cyberharassment, it was reported that about 9 percent of youth were harassed online in the year before the survey.¹³ On the other hand, a Canadian study, utilizing the same definition of cyberbullying and a large and diverse sample of middle- and high-school students, found that nearly half (49.5 percent) indicated that they had been bullied online and about 34 percent had bullied others online.¹⁴ Such a discrepancy could be explained by differences in the methods of cyberbullying assessment. Nevertheless, these results indicated that cyberbullying and victimization is a common and growing problem among young people.

In terms of risk factors of cyberbullying and victimization among adolescents, information provided from the literature is scarce. There seemed to be a gender differential dependent on categories of bullying.¹² Young males were more likely to be involved in bullying only (odds ratio [OR]=1.73, 95% confidence interval [CI]=1.21–2.45) and less likely to be a victim only (OR=0.71, 95% CI=0.54–0.93) in comparison to the those who had never been involved in cyberbullying or been a victim.¹² Unlike traditional bullying, being overweight or underweight were not associated with cyberbullying and victimization.¹⁵ However, smoking and drinking were found to be associated with cyberbullying and victimization.¹⁶ Several psychosocial characteristics, such as hostility, anger, and being diagnosed with attention deficit and hyperactivity disorder (ADHD) or Asperger syndrome, were also suggested to be related to cyberbullying and victimization.^17–19 For the consequences of cyberbullying, the literature suggests that there are similarities with traditional bullying. In a large-scale Australian study among school children, it was demonstrated that cybervictimization is associated with symptoms of stress.²⁰ Other studies have also shown that being a victim of cyberbullying is associated with negative emotions, fear, and a greater sense of helplessness. ^21,22 For health outcomes, cyberbullying and victimization have been shown to be related to depression, anxiety, and self-harm.^23–25

In recent years, we have seen a proliferation of online games. Moreover, it has been suggested that the gaming experience of online games, particularly, the Massively Multiplayer Online Role-Playing Games, is different to that of the traditional offline video games.²⁶ This experience maybe the reason of attracting people to these games.²⁶ This has provided a reason for concern in terms of the effect of violent game exposure and online aggressive behaviors, particularly among young people. As aforementioned, there are sufficient evidence for the relationship between playing violent video games and aggressive behavior. It would be logical to deduce that violent online games may have a similar effect as offline violent video games on aggressive behavior and manifest as cyberbullying and victimization. A search in the related literature has revealed little information on the relationship between exposure to violent online games and cyberbullying and victimization among adolescents. Hence, the aim of this exploratory study is to bridge the gap in knowledge on the relationship between exposure to violent online games and cyberbullying and victimization among adolescents. It is hypothesized that exposure to violent online games is associated with cyberbullying and victimization.

Materials and Methods

This cross-sectional health survey was conducted in two main cities, Wuxi and Wuhu cities of the Jiangsu and Anhui Provinces in Northeast China in October 2011. Both are capital cities of their Provinces and are highly populated. There were about 90 and 25 high schools in these two cities, respectively. Institute ethics approval for the study was granted by the Wuxi Mental Health Institute, Jiangsu Academy of Psychological Sciences.

The sample consisted of high-school students aged between 13–18 years with the total student population attending high schools in the two cities as the sample frame. The entire list of high schools was obtained from local education departments of the two cities. The sample was generated using a two-stage random cluster sampling technique with stratification according to the population size of high-school students in these two cities. First, using individual schools as primary sampling unit, a number of schools were randomly selected with a probability, which is proportional to the size of the target population in each school. Second, using the class as the secondary sampling unit, different clusters of students were randomly selected from each of the selected schools. A sample size estimation was accordance to the study design.²⁷

The health survey was conducted within 2 weeks on campus at different schools. Students and their parents of the selected classes from different schools were informed of the survey with a written information letter. They were invited to participate in the study with a consent form signed by their parents or carers before the survey. During the survey, students were to fill in a self-reported questionnaire designed specifically for the study.

Cyberbullying and victimization were assessed using the E-victimization (E-VS) and E-bullying (E-BS) scales that were designed based on the conceptual model of the Aggression and Victimization Scale by Orpinas and Horne, but translating it to the online environment.²⁸ These two scales were designed as a self-reported instrument with 6 items each using a response set with a rating from 0 to 6 corresponding to a range of 0 times to 6 times or more was used. An example question asked the respondent to indicate: “In the past 7 days, how many times did someone threaten you using emails, texting, short messages, on a Web site such as Facebook, etc.?” Results obtained from a validation study suggested good internal reliability with Cronbach's alpha ranging from 0.55 to 0.96 that were largely not affected by sex. Correlations between E-VS and Centre for Epidemiological Studies-Depression for Children (CES-DC)²⁹ as well as the Zung's Anxiety Scales (SAS)³⁰ showed positive and significant relationships. Both CES-DC and SAS were validated and commonly used instruments with good internal reliability of 0.84 and 0.85, respectively. The test–retest reliability of CES-DC was moderate with an intra-class correlation of 0.54, and the split half reliability of SAS was estimated to be 0.71. Each scale was dichotomized into two categories as Involved in cyberbullying and Never for the E-BS and Being victimized and Never for the E-VS. There were two main reasons for dichotomized measure into categories. First, the literature clearly indicated that there are different categories of individuals involved in cyberbullying; namely, victims only, perpetrators only, and perpetrator-victims. Since each scale was designed to measure one specific construct, either victimization or bullying, it would be difficult to identify different cyberbullying types, particularly, the perpetrator-victims without dichotomizing the scales. Second, using a dichotomized scale for assessing cyberbullying was in consistent with the practice of previous studies in the area.^12–16 For the ease of comparisons of results with other studies, it would be prudent to use the same methodology for measuring the outcome variable. To satisfy the repetitive characteristic of bullying, the Never category contained those responded 0 or 1 time only to all items in the scales. A composite variable was also created from these two exposure variables with four groups; namely, Bully-Victim; Victim; Bully; and Nonbully-victim.

In terms of exposure to violent online games, a series of questions were included in the questionnaire to elicit information on students' cyberbehavior, including Internet access, involvement with online games, duration of playing online games, playing violent online games, and the most frequent games played online. Exposure to violent online games was assessed by the response to a specific question asking students to self-nominate the degree of violent content in the games they were playing. The degree of violence exposure was measured using a 5-point Likert scale ranging from 0 as Not violent at all to 5 as Extremely violent. For ease of analysis, the exposure measure was again dichotomized into No and little and Moderate and Severe exposures. To assess the validity of the exposure variable using a single question, the following procedures were implemented. The content of the reported games played online were examined by a panel of mental health experts using a 5-point rating scale ranging from nonviolent to extreme violent. The ratings of experts and the self-nominated degree of violent content were subject to being examined using Pearson correlation analysis. Results indicated a moderate and significant correlation between two scores (r=0.521, p<0.001). Due to the fact that the online gaming variables were highly correlated, only exposure to violent online games was included in the final analysis to avoid problems of colinearity.

Other information collected in the survey included demographics, whether the respondent was a single child, parental education levels, health behaviors, including body weight and height, alcohol drinking, parental drinking, and gambling. Other potential confounding factors of bullying and victimization, such as anger, hostility, ADHD status, and sense of belonging were also assessed by Anger and Hostility subscales of the Aggression Questionnaire and the DSM-IV Symptom subscale of the Conners Rating Scale-Revised (CRS-R).^31,32 These scales were also validated with good psychometric properties of a Cronbach's alpha of 0.82 and 0.94, respectively. The sense of belonging was assessed using the Sense of Belonging scale designed for the Program of International Student Assessment in 2000 (PISA).³³ Depression, anxiety, and intentional self-harm of students were assessed using the CES-DC and the Zung Self-Rating Anxiety Scale.^29,30

Data were analyzed using the Stata V10.0 statistical software program.³⁴ Since the study was of a cluster sampling design, data were set up with the survey design function utilizing the svy commands for handling the cluster sampling effect. The svy algorithm assigned appropriate weight to each data point in the dataset according to the study design using the sizes of the primary and secondary samples as weighting factors. Bivariate analyses were conducted to examine the unadjusted relationships between exposure to violent online games, other variables of interest, and cyberbullying and victimization. The majority of potential risk factors were categorical or ordinal by nature; thus, comparisons across groups were conducted. Equality of means among groups was examined using F-tests with adjustments for the cluster sampling design. As cyberbullying and victimization was a categorical variable with four subgroups, multinomial logistic regression modeling was then applied to investigate associations between violent online games exposure, selected potential risk factors, and cyberbullying and victimization with adjustment for the cluster sampling effect. For the inclusion of any variable in the initial regression model, the criteria of a bivariate association with p<0.20 was used. This was to ensure that all potential risk factors were included in the multiple regression models for model reduction. Multinomial logistic regression models were constructed with a backward elimination process aiming to derive a parsimonious model that contained significant variables only. A significance level of 5 percent was used for hypothesis testing. The interaction terms of significant variables retained in the model were then tested for their significance. A significance level of 1 percent was used for hypothesis testing of all interaction terms. This was to prevent possible type I error with the inclusion of the interaction terms as well as variables from which the interaction terms were derived.

Results

A total of 1,278 students responded to the survey providing usable information. This represented a response rate of 94 percent. The demographics of the sample were presented in Table 1 with a mean age of 14.7 (SD=0.9) and 619 (48.4 percent) males. Other characteristics, including cyberbullying and victimization of the respondents were also summarized in Table 1. In terms of cyberbullying and victimization, the majority of respondents did not experience any cyberbullying or victimization in the last 7 days (n=933, 74.3 percent), 184 (14.4 percent) respondents reported to be victimized via cyberspace, 31 (2.9 percent) admitted that they had bullied others, and 117 (8.4 percent) reported to be both bullies and victims. For exposure to online games, 486 (41.2 percent) indicated that they had spent less than 1 hour playing online games each day, 342 (29.2 percent) 1–2 hours, and 250 (21.5 percent) 3 hours or more. Of those who did play online games, 222 (26.0 percent) reported that they had mostly played combat and war games, 100 (10.7 percent) gambling games, and the majority played other games (n=562, 63.4 percent). In terms of the extent of violence of these online games, 187 (15.3 percent) considered the games they were playing to be of moderate to severe violence.

Table 1.

Frequency (Percent) ^a or Mean (SE) of Students' Demographics, Personal and Familial Risk Factors, Psychosocial Measures, Internet Access, Violence Online Game Exposure, and Health Outcomes by Cyberbullying and Victimization Status and Results on Comparisons Across Bullying and Victimization Status (n=1278)

	Cyberbullying and victimization status
Students' characteristics	Bully and victim (n=117)	Victim (n=184)	Bully (n=31)	Non-bully-victim (n=933)	Results
Demographics
City
Wuxi	51 (73.3)	95 (79.2)	21 (88.2)	495 (80.1)	χ²₃=4.37, p=0.143
Wuhu	66 (26.7)	89 (20.8)	10 (11.8)	438 (19.9)
Age group
<14 years	37 (29.4)	67 (35.5)	7 (20.3)	333 (31.7)	χ²₆=10.26, p=0.280
14–15 years	49 (45.7)	79 (41.9)	10 (32.9)	370 (41.1)
16 years or older	31 (24.9)	38 (22.7)	14 (46.7)	226 (27.3)
Sex
Male	62 (50.9)	84 (42.4)	16 (46.1)	457 (50.2)	χ²₃=3.98, p=0.408
Female	55 (49.1)	100 (57.6)	15 (53.9)	476 (49.8)
Single child
Yes	73 (75.1)	121 (69.6)	21 (76.5)	662 (75.9)	χ²₃=3.19, p=0.393
No	42 (24.9)	61 (30.4)	8 (23.5)	261 (24.1)
Father's education level
Primary or lower	14 (12.9)	32 (14.7)	3 (9.6)	116 (9.2)
Middle school	56 (38.1)	81 (42.3)	14 (40.7)	383 (35.7)	χ²₉=14.23, p=0.433
High school/college	39 (40.6)	54 (31.2)	10 (39.0)	321 (40.0)
University or higher	8 (8.4)	17 (11.8)	4 (10.8)	112 (15.0)
Mother's education level
Primary or lower	36 (20.8)	60 (27.2)	7 (23.4)	232 (18.1)
Middle school	49 (40.5)	76 (41.8)	15 (44.9)	376 (39.4)	χ²₉=17.23, p=0.170
High school/college	26 (32.2)	37 (24.2)	4 (16.8)	262 (33.9)
University or higher	6 (6.7)	11 (6.8)	5 (15.0)	61 (8.6)
Risk factors
BMI
Under weight	47 (42.5)	82 (45.3)	9 (31.8)	406 (40.2)	χ²₆=5.48, p=0.528
Normal	64 (54.8)	88 (50.5)	20 (59.8)	459 (53.7)
Over weight/obese	4 (2.7)	7 (4.2)	2 (8.4)	45 (6.1)
Alcohol consumption in the last 3 months
Yes	29 (29.3)	21 (13.4)	11 (31.1)	109 (13.1)	χ²₃=27.25, p<0.001
No	88 (70.7)	162 (86.6)	20 (68.9)	822 (86.9)
Parents binge drank in the last 3 months
Yes	33 (23.7)	30 (17.8)	2 (2.4)	126 (10.2)	χ²₃=25.40, p<0.001
No	84 (76.3)	154 (82.2)	29 (97.6)	807 (89.9)
Parents gambled in the last 3 months
Yes	29 (20.0)	306(17.4)	5 (8.9)	157 (12.6)	χ²₃=7.26, p=0.059
No	88 (80.0)	148 (82.6)	26 (91.1)	776 (87.4)
Psychosocial measures
Anger	15.7 (0.50)	14.8 (0.39)	14.4 (1.15)	13.1 (0.19)	F_(3,26)=21.89, p<0.001
Hostility	16.9 (0.67)	14.5 (0.50)	13.4 (0.57)	13.2 (0.19)	F_(3,26)=16.90, p<0.001
ADHD score	13.4 (1.08)	9.6 (0.56)	9.9 (1.17)	8.5 (0.41)	F_(3,26)=8.16, p<0.001
Sense of belonging	18.5 (0.42)	18.6 (0.29)	19.1 (0.79)	19.2 (0.15)	F_(3,26)=4.74, p=0.009
Internet access, online gaming, and violence exposure
Owing a computer at home
Yes, use by myself	39 (38.6)	54 (37.2)	15 (50.9)	215 (27.5)
Yes, used by family members	54 (49.1)	98 (51.4)	12 (38.3)	489 (56.0)	χ²₆=19.39, p=0.025
No	22 (12.3)	30 (11.4)	4 (10.8)	217 (16.5)
Hours of playing online game/day
>3 hours	37 (34.3)	41 (24.4)	10 (26.9)	158 (19.1)
1–2 hours	39 (34.1)	53 (27.9)	11 (37.1)	238 (28.9)	χ²₉=32.66, p=0.029
<1 hour	33 (29.5)	79 (43.8)	10 (36.0)	358 (41.9)
No access	5 (2.1)	6 (3.9)	0 (0.0)	120 (10.0)
Types of online game played
Combat/war games	23 (21.6)	25 (13.8)	10 (33.3)	161 (21.2)
Gambling games	13 (13.1)	14 (7.9)	3 (6.6)	69 (8.0)	χ²₉=18.02, p=0.165
Other games	59 (50.8)	102 (58.8)	15 (51.5)	380 (48.3)
Did not play games	14 (14.6)	33 (19.6)	2 (8.5)	173 (22.5)
Exposure to violent online games
Moderate/severe	37 (33.0)	30 (13.6)	11 (33.3)	108 (12.8)	χ²₃=35.59, p<0.001
No/little	74 (67.0)	142 (86.4)	17 (66.7)	659 (87.2)
Health outcomes
Ever committed self harm in the last 3 months
Yes	36 (33.3)	38 (22.7)	8 (24.5)	111 (11.5)	χ²₃=47.35, p<0.001
No	80 (66.7)	145 (77.3)	23 (75.5)	821 (88.5)
Depression
Yes	74 (60.9)	101 (58.8)	10 (29.9)	368 (40.7)	χ²₃=34.51, p<0.001
No	39 (39.1)	78 (41.2)	21 (70.1)	542 (59.3)
Anxiety
Yes	17 (15.2)	20 (12.7)	3 (9.6)	50 (5.5)	χ²₃=21.57, p=0.049
No	99 (84.8)	159 (87.3)	28 (90.4)	867 (94.5)

Due to rounding, not all total percentages were added to 100 percent; percentages were adjusted for cluster sampling effect.

ADHD, attention deficit and hyperactivity disorder.

The bivariate relationships between cyberbullying and victimization, students' demographics, health conditions and behaviors, Internet access, online game playing, and exposure to violent online games, as well as health outcomes were examined. The results were also summarized in Table 1. As shown, without adjustment for other variables, cyberbullying and victimization were significantly associated with a number of variables. These included exposure to violent online games, duration of playing online games, alcohol consumption in the last 3 months, parents drunk in the last 3 months, anger, hostility, ADHD, and a sense of belonging. In terms of detrimental health effects, there were also unadjusted significant associations between cyberbullying/victimization and depression (χ²₃=34.51, p<0.001) and intentional self-harm (χ²₃=47.35, p<0.001). Anxiety was marginally associated with cyberbullying/victimization (χ²₃=21.57, p=0.049). These variables, except for the health outcomes, were included in further analyses. Cities, maternal education levels, and parental gambling were also selected due to the reason that associations between these variables and cyberbullying and victimization attained a significance level ≤0.20. Other Internet-related variables were not included in further analyses due to the aforementioned reasons.

The results obtained from the multinomial logistic regression analyses were presented in Table 2. Five variables remained significant in the final model after adjusting for each other. The results suggested that, after adjusting for other potential confounding factors, including time spent online, moderate to severe exposure to violent online games was significantly associated with bullying–victimization (t=2.93, p=0.007) and bullying only (t=2.64, p=0.014), but not with victimization only. Students who had been involved in cyberbullying as well as being victimized were two times as likely to have been moderately to severely exposed to violent online games (OR=2.04, 95% CI=1.24–3.35). For those who had been involved in bullying others via cyberspace, it was nearly 4 times as likely that they had been exposed to violent online games moderately or severely (OR=3.60, 95% CI=1.33–9.72). An examination of the interaction terms of the exposure variable and other potential confounding variables indicated that none were significant.

Table 2.

Odds Ratios (95% Confidence Interval) and Significance Obtained from the Multinomial Logistic Regression of Violence Exposure, Other Potential Confounding Factors on Cyberbullying and Victimization

Variables in the final model	OR (95% CI) ^a	Significance
Cyberbullying and victimization
Violence exposure	2.20 (1.36–3.55)	t =3.36, p =0.002
Parental binge drinking	2.16 (1.20–3.88)	t =2.69, p =0.012
Anger	1.04 (0.97–1.12)	t=1.20, p=0.239
Hostility	1.07 (1.01–1.14)	t =2.36, p =0.026
ADHD	1.04 (0.99–1.09)	t=1.69, p=0.102
Cyberbullying
Violence exposure	3.75 (1.50–9.35)	t =2.96, p =0.006
Parental binge drinking	0.22 (0.05–1.00)	t=−2.09, p=0.050
Anger	1.11 (0.95–1.29)	t=1.42, p=0.166
Hostility	0.94 (0.88–1.00)	t=−2.04, p=0.051
ADHD	1.02 (0.95–1.08)	t=0.46, p=0.651
Cybervictimization
Violence exposure	0.99 (0.45–2.19)	t=−0.02, p=0.986
Parental binge drinking	1.69 (1.05–2.74)	t =2.25, p =0.033
Anger	1.09 (1.05–1.13)	t =5.23, p <0.001
Hostility	1.02 (0.95–1.09)	t=0.57, p=0.577
ADHD	0.99 (0.96–1.03)	t=−0.28, p=0.782

Bold indicates significant results.

Non-bully-victim group as the referent group; ORs were calculated with adjustment to cluster sampling effect.

OR, odds ratio; CI, confidence interval.

Discussion

This study aims to examine the relationship between exposure to violent online games and cyberbullying and victimization among a population of young people in Northeast China. The results suggested that moderate to severe exposure to violent online games was significantly associated with being a perpetrator and victim as well as a perpetrator only of cyberbullying. The results indicated there were no significant associations between exposure to online game violence and being a victim of cyberbullying.

Due to the lack of a similar study, a comparison of results is difficult. However, the point estimate of cyberbullying and victimization obtained from this study could be compared to that reported in the literature, particularly from studies conducted in the United States.^12,13 For each subtype of bullying and victimization, the pattern of distribution remains similar with a larger proportion of victims of bullying at 14.4 percent, followed by perpetrators-victims of 8.4 percent, with only 2.9 percent of perpetrators only. Also, in consistent with the literature are the associations between cyberbullying and victimization and depression and intentional self-harm. The results of this study provided further evidence of the detrimental effects of cyberbullying and victimization on the mental health among young people.

There could be many explanations for the associations between exposure to violent online games and cyberbullying and victimization. The most intuitive one is that exposure to violence online games, as to other violence through media, will enhance the aggressive tendency of an individual.⁹ Anderson and Bushman have put forth the General Aggression Model (GAM) for the formation of the long-term effects of exposure to video game violence.⁵ The model stipulates that repeated violent game playing will lead to an increase in aggressive belief and attitude; enhancement of aggressive perceptual schemata; deepening of aggression desensitization; and thus, an increase in aggressive personality. This, in turn, will manifest as aggressive behavior when the environmental trigger is available.⁵ The results obtained from this study renders some support to the GAM for the following two reasons. First, exposure to violent online games is significantly associated with anger and hostility, and both are measures of an aggressive personality. Second, exposure to violent online games has increased the odds of being a perpetrator as well as a perpetrator-victim of cyberbullying, but not for victims of cyberbullying. One remaining issue in this explanation is the interaction terms between exposure and anger, as well as exposure and hostility as are both insignificant in the model. These results seem to suggest a direct effect of violent online game exposure on cyberbullying behavior after adjusting for anger and hostility, but not mediated by aggressive personality. This is worthy of further investigation.

The results obtained from this study have a direct implication for the detection as well as the prevention of mental problems among young people. The results of this study suggest that cyberbullying is common among Chinese high-school students as those in the United States. Parents, carers, teachers, and school principals should be aware of the potential harmful effects of the exposure to violent online games on the psychosocial aspects of children's cognitive and emotional development. This can be achieved through greater information sharing via the media and the school system. Psychological distress and depressive symptoms might be a signal of possible involvement of cyberbullying and victimization. The results also have implications on policy development. It has been noted by the Australian government that censorship laws and regulations on Internet materials, particularly relating to violence, vary across many developed countries that render them ineffective in protecting children and young people on the Internet.³⁵ Should the potential causal link between exposure to online violent games and cyberbullying and victimization be demonstrated, this information could possibly be utilized to facilitate evidence-based policy making. It could be used to support the argument for international regulations on violent online game censorship.

As in all studies, there are strengths and weaknesses in this study. This is a population-based study that includes a random sample of students from two large cities utilizing a two-stage cluster random sampling technique. An appropriate statistical analytical approach has been used to adjust for the effect of cluster sampling. The use of a standardized and validated assessment instrument for cyberbullying minimized some measurement biases. Some potential limitations have also been identified in this study. First, information on violence exposure is collected using a single question via self-reporting. It is an assessment of self-perception of violence exposure, not a quantitative measure of real life exposure. Hence, this will constitute an assessment bias in the exposure variable although it would most likely be a nondifferential bias. Second, a cross-sectional study could be considered as an appropriate design for exploring potential risk factors for a condition or disease. However, the evidence provided from such a study can only be considered as associative and it is insufficient to draw any causal inference.³⁶ This study can be considered as an exploratory study to identify the potential risk factors for cyberbullying and victimization among adolescents. Future studies could be conducted with a better design, such as a longitudinal cohort study, to elucidate whether the association is of a causal nature. To improve the assessment of exposure to violent online games, it would be prudent to employ a more objective method, such as utilizing content descriptor ratings of games provided by the U.S. Entertainment Software rating Board (ESRB).

Footnotes

Author Disclosure Statement

All authors ensure that there is no conflict of interest of any kind involved in the production of this manuscript nor is the study associated with any commercial bodies. No competing financial interests exist.

References

Berkowitz

, LePage

. Weapons and aggression-eliciting stimuli. Journal of Personality and Social Psychology, 1967; 7:202–207.

Bandura

. 1973. Aggression: a social learning analysis. Englewood Cliffs, NJ: Prentice-Hall.

Anderson

, Dill

. Video games and aggressive thoughts, feelings, and behaviour in the laboratory and in life. Journal of Personality and Social Psychology, 2000; 78:772–790.

Anderson

, Bushman

. Human aggression. Annual Review of Psychology, 2002; 53:27–31.

Anderson

, Bushman

. Effects of violent video games on aggressive behaviour, aggressive cognition, aggressive affect, physiological arousal, and prosocial behaviour: a meta-analytic review of the scientific literature. Psychological Science, 2001; 12:353–359.

Huesmann

. The impact of electronic media violence: scientific theory and research. Journal of Adolescent Health, 2007; 41,6 Suppl:S6–S13.

Mitrofan

, Paul

, Spencer

. Is aggressive in children with behavioural and emotional difficulties associated with television viewing and video game playing? A systematic review. Child: Care, Health, and Development, 2008; 35:5–15.

Ferguson

, Kilburn

. The public health risk of media violence: a meta-analytic review. Journal of Pediatrics, 2009; 154:759–763.

Anderson

, Ihori

, Bushman

et al. Violent video game effects on aggression, empathy, and prosocial behaviour in Eastern and Western countries: a meta-analytic review. Psychological Bulletin 2010, 136; 151–173.

10.

Eisnberg

, Aalsma

. Bullying and peer victimization: position paper of the Society for Adolescent Medicine. Journal of Adolescent Health, 2005; 36:88–91.

11.

. New bottle but old wine. A research of cyber bullying in schools. Computer and Human Behaviour, 2007; 23:1777–1791.

12.

Wang

, Iannotti

, Nansel

. School bullying among US adolescents: physical, verbal, relational and cyber. Journal of Adolescent Health, 2009; 45:368–375.

13.

Wolak

, Mitchell

, Finkelhor

. Does online harassment constitute bullying? Journal of Adolescent Health, 2007; 41:S51–S58.

14.

Mishna

, Cook

, Gadalla

et al. Cyber bullying behaviours among middle and high school students. American Journal of Orthopsychiatry, 2010; 80:362–374.

15.

Wang

, Iannotti

, Luk

. Bullying victimization among underweight and overweight U.S. youth: differential associations for boys and girls. Journal of Adolescent Health, 2010; 47:99–101.

16.

Vieno

, Gini

, Santinello

. Different forms of bullying and their association to smoking and drinking behaviour in Italian adolescents. Journal of School Health, 2011; 81:393–399.

17.

Pornai

, Wood

. Peer and Cyber aggression in secondary school students: the role of moral disengagement, hostile attribution bias, and outcome expectancies. Aggressive Behavior, 2010; 36:81–64.

18.

Sontag

, Clemans

, Graber

et al. Traditional and cyber aggressors and victims: a comparison of psychosocial characteristics. Journal of Youth Adolescence, 2011; 40:392–404.

19.

Kowalski

, Fedina

. Cyber bullying in ADHD and Asperger syndrome populations. Research in Autism Spectrum disorders, 2011; 5:1201–1208.

20.

Rigby

. Peer Victimisation at school and the health of secondary school students. British Journal of Educational Psychology, 1999; 69:95–104.

21.

Ortega

, Elipe

, Mora-Merchan

et al. The emotional impact of victims of traditional bullying and cyberbullying: a study of Spanish adolescents. Journal of Psychology, 2009; 217:197–204.

22.

Spears

, Slee

, Owens

et al. Behind the scenes and screens: insights into the human dimension of covert and cyberbullying. Journal of Psychology, 2009; 217:189–196.

23.

Wang

, Nansel

, Iannotti

. Cyber and traditional bullying: differential association with depression. Journal of Adolescent Health, 2011; 48:415–417.

24.

Perren

, Dooley

, Shaw

et al. Bullying in school and cyberspace: associations with depressive symptoms in Swiss and Australian adolescents. Child and Adolescent Psychiatry and Mental Health, 2010; 4:28. www.capmh.com/contnet/4/1/28. 2012 Jan 10.

25.

Hay

, Meldrum

. Bullying victimization and adolescent self-harm: testing hypotheses from general strain theory. Journal of Youth Adolescence, 2010; 39:446–459.

26.

Kelly

. 2004. Massively multiplayer online role-playing games: the people, the addiction and the playing experience. Jefferson, NC: Mcfarland & Co, Inc.

27.

Kish

. 1965. Survey sampling. New York: John Wiley and Sons.

28.

Orpinas

, Horne

. 2006. Creating a positive school climate and developing social competence. Washington, DC: American Psychological Association.

29.

Weissman

, Orvaschel

, Padian

. Children's symptom and social functioning self-report scales: comparison of mothers' and children's reports. Journal of Nervous and Mental Diseases, 1980; 168:736–740.

30.

Zung

. A self-rating depression scale. Archive of General Psychiatry, 1965; 12:63–70.

31.

Buss

, Perry

. The aggression questionnaire. Journal of Personality and Social Psychology, 1992; 63:452–459.

32.

Conners

. 2000. Conners' rating scales-revised. New York: Multi-Health Systems, Inc.

33.

Willms

. Student engagement at school: a sense of belonging and participation. Results from PISA, 2000. OECD, 2003. www.oecd.org/dataoecd/42/35/33689437.pdf. 2012 Jan 15.

34.

StataCorp. 2007. Stata statistical software: release 10.0. College Station, TX: Stata Corporation.

35.

Electronic Frontiers Australia. 2002. Internet censorship: law and policy around the world. www.efa.org.au/Issues/Censor/cens3.html. 2012 Jan 17.

36.

Rothman

, Greenland

. 1988. Morden epidemiology, 2nd. Philadelphia, PA: Lippincott Williams & Wilkins.