Computers in Human Behavior 63 (2016) 650e658
Contents lists available at ScienceDirect
Computers in Human Behavior
journal homepage: www.elsevier.com/locate/comphumbeh
Video games as virtual teachers: Prosocial video game use by children
and adolescents from different socioeconomic groups is associated
with increased empathy and prosocial behaviour
Brian Harrington*, Michael O’Connell
School of Psychology, University College Dublin, Newman Building, Belfield, Dublin 4, Ireland
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 December 2015
Received in revised form
3 April 2016
Accepted 23 May 2016
Available online 8 June 2016
Objective: The main aim of this study was to determine if there was a positive relationship between
prosocial video game use and prosocial behaviour in children and adolescents.
Method: This study had a cross-sectional correlational design. Data were collected from 538 9e15 year
old children and adolescents between March and December 2014. Participants completed measures of
empathy, prosocial behaviour and video game habits. Teachers rated the prosocial behaviour of participants. The socioeconomic status of participants was also gathered.
Results: Multiple linear regressions were conducted on these data. Prosocial video game use was positively associated with the tendency to maintain positive affective relationships, cooperation and sharing
as well as empathy. This association remained significant after controlling for gender, age, school type
(disadvantaged/non-disadvantaged), socioeconomic status, weekly game play and violent video game
use.
Conclusions: These findings provide evidence that prosocial video game use could develop empathic
concern and improve affective relationships in a diverse population of young people.
© 2016 Elsevier Ltd. All rights reserved.
Keywords:
Prosocial video game use
Violent video game use
Prosocial behaviour
Empathy
Socioeconomic status
Young people
1. Introduction
The increasing engagement of young people with media
including video games is well documented (Rideout, Foehr, &
Roberts, 2010). Computer and video game sales in the US have
risen from 7 billion dollars in 2003 to 15.4 billion dollars in 2014
(Entertainment Software Association, 2015). Some researchers
have suggested that video games could be used as teaching resources in schools as these games are based on learning principles
that allow players to be producers rather than consumers (Gee,
2003). In this context the use of video games in both educational
and clinical settings has received attention recently from researchers (Granic, Lobel, & Engels, 2014).
Anderson and Bushman (2001) ask if it is possible to create
engaging video games “to teach and reinforce nonviolent solutions
to social conflicts” (Anderson & Bushman, 2001, p.359). According
to researchers in this area, a prosocial video game is a game in
* Corresponding author.
E-mail addresses: brian.harrington@ucdconnect.ie (B. Harrington), michael.f.
oconnell@ucd.ie (M. O’Connell).
http://dx.doi.org/10.1016/j.chb.2016.05.062
0747-5632/© 2016 Elsevier Ltd. All rights reserved.
which the player must help and cooperate in order to succeed.
Examples of games with these characteristics that have been used
in previous research are Animal Crossing, Super Mario Sunshine, Zoo
Vet and Lemmings (Gentile et al., 2009; Greitemeyer & Osswald,
2010)1.
The General Learning Model (GLM) (Gentile et al., 2009) proposes that each experience (eg. playing a video game) an individual
has is a learning trial which temporarily alters cognitions, emotions
and levels of physiological arousal. The GLM proposes that two
short-term processes explain prosocial video game effects. Firstly,
the cognitive effect of priming scripts predicts that games with
1
Previous studies have used prosocial video games such as Lemmings
(Greitemeyer & Osswald, 2010) in which there is no violence and the player performs prosocial acts such as protecting a lemming from harm. However, content
analysis of 33 best-selling video games found that 79% of these games had some
form of violent content (Dietz, 1998). Therefore as games with only prosocial
content are less common, the present study uses the variable ‘prosocial video game
use’ to refer to prosocial behaviour within a game and ‘violent video game use’ to
refer to violent behaviour within a game. For example, in the game Minecraft it is
possible to cooperate with other players and construct buildings; however it also
possible to fight creatures.
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
prosocial content will result in prosocial behavioural scripts being
primed and rehearsed. Secondly, changes in cognitions, feelings
and levels of physiological arousal while playing a prosocial video
game are reciprocally reinforced through both classical and operant
conditioning.
Repeated practice of video games could produce certain longterm effects such as changes to precognitive and cognitive constructs, cognitive-emotional constructs and affective traits. This
model when applied to prosocial video game use predicts that a
game which requires the player to use prosocial behaviours to
succeed will create an increase in prosocial behaviours in the player
immediately following completion of the game. The repeated
practice producing these short-term effects could change personality traits in the individual playing prosocial video games.
Conversely the amount of time spent playing violent video games
could result in long-term aggressive behaviour according to the
learning mechanism described in this model (Gentile et al., 2009).
A recent meta-analysis has provided evidence that video games
have social outcomes (Greitemeyer & Mugge, 2014). This metaanalysis and other recent studies have concluded that violent
video game use leads to desensitization and aggression while
prosocial video game use increases empathy and therefore prosocial behaviour (Gentile et al., 2009; Greitemeyer & Mugge, 2014;
Prot et al. 2014; Gentile, Khoo, Prot & Anderson, 2014).
1.1. Empathy and prosocial video game use
Researchers have suggested that the relationship between prosocial video game use and prosocial behaviour could be mediated
by empathy as opposed to accessibility to prosocial thoughts
(Bartlett & Anderson, 2013). Previous correlational research into
prosocial video game effects in children and adolescents has found
a significant positive association between prosocial video game use
and empathy (Gentile et al., 2009). A recent longitudinal study
found that prosocial video game use was a significant predictor of
prosocial behaviour and that this change was mediated by empathy
(Prot et al., 2014). Therefore in the context of previous research it is
reasonable to expect that prosocial video game use should be
positively associated with empathy.
1.2. Theoretically relevant confounding variables such as
sociodemographic factors and weekly game play
Research has shown that when controlling for long-term causal
factors for aggressive behaviour, such as personality and environmental factors, violent video game effects can disappear (Ferguson,
San Miguel, Garza, & Jerabeck, 2012). Therefore in the case of
prosocial video game effects, it is theoretically possible that when
controlling for sociodemographic factors and weekly gameplay that
prosocial video game effects could disappear.
The following independent variables could theoretically explain
part of the variance in prosocial behaviour: age, gender, socioeconomic status (SES), school status and weekly gameplay. The relationship between age and prosocial behaviour has been extensively
studied ranging from the impact of adverse childhood experiences
on prosocial behaviour (Caprara & Pastorelli, 1993) to factors
influencing the development of prosocial behaviours in childhood
and adolescence (Eisenberg & Mussen, 1989). Gender differences in
prosocial behaviour have focussed on the agentic theory of male
gender role models (Eagly & Crowley, 1986) as well as differences in
the levels of prosocial behaviour in male and female children
(Calvo, Gonzalez, & Martorell, 2014). While experimental research
found that lower levels of social status were associated with higher
levels of prosocial behaviour (Guinote, Cotzia, Sandhu &Siwa,
2015), clinical and developmental psychologists have noted the
651
difficulty that parents in socially disadvantaged communities have
in reinforcing prosocial behaviours in their children (Kazdin, 1987).
A study examining the effect of family, school and classroom
ecologies on children’s social and emotional development found
that first grade children who attended schools in disadvantaged
communities had lower levels of prosocial behaviour (Hoglund &
Leadbetter, 2004). Screen time in the form of weekly game play
has also been found to be negatively associated with prosocial
behaviour (Gentile et al., 2009). Therefore weekly game play could
also explain some of the variance in prosocial behaviour.
If the relationship between prosocial video game use and prosocial behaviour remains significant after controlling for the
abovementioned theoretically relevant independent variables it
could be argued that this provides stronger evidence for a prosocial
video game effect (Prot & Anderson, 2013).
1.3. Violent video game use and prosocial behaviour
Numerous studies have identified relationships between violent
video game use and aggressive behaviour (Anderson et al., 2010;
Gentile et al., 2014). There have also been a number of studies
suggesting that violent video game use is associated with decreases
in prosocial behaviour (Anderson et al., 2010; Gentile et al., 2009).
Therefore, based on previous research, it is reasonable to expect
that violent video game use will be negatively associated with
prosocial behaviour in children and adolescents.
1.4. The present study
Previous studies into violent and prosocial video game effects
have generally accessed normative community-based samples
(Anderson, Gentile, & Buckley, 2007). Boxer, Huesmann,
Bushman, O’Brien and Moceri (2008) sought to address this
deficit in relation to violent media effects by including a sample
of juvenile deliquents in a study into the relationship between
violent media use and involvement in violent acts. In addition
numerous studies have investigated the video game use of specific clinical samples such as individuals with Autistic Spectrum
Disorder (Mazurek & Engelhardt, 2013). A recent study investigated the role of low educational ability as a risk factor for
playing violent video games (Bijvank, Konijn, & Bushman, 2012).
Prot et al. (2014) note that in studies investigating prosocial
video game effects in both children and adolescents that socioeconomic status (SES) and parental education were measured.
However, neither of these variables were controlled for in the
statistical analysis. Therefore to our knowledge this is the first
correlational study into prosocial video game effects to statistically control for both SES and school status.
The present study primarily aimed to determine if there was a
positive relationship between prosocial video game use and prosocial behaviour in children and adolescents. In addition three
related objectives were pursued. These objectives related to theoretically relevant variables that were identified based on a review of
the literature on both video game effects and prosocial behaviour.
These three objectives were as follows:
Objective 1 : To determine if prosocial video game use was positively associated with empathy in children and
adolescents.
Objective 2 To determine if the relationship between prosocial
video game use and prosocial behaviour remained
significant after controlling for theoretically relevant
variables such as sociodemographic variables and
weekly game play.
652
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
Objective 3 To determine if there was a negative relationship between violent video game use and prosocial behaviour
in children and adolescents.
2. Material and methods
2.1. Participants
The population under study were students (N ¼ 538) from 10
schools in the Republic of Ireland. Participants ranged in age from 9
to 15 years, (M ¼ 11.6 years, SD ¼ 1.44). There were more males
(n ¼ 315 (59%)) than females (n ¼ 223 (41%)) in the sample. Five of
the schools in the sample were co-educational, two of the schools
were all male schools and three of the schools were all female
schools.
Four of these schools were located in a city in western Ireland,
while the remaining six schools were located in a city in eastern
Ireland. Five of these schools could be described as socioeconomically disadvantaged, based on either a formal DEIS2 rating
(four) or in one case, based on the analysis of professionals
working with the school. In addition to five socioeconomically
disadvantaged schools (two primary, three post-primary), a private primary school, two Gaelscoileanna, an Educate Together
primary school and a mainstream post-primary school were
included in this sample3.
Snowballing or chain referral was used to access this sample.
Snowball sampling is a technique widely used to reach populations
that are generally difficult to access (Biernacki & Waldorf, 1981).
2.2. Procedure
Data were collected between March and December 2014.
The average overall response rate was 52% (range from 83% to
17%). The average response rate in the disadvantaged schools
was 49% while the average response rate in the nondisadvantaged schools was 53%. The response rates for teacher
questionnaires was 97%. The response rate for the measure of SES
from parents/guardians who consented for their child to participate was 75%.
Participants completed measures of computer/video game
habits, empathy and prosocial behaviour. The researcher administered a battery of these questionnaires to each class group in the
participating schools. Participants also received a glossary
explaining potentially difficult words and phrases in the questionnaires. Class teachers were also asked to rate the prosocial
behaviour of the participating students in their class group. Details
regarding the measures used in this study are provided in the
following section.
2
DEIS schools are schools in the Republic of Ireland that are designated disadvantaged and therefore allocated additional resources. The DEIS initiative used the
following definition of educational disadvantage in the Education Act (1998) to
guide the implementation of this project when it first began in 2005: “… the impediments to education arising from social or economic disadvantage which prevent students from deriving appropriate benefit from the education in schools”
(https://www.education.ie).
3
A large number of the primary and post-primary schools in the Republic of
Ireland are under the patronage of the Catholic church and are funded by the
Department of Education and Skills. The main language of instruction is English.
However there are a number of exceptions. A Gaelscoil is a school in which all the
instruction is carried out through the Irish language (Irish Gaelic). An Educate
Together school is a multi-denominational school. The Department of Education
and Skills pays the salaries to teachers in Gaelscoilenna and Educate Together
Schools. A private school in the Republic of Ireland is self-funded and does not
receive funding from the Department of Education and Skills (www.education.ie).
2.3. Measures
2.3.1. Prosocial video game use
Computer/video game habits were measured using an adapted
version of the Computer/Video Game Habits Questionnaire (Prot
et al., 2014). In order to measure prosocial video game use the
following procedure was adopted. Participants named the three
games that they played most frequently. Participants chose one day
during the week and 1 day at the weekend (Example: Pick one day
during the week: (Monday, Tuesday, Wednesday, Thursday or
Friday)______________. How many hours do you play this game on
that day?). Participants were given a choice of times from None to
More than 10 hours. Participants completed two items that rated
each game on a 4-point Likert scale from Never to Almost Always.
(Example: “How often do you help others in this game?”). Responses were coded from 0 (Never) to 3 (Almost Always). Participants’ reports of hours gaming during the week were multiplied by
five while reports of weekend gaming were multiplied by two in
order to calculate total weekly hours playing a particular game.
Total weekly hours were then multiplied by the video game ratings
to compute a score for weekly prosocial video game use. Prosocial
video game scores were then divided by three to obtain an average
prosocial video game score. This average score was then used as the
variable ‘prosocial video game use’ in data analysis. 4
2.3.2. Violent video game use
In order to measure violent video game use participants
completed two items that rated the violent content of each game on
a 4-point Likert scale from Never to Almost Always (Example: “How
often do you shoot or kill creatures in this game?”). Violent video
game use was measured using the same procedure that had been
used to measure prosocial video game use.
As has been referred to in the Introduction many games involve
both prosocial and violent behaviours. The procedure for rating
games as described above allowed a participant to simultaneously
rate a game in relation to both the degree of violent and prosocial
behaviour in the game. For example a video game such as Clash of
Clans involves prosocial behaviours such as protecting members of
one’s own clan as well as violent behaviours such as fighting enemy
clans. This issue will be addressed in more detail in the Discussion
section.
2.3.3. Weekly game play
The weekly hours spent playing each game were divided by
three to obtain a measure of average game time (Weekly game play
a ¼ 0.93) When mean scores for weekly gameplay were compared
to international studies this method of calculating weekly gameplay produced mean scores that were consistent with international
evidence (Rideout et al., 2010).
2.3.4. Empathy
Empathy was measured using the 16-item Children’s Empathic
Attitudes Questionnaire (CEAQ) (Funk, Fox, Chan, & Curtiss, 2008).
Funk et al. (2008) note that this scale is a measure of cognitive
empathy a construct which the authors conceptualise as ‘empathic
attitudes’. The CEAQ is designed to measure attitudes and likely
behaviour in children in relation to empathic responding (Example:
4
A Cronbach’s Alpha of 0.78 was obtained as a measure of internal reliability of
prosocial video game use. While this is an acceptable level of internal reliability, 141
cases were excluded from this analysis. As some participants only listed one game
in Section C, Sections D and E of some questionnaires were not completed.
Therefore due to incomplete data interpretation of the internal reliability of this
scale is problematic. A Cronbach’s Alpha of 0.75 was obtained for violent video
game use. 141 cases were excluded in this instance.
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
“I feel sorry for kids who can’t find anyone to hang out with”.)
a ¼ 0.74.
2.3.5. Helping behaviour, cooperation and sharing, affective
relationships and normative behaviour
Helping behaviour, cooperation and sharing, affective relationships and normative behaviour were assessed using the 40 item
Prosocial Orientation Questionnaire (POQ) (Cheung, Ma, & Shek,
1998). The original scale was adapted to an Irish context after
consultation with professionals working with participants from the
current sample. Helping behaviour was measured with an 11-item
subscale from the POQ (Example: “I would spend time and money
to help those in need.”) a ¼ 0.63. Co-operation and sharing were
measured using a 7-item subscale from the POQ (Example: “I feel
jealous when my friends win an award or prize.”) a ¼ 0.50. The
tendency to maintain friendly, affective and sympathetic relationships with family and peers was measured with an 11-item subscale from the POQ (Example: “I always argue with my family)
a ¼ 0.66. The tendency to comply with social norms (normative
relationships) was measured with an 11-item subscale from the
POQ (Example: “I am always on time.”) a ¼ 0.63.
2.3.6. Prosocial behaviour (teacher evaluation)
The 5-item Prosocial Behaviour Subscale of The Strengths and
Difficulties Questionnaire (Teacher version) (SDQ) (Goodman,
1997) was used for the teacher evaluation of the participants’prosocial behaviour. 43 teachers took part in this study. These
teachers were asked to evaluate the prosocial behaviour of
participating students in their classes in order to control for biases
associated with self-report. The teachers were given the following
instructions: “Please put a tick in the box which most accurately
describes your student: Not True, Somewhat True, Certainly True.
(Example: Item 3: Helpful if someone is hurt, upset or feeling ill)
a ¼ 0.89.
2.3.7. Socioeconomic status
On the information sheet/consent form that was sent to all
participants, parents/guardians were given the option of giving
information in relation to their occupational status. The occupations of parents/guardians were coded for socioeconomic status
(SES) using an Irish census based social class scale (O’Hare, Whelan,
& Commins, 1991).5
2.4. Partial missingness
In relation to missing data three separate strategies were used.
Firstly, in relation to the CEAQ and POQ missing values were left
blank. Multiple value imputation was used for these measures as
the missing values were“Missing at Random” (MAR). MAR is also
referred to as ignorable non-response. According to Tabachnick and
Fidell (2014) attention should be paid to the pattern rather than the
amount of missing data. In the present study missing values were
randomly distributed throughout the data matrix, therefore
5
This information was coded on the following ordinal scale: “Social Class 1:
Higher professional and higher managerial; proprietors and farmers owning 200 or
more acres; Social Class 2: Lower professional and lower managerial; proprietors
and farmers owning 100e199 acres; Social Class 3: Other non-manual and farmers
owning 30e49 acres; Social Class 4: Skilled manual and farmers owning 30e49
acres; Social Class 5: Semi-skilled manual and farmers owning less than 30 acres;
Social class 6: Unskilled manual” (O’Hare et al., 1991, p.142). Each Social Class was
coded with a corresponding number, eg. Social Class 1 ¼ 7, Social Class 2 ¼ 6.
Participants who were unemployed were coded as 1. Where two parents/guardians
gave their occupations, the occupation in the higher social class was used to code
SES.
653
missing data could be predicted using other variables in the data
set. The multiple value imputation function was used in SPSS 20 for
this purpose.
Secondly, a different strategy was used in relation to missing
data in the Computer/Video Game Habits Questionnaire (Prot et al.,
2014). The majority of the missing values in this measure were not
MAR. This was due to the fact that participants who did not play
games were instructed to leave the questionnaire blank. A small
number of missing values that were MAR (eg. Participant who
played video games but had omitted an item rating the game’s
content) were inputted by the researcher using knowledge of the
game’s content. ‘Prior knowledge’is a strategy used to input missing
data in situations in which the researcher has sufficient knowledge
to input missing values (Tabachnick & Fidell, 2014). Other cases in
which the same game was rated were examined by the researcher.
In addition video clips of gameplay were watched by the researcher
to obtain knowledge of the game’s content.
Thirdly, missing values for the SES variable were inputted
using mean substitution. A mean of 5.25 was inputted into the
data set. This corresponded approximately to Social Class 3,
which was coded as 5. These values were not MAR as it was
difficult to ascertain if parents/guardians did not see this item on
the information sheet or decided not to disclose this information.
A total of 133 parents/guardians out of the 538 participants did
not disclose or omitted to disclose their occupations on the information sheet, which meant that 24.7% of the SES data was
missing. While some authors caution against the use of mean
substitution when there are a large percentage of missing cases
(Tabacknick&Fidell, 2014), for the purposes of multiple linear
regression listwise deletion would have reduced the number of
cases substantially.
Finally, in relation to the SDQ, there were missing items for 17
participants. These 17 cases were excluded from data analysis. The
majority of the missing cases were due to teachers omitting to
complete the questionnaire in relation to particular students, while
a small number of cases were excluded due to the teacher submitting an incomplete questionnaire. As this variable was not being
used in the multiple linear regression it was not necessary to increase the number of cases.
2.5. Ethics
Ethical approval for this project was received on 14th November
2013 from the University College Dublin Human Research Ethics
Committee. Parents/guardians were required to give written consent before their child could participate in the study. Participants
were also asked to give their assent by signing an assent form on
the day of data collection.
3. Results
Table 1 displays mean scores, standard deviations and range of
scores for the main scales of interest.
Table 2 displays bivariate correlations between video game use
and a variety of prosocial behaviours. The negative correlations
between prosocial video game use and helping behaviour, normative behaviour, empathy and the teacher evaluation of prosocial
behaviour appear to contradict the predictions of the GLM which
predicts a positive association between prosocial video game use
and prosocial behaviour.
However as can be seen in Table 5 prosocial video game use had
a significant positive association with cooperation and sharing, the
tendency to maintain positive affective relationships and empathy
in the multiple linear regressions. It is possible that this is a suppression effect (Tzelgov & Henik, 1991).
654
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
Table 1
Mean scores, standard deviations and range of scores for main scales of interest.
Variable
N
M
SD
Range
Socio-economic status
Prosocial video game use
Violent video game use
Weekly game play
Empathy
Helping behaviour
Co-operation and sharing
Affective relationships
Normative behaviour
Prosocial behaviour (Teacher evaluation)
538
532
532
538
537
530
537
537
537
521
5.25
28.99
23.18
9.13
21.75
34.55
23.15
38.77
35.79
8.23
1.33
48.08
45.96
11.06
5.08
3.66
2.55
3.55
3.92
2.24
1e7a
0e504b
0e504c
0 he84 h
0e32 (max ¼ 32)
17e42 (max ¼ 44)
13e28 (max ¼ 28)
22e44 (max ¼ 44)
23e44 (max ¼ 44)
0e10 (max ¼ 10)
a
SES (1 ¼ unemployed; 2 ¼ social class 6; 3 ¼ social class 5; 4 ¼ social class 4; 5 ¼ social class 3; 6 ¼ social class 2; 7 ¼ social class 1). See Section 2.3 Measures for more
detailed information in relation to professions corresponding to each social class.
b
Scores are calculated by multiplying average ratings over 3 games by average time playing a game. Therefore a score of 0 approximates to either ratings of 0 for prosocial
content or no video game play. A score of 510 approximates to 84 hours weekly game play multiplied by an average prosocial rating of 6 (e.g. a rating of “Almost Always” for
each of the items measuring prosocial video game use).
c
The procedure for calculating violent video game use is identical to the procedure for calculating prosocial video game use. See footnote b above.
Table 2
Bivariate correlations between video game use and prosocial behaviours.
Variable
Helping behaviour Cooperation and sharing Affective relationships Normative behaviour Empathy Prosocial behaviour (teacher evaluation)
Prosocial video game use 0.11*
Violent video game use 0.16**
Weekly game play
0.17**
**
p < 0.001
*
0.07
0.16**
0.14**
0.05
0.16**
0.14**
3.1. Empathy and prosocial video game use
Empathy was regressed onto the variable prosocial video game
use after controlling for gender, age, school type, SES, weekly game
play and violent video game use. The value of R2 for empathy indicates that approximately 17.7% of the variance in the dependent
variable, can be attributed to the variance of the independent variables (See Table 4). Prosocial video game use was positively related
to empathy (b ¼ 0.308, p < 0.001) in a multiple linear regression
(See Table 5).
Table 3
Bivariate correlations between participants’ self-report of prosocial behaviour and
teachers’ evaluation of participants’ prosocial behaviour.
Variable (Participants’ self-report)
Prosocial behaviour (teacher evaluation)
Helping behaviour
Cooperation and sharing
Affective relationships
Normative behaviour
0.20**
0.21**
0.15**
0.25**
p < 0.001.
0.10*
0.22**
0.21**
0.11*
0.16**
0.14**
p < 0.05.
In addition the correlation between prosocial and violent video
game use was high (r ¼ 0.75, p < 0.01). Multicollinearity occurs
when different predictors in a multiple linear regression are highly
inter-related. In this case the individual predictors become redundant as all the predictors are measuring the same construct.
Multicollinearity diagnostics were carried out. Variance Inflation Factors were less than 10 in relation to each of the regression
coefficients. Therefore the assumption of multicollinearity was not
violated (Tabachnick & Fidell, 2014).
Finally, Pearson’s Product Moment Correlations revealed that
the teachers’ evaluation of students’ prosocial behaviour was
positively correlated with the students’ self-report of helping
behaviour (r ¼ 0.20, p < 0.001), cooperation and sharing (r ¼ 0.21,
p < 0.001), affective relationships (r ¼ 0.15, p < 0.001) and
normative behaviour (r ¼ 0.25, p < 0.001) (See Table 3).
**
0.12**
0.21**
0.17**
3.2. Theoretically relevant confounding variables such as
sociodemographic factors and weekly game play
Each prosocial behaviour was regressed onto the variable prosocial video game use after controlling for the following theoretically relevant confounding variables: gender, age, school type, SES,
weekly game play and violent video game use. While all of the
regression models were significant, the R2 values for cooperation
and sharing as well as affective relationships are of particular interest given the positive associations between prosocial video game
use and these variables in the regression models. The value of R2 for
cooperation and sharing indicates that 8.6% of the variance in the
dependent variable can be attributed to the variance of the independent variables. The value of R2 for affective relationships indicates that approximately 9.6% of the variance in the dependent
variable can be attributed to the variance of the independent variables (See Table 4). Prosocial video game use was positively related
to cooperation and sharing (b ¼ 0.190, p < 0.016) as well as the
tendency to maintain affective, friendly and sympathetic relationships (b ¼ 0.222, p < 0.005) in the multiple linear regressions (See
Table 5).
3.3. Violent video game use and prosocial behaviour
Violent video game use was negatively associated with the
tendency to comply with social norms (b ¼ 0.243, p < 0.003), the
tendency to maintain affective, friendly and sympathetic relationships (b ¼ 0.189, p < 0.019) as well as empathy (b ¼ 0.153,
p < 0.045) (See Table 5).
3.4. Results and the GLM
These results are consistent with some of the predictions of the
GLM which predicts that prosocial video game use will be positively
associated with prosocial behaviour and that violent video game
use will be negatively associated with prosocial behaviour. However previous research has found associations between prosocial
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
655
Table 4
Model summary for multiple linear regressions for prosocial behaviours.
Dependent variable
R
R2
Adjusted R2
Std. error of the estimate
Helping
Cooperation and sharing
Normative behaviour
Affective relationships
Empathy
0.292
0.294
0.269
0.309
0.421
0.085
0.086
0.073
0.096
0.177
0.073
0.074
0.060
0.084
0.166
3.524
2.456
3.797
3.397
4.643
Independent variables: Gender, age, school type, SES, violent video game use, prosocial video game use, weekly game play.
video game use and other prosocial behaviours such as helping
behaviour (Gentile et al., 2009; Prot et al., 2014).
4. Discussion
as cooperation and sharing as well as the tendency to maintain
positive affective relationships. The significant association between
prosocial video game use and the abovementioned prosocial behaviours while controlling for confounding variables strengthens
the evidence of a prosocial video game effect.
4.1. Discussion of findings
The main findings from this study indicate a positive and significant relationship between prosocial video game use and the
following dependent variables: cooperation and sharing, the tendency to maintain positive affective relationships as well as
empathy. Previous studies investigating the relationship between
prosocial video game use and prosocial behaviour in children have
found a positive relationship between prosocial video game use
and prosocial behaviour (Gentile et al., 2009; Prot et al., 2014).
These studies found that prosocial video game use was positively
associated with cooperation, helping behaviour and empathy in
children and adolescents (Gentile et al., 2009; Prot et al., 2014).
These studies did not measure normative behaviour or affective
relationships. In sections 4.1.1 to 4.1.3 the findings of the present
study will be discussed in relation to the three objectives outlined
in the Introduction.
4.1.1. Objective 1: To determine if prosocial video game use was
positively associated with empathy in children and adolescents
The positive association between prosocial video game use and
empathy in the multiple linear regression model is consistent with
previous research which found that the relationship between
prosocial video game use and prosocial behaviour was mediated by
empathy (Prot et al., 2014). Bartlett and Anderson (2013) propose
that “the affective processing route may be the most influential
route in predicting short-term media effects on prosocial behaviour; however more work and replication is needed to support this
claim” (Bartlett & Anderson, 2013, p.14). The findings from the
present study support Bartlett and Anderson’s (2013) proposition.
4.1.2. Objective 2: To determine if the relationship between
prosocial video game use and prosocial behaviour remained
significant after controlling for theoretically relevant variables such
as sociodemographic variables and weekly game play
Multiple linear regressions were carried out with the following
dependent variables measuring prosocial behaviours: helping
behaviour, cooperation and sharing, normative behaviour, affective
relationships. The following independent variables were controlled
for: gender, age, school type, SES, weekly game play and violent
video game use. As has been discussed in the Introduction, each
independent variable could theoretically explain part of the variance in the dependent variables measuring prosocial behaviour.
If the relationship between prosocial video game use and prosocial behaviour remains significant after controlling for theoretically relevant independent variables it could be argued that this
provides stronger evidence for a prosocial video game effect (Prot &
Anderson, 2013). In the multiple linear regressions prosocial video
game use was positively associated with prosocial behaviours such
4.1.3. Objective 3: To determine if there was a negative relationship
between violent video game use and prosocial behaviour in children
and adolescents
Finally, the negative relationship between violent video game
use and a variety of prosocial behaviours is consistent with findings
from previous research (Anderson et al., 2010; Gentile et al., 2009).
In the present study violent video game use was negatively associated with affective relationships, normative behaviour and
empathy.
4.1.4. Suppression
Another unique aspect of the present study is the issue of suppression. One of the assumptions underlying multiple linear
regression is that the independent variables are highly correlated
with the dependent variable and have low correlations among
themselves. However if an independent variable has a low correlation with the dependent variable and a high correlation with
another independent variable and then is a significant variable in
the multiple linear regression, suppression has occurred (Hinkle,
Wiersma, & Jurs, 1994).
In the present study prosocial and violent video game use were
highly correlated with each other and had a low correlation or were
not significantly associated with each of the dependent variables.
As has been discussed earlier, prosocial and violent video game use
were positively and negatively associated respectively with
empathy and a variety of prosocial behaviours in the multiple linear
regressions. Therefore it is possible that this is due to suppression.
For example, in the present study prosocial video game use was
negatively associated with empathy in a bivariate correlation.
Nevertheless, prosocial video game use was positively associated
with empathy in the multiple linear regression. This is an example
of negative suppression (Tzelgov & Henik, 1991). Violent video
game use was negatively associated with empathy both in a
bivariate correlation and the multiple linear regression. Therefore
suppression effects did not occur in relation violent video game use.
In the present study prosocial and violent video game use were
highly correlated. Participants engaged in both prosocial and violent video game use. Therefore it could be concluded that the
participants were subject to the long-term influence of both prosocial and violent video game content.
For example, a video game such as Call of Duty involves prosocial
behaviours such as cooperating with members of an army unit as
well as violent behaviours such as fighting enemy armies. In the
bivariate correlations, the negative correlation between prosocial
video game use and empathy was lower than the correlation between violent video game use and empathy. It could be argued that
prosocial video game use has a protective role against the effects of
violent video game use causing less of a decline in empathy.
656
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
Table 5
Regression coefficients for associations between prosocial video game use and prosocial behaviour.
Prosocial variable
Helping behaviour
Gendera
Age
School typeb
SESc
Weekly game play
Violent video game use
Prosocial video game use
Cooperation and sharing
Gender
b
t
p
0.210
0.051
0.051
0.043
0.174
0.068
0.137
4.766
1.181
1.054
0.926
1.705
0.843
1.736
0.001
0.238
0.292
0.355
0.089
0.400
0.083
4.348
0.001
3.456
0.001
0.407
0.684
0.549
0.583
1.424
0.155
1.942
0.053
2.416
0.016
3.388
0.001
1.384
0.167
0.397
0.692
1.690
0.092
0.029
0.977
2.995
0.003
1.138
0.256
2.911
0.004
3.923
0.001
1.454
0.147
3.073
0.002
1.451
0.147
2.363
0.019
2.845
0.005
7.599
0.001
0.153
0.878
3.051
0.002
1.060
0.290
2.098
0.036
2.005
0.045
4.127
0.001
0.190
Age
0.148
School type
0.019
SES
0.025
Weekly game play
0.145
Violent video game use
0.157
Prosocial video game use
0.190
Normative behaviour
Gender
0.150
Age
0.060
School type
0.019
SES
0.078
Weekly game play
0.003
Violent video game use
0.243
Prosocial video game use
0.090
Affective relationships
Gender
0.127
Age
0.167
School type
0.069
SES
0.140
Weekly game play
0.147
Violent video game use
0.189
Prosocial video game use
0.222
Empathy
Gender
0.316
Age
0.006
School type
0.138
SES
0.046
Weekly game play
0.202
Violent video game use
0.153
Prosocial video game use
0.308
a
b
c
Gender (1 ¼ male; 2 ¼ female).
School type (1 ¼ disadvantaged; 2 ¼ non-disadvantaged).
SES(1 ¼ unemployed; 2 ¼ social class 6; 3 ¼ social class 5; 4 ¼ social class 4; 5 ¼ social class 3; 6 ¼ social class 2; 7 ¼ social class 1).
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
However when the variance associated with violent video game
use is controlled for in the multiple linear regression the positive
contribution of prosocial video game use to the variance associated
with empathy becomes apparent.
Finally, it should be noted that the area of suppression in relation to multiple linear regression is a complex topic and the
abovementioned conclusions are possible interpretations of the
data. These findings further outline the complexity of studying
video game use in young people. Behaviour in the virtual reality of a
video game environment cannot be neatly compartmentalised into
a dichotomy of purely prosocial or violent behaviour. In the same
way that individual human behaviour can contain both prosocial
and violent components, behaviour within a video game environment can be influenced by both prosocial and violent motives.
4.2. Limitations
This study had a number of methodological weaknesses. Firstly,
the sample was a convenience sample rather than one drawn by
random sampling. Although the sample was not necessarily
representative it was diverse, drawn from ten schools representing
various socio-economic groups. Therefore it could be argued that
the sample was an accurate reflection of the diversity of 9e15 year
old young people. Secondly, the internal reliability estimates of a
number of the measures of prosocial behaviour were below 0.70.
However previous studies which have used these measures of
prosocial behaviour have found similar levels of internal reliability
(Cheung et al., 1998). Thirdly, the present study used a number of
self-report measures which carries the risk of participants having a
social desirability bias. Researchers have noted that prosocial behaviours are highly socially desirable (Eisenberg & Mussen, 1989).
Nevertheless attempts were made to control for self-report by
including a teacher evaluation of the participants’ prosocial
behaviour, which was positively correlated with the
participants’self-report of prosocial behaviour. Although video
game use was also measured by self-report, comparison of expert
ratings and participant’s ratings of video game ratings have been
highly correlated in previous research (Gentile at al, 2009).
Finally, a significant methodological weakness of this study was
its cross-sectional correlational design. Difficulties establishing
causation mean that it could be argued that children with a preexisting prosocial orientation may choose to play prosocial video
games. However it could also be argued that children tend to
choose to play games due to their popularity and quality of game
play rather than explicitly choosing a game based on its content.
Therefore a child with high levels of prosocial behaviour could
choose to play a violent video game due to the power of market
forces such as advertising (Calvert, 2008).
4.3. Theoretical issues
In the following section a number of theoretical issues in relation to prosocial video game effects will be discussed. The present
study has been guided by the predictions of the General Learning
Model (GLM) (Gentile et al., 2009). The GLM is an extension of the
General Aggression Model (GAM) (Anderson & Bushman, 2001).
However while the GAM can explain violent video game effects, the
GLM has a broader scope that can be used to explain other issues
such as prosocial video game effects and gender stereotypes in
games. Both the GAM and the GLM are integrative theories. Each
model integrates elements of five different socio-cognitive theories
of personality theories in an attempt to explain video game effects.
These five theories are Cognitive Neo-Associative Theory
(Berkowitz, 1984), Excitation Transfer Theory (Zillmann, 1971),
Social Learning Theory (Bandura, 1977), Script Theory (Huesmann,
657
1986) and Social Information Processing (Crick & Dodge, 1994) (as
cited in Bartlett & Anderson, 2013).
Gentile et al. (2009) note that prosocial and antisocial behaviour
are not binary constructs. It is possible to be hostile towards enemies while behaving prosocially towards friends. As has been
noted in the Introduction, the GLM proposes that two short-term
processes explain prosocial video game effects. Firstly, the cognitive effect of priming scripts predicts that games with prosocial
content will result in prosocial behavioural scripts being primed
and rehearsed. Secondly, changes in cognitions, feelings and levels
of physiological arousal while playing a prosocial video game are
reciprocally reinforced through both classical and operant
conditioning.
However there is a possible theoretical weakness in relation to
the GLM and prosocial video game effects. In the present study
prosocial video game effects were strongly associated with affective
processing. The GLM is a theory which while containing emotional
constructs has a strong focus on cognitive constructs consistent
with the social cognitive theories of personality, which are integrated in this model (Prot et al., 2014). It could be argued that
theories focussed on an individual’s emotional response to a
stimulus might explain prosocial video game effects more clearly.
Two theories from positive psychology could be advanced to
explain prosocial video game effects.
Elevation has been defined as “an emotion triggered by people
behaving in a virtuous, pure, or superhuman way” (Haidt, 2003,
p.281). Observing video game clips of individuals such as Mother
Teresa behaving prosocially has induced elevation in participants in
experimental studies (Haidt, 2003). It is theoretically plausible that
a prosocial video game such as Peacemaker (http://www.
peacemakergame.com) might induce elevation in players which
could mediate prosocial video game effects.
The Broaden and Build Theory of Positive Emotion (Friedrickson,
2001) hypothesises that positive emotion broadens thought-action
repertoires in the actual moment. Therefore experiencing positive
emotion through participating in prosocial video game play could
create an upward spiral of positive emotion. This broadening of
thought-action repertoires could potentially mediate the performance of prosocial acts.
4.4. Practical implications
The practical implications of prosocial video games are
numerous. Video games have been used to train visual skills in
adults (Achtman, Green, & Bavelier, 2008), to teach civics to middle
school students (www.icivics.org) and to teach geometry and social
studies to elementary school students (https://minecraftedu.com).
The video game Secret Agent Society is used for social skills instruction for children with Asperger’s Syndrome (http://www.sstinstitute.net). Based on the evidence from the present study
video games with prosocial content could be used by educators to
develop empathic concern and improve affective relationships in a
diverse population of youth.
4.5. Future research
Disadvantaged populations could particularly benefit from the
use of prosocial video games in educational and clinical settings.
Socioeconomic disadvantage is associated with lower levels of academic achievement (Mc Loyd, 1998). Furthermore, longitudinal
research has found that prosocial behaviour in childhood predicted
academic achievement in adolescence (Caprara, Barbaranelli,
Pastorelli, Bandura, & Zimbardo, 2000). Video games do not
depend exclusively on formal literacy and numeracy to teach skills
and convey social messages. Therefore video games with prosocial
658
B. Harrington, M. O’Connell / Computers in Human Behavior 63 (2016) 650e658
content could become a vital pedagogical tool in the educational
provision for youth from disadvantaged communities. Future
research in the area of prosocial video game effects could address
this gap in the research by using experimental and longitudinal
designs in order to establish causal relationships.
5. Conclusions
These findings are consistent with previous research which
found that prosocial video game use was positively associated with
prosocial behaviour and empathy in children and adolescents
(Gentile et al., 2009; Prot et al., 2014). However, the findings from
the present study suggest that emotions rather than cognitions
could explain prosocial video game effects. Therefore future studies
using models from positive psychology such as Haidt’s (2003)
construct of elevation or the Broaden and Build Theory of Positive
Emotion (Friedrickson, 2001) could explain prosocial video game
effects in children and adolescents more clearly.
Video games can be conceived as ‘virtual teachers’ that can
teach both prosocial and anti-social behaviours. The findings from
the present study contribute to an increasing body of evidence that
is succinctly summarised in the maxim: “Video games are exemplary teachers” (Gentile & Gentile, 2008). Parents and educators
should bear this maxim in mind when weighing up the risks and
benefits of these virtual teachers in relation to the healthy development of the young people in their care.
References
Achtman, R. L., Green, C. S., & Bavelier, D. (2008). Video games as a tool to train
visual skills. Restorative Neurology and Neuroscience, 26(4e5), 435e446.
Anderson, C. A., & Bushman, B. J. (2001). Effects of violent video games on
aggressive behavior, aggressive cognition, aggressive affect, physiological
arousal, and prosocial behavior: a meta-analytic review of the scientific literature. Psychological Science, 12, 353e359.
Anderson, C. A., Gentile, D. A., & Buckley, K. E. (2007). Violent video game effects on
children and adolescents: Theory, research, and public policy. New York: Oxford
University Press.
Anderson, C. A., Shibuya, A., Ihori, N., Swing, E. L., Bushman, B. J., Sakamoto, A., et al.
(2010). Violent video game effects on aggression, empathy, and prosocial
behavior in Eastern and Western countries: a meta-analytic review. Psychological Bulletin, 136(2), 151e173.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Bartlett, C. P., & Anderson, C. A. (2013). Examining media effects: the general
aggression and general learning models. In E. Scharrer (Ed.), Media effects/Media
psychology: Vol. 5. The international encyclopedia of media studies (pp. 1e20).
London: Blackwell.
Berkowitz, L. (1984). .Some effects of thoughts on anti- and prosocial influences of
media events: a cognitive neoassociation analysis. Psychological Bulletin, 95(3),
410e427.
Biernacki, P., & Waldorf, D. (1981). Snowball sampling: problems and techniques of
chain referral sampling. Sociological Methods and Research, 10(2), 141e163.
Bijvank, M. N., Konijn, E. A., & Bushman, B. J. (2012). “We don’t need no education”:
video game preferences, video game motivations, and aggressiveness among
adolescent boys of different educational ability levels. Journal of Adolescence,
35(1), 153e162.
Boxer, P., Huesmann, R. L., Bushman, B. J., O’Brien, M., & Moceri, D. (2008). The role
of violent media preference and cumulative developmental risk for violence
and general aggression. Journal of Youth and Adolescence, 38, 417e428.
Calvert, S. L. (2008). Children as consumers: advertising and marketing. The Future
of Children, 18(1), 205e234.
Calvo, A. J., Gonzalez, R., & Martorell, M. C. (2014). Variables relacionadas con la
conducta prosocial en la infancia y adolescencia: personalidad, autoconcepto y
nero/Variables related to prosocial behaviour in childhood and adolescence:
ge
personality, self-concept and gender. Infancia y Aprendizaje: Journal for the Study
of Education and Development, 24(1), 95e111.
Caprara, G. V., Barbaranelli, C., Pastorelli, C., Bandura, A., & Zimbardo, P. (2000).
Prosocial foundations of children’s academic achievement. Psychological Science, 11(4), 302e306.
Caprara, G. V., & Pastorelli, C. (1993). Early emotional instability, prosocial behaviour, and aggression: some methodological aspects. European Journal of Personality, 7, 19e36.
Cheung, P. C., Ma, H. K., & Shek, T. L. D. (1998). Conceptions of success: their
correlates with prosocial orientation and behaviour in Chinese adolescents.
Journal of Adolescence, 21, 31e42.
Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social informationprocessing mechanisms in children's social adjustment. Psychological Bulletin,
115(1), 74e101.
Dietz, T. L. (1998). An examination of violence and gender role portrayals in video
games: implications for gender socialization and aggressive behavior. Sex Roles.,
38(5e6), 425e442.
Eagly, A. H., & Crowley, M. (1986). Gender and helping behavior: a meta-analytic
review of the social psychological literature. Psychological Bulletin, 100(3),
283e308.
Eisenberg, N., & Mussen, P. H. (1989). The roots of prosocial behaviour in children.
Cambridge: Cambridge University Press.
Entertainment Software Association. (2015). Sales, demographic and usage data:
Essential facts about the video game industry.
Ferguson, C. J., San Miguel, C., Garza, A., & Jerabeck, J. M. (2012). A longitudinal test
of video game violence influences on dating and aggression: a 3-year longitudinal study of adolescents. Journal of Psychiatric Research, 46(2), 141e146.
Friedrickson, B. L. (2001). The role of positive emotions in positive psychology: the
broaden-and-build theory of positive emotions. American Psychologist, 56(3),
218e226.
Funk, J. B., Fox, C. M., Chan, M., & Curtiss, K. (2008). The development of the Children’s Empathic Attitudes Scale using classical and Rasch analyses. Journal of
Applied Developmental Psychology, 29, 187e198.
Gee, J. P. (2003). What video games have to teach us about learning and literacy.
Computers in Entertainment, 1, 1e4.
Gentile, D. A., Anderson, C. A., Yukawa, S., Ihoro, N., Kam Ming, L., Shibuya, A., et al.
(2009). The effects of prosocial video games on prosocial behaviors: international evidence from correlational, longitudinal and experimental studies.
Personality and Social Psychology Bulletin, 35(6), 752e763.
Gentile, D. A., & Gentile, J. R. (2008). Violent video games as exemplary teachers: a
conceptual analysis. Journal of Youth and Adolescence, 37(2), 127e141.
Gentile, D. A., Li, D., Khoo, A., Prot, S., & Anderson, C. A. (2014). Mediators and
moderators of long-term effects of violent video games on aggressive behavior.
JAMA Pediatrics, 168(5), 450e457.
Goodman, R. (1997). The strengths and difficulties questionnaire: a research note.
Journal of Child Psychology and Psychiatry, 38(5), 581e586.
Granic, I., Lobel, A., & Engels, R. C. M. E. (2014). The benefits of playing video games.
American Psychologist, 69(1), 66e78.
Greitemeyer, T., & Mugge, D. (2014). Video games do affect social outcomes:A metaanalytic review of the effects of violent and prosocial video game play. Personality and Social Psychology Bulletin, 40(5), 578e589.
Greitemeyer, T., & Osswald, S. (2010). Effects of prosocial video games on prosocial
behaviour. Journal of Personality and Social Psychology, 98(2), 211e221.
Guinote, A., Cotzia, I., Sandhu, S., & Siwa, P. (2015). Social status modulates prosocial
behavior and egalitarianism in preschool children and adults. PNAS, 122(3),
731e736.
Haidt, J. (2003). Elevation and the positive psychology of morality. In C. L. M. Keyes,
& J. Haidt (Eds.), Flourishing: Positive psychology and the life well-lived (pp.
275e289). Washington DC: American Psychological Association.
Hinkle, D. E., Wiersma, W., & Jurs, S. G. (1994). Applied statistics for the behavioral
sciences. Boston: Houghton Mifflin.
Hoglund, W. L., & Leadbetter, B. J. (2004). The effects of family, school, and classroom ecologies on changes in children’s social competence and emotional and
behavioral problems in first grade. Developmental Psychology, 40(4), 533e544.
Huesmann, L. R. (1986). Psychological processes promoting the relation between
exposure to media violence and aggressive behavior by the viewer. Journal of
Social Issues, 42(3), 125e139.
Kazdin, A. E. (1987). Treatment of antisocial behavior in children: current status and
future directions. Psychological Bulletin, 102(2), 187e203.
Mazurek, M. O., & Engelhardt, C. R. (2013). Video game use in boys with autism
spectrum disorder, ADHD, or typical development. Pediatrics, 132(2), 260e266.
Mc Loyd, V. C. (1998). Socioeconomic disadvantage and child development. American Psychologist, 53(2), 185e204.
O’Hare, A., Whelan, C. T., & Commins, P. (1991). The development of an Irish censusbased social class scale. Economic & Social Review, 22(2), 135e156.
Prot, S., & Anderson, C. A. (2013). Research methods, design and statistics in media
psychology. In K. E. Dill (Ed.), The Oxford handbook of media psychology (pp.
109e136). New York: Oxford University.
Prot, S., Gentile, D. A., Anderson, C. A., Suzuki, K., Swing, E., Lim, K. M., … Lam, B. C. P.
(2014). Long-term relations among prosocial-media use, empathy and prosocial
behavior. Psychological Science, 25(2), 358e368.
Rideout, V. J., Foehr, U. L., & Roberts, D. F. (2010). Generation M: Media in the lives of
8-18 year-olds. Menlo Park, CA: Henry J. Kaiser Family Foundation.
Tabachnick, B. G., & Fidell, L. S. (2014). Using multivariate statistics. Edinburgh:
Pearson Education Limited.
Tzelgov, J., & Henik, A. (1991). Suppression situations in psychological research:
definitions, implications, and applications. Psychological Bulletin, 109(3),
524e536.
Zillmann, D. (1971). Excitation transfer in communication-mediated aggressive
behavior. Journal of Experimental Social Psychology, 7(4), 419e434.
Psychology of Popular Media Culture
2019, Vol. 8, No. 1, 76 – 87
© 2017 American Psychological Association
2160-4134/19/$12.00 http://dx.doi.org/10.1037/ppm0000159
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Gaining a Competitive Edge: Longitudinal Associations Between
Children’s Competitive Video Game Playing, Conduct Problems, Peer
Relations, and Prosocial Behavior
Adam Lobel
Rutger C. M. E. Engels
University of Geneva
Trimbos Instituut, Utrecht, the Netherlands
Lisanne L. Stone
Isabela Granic
Pro Persona, Nijmegen, the Netherlands
Radboud University
Playful competition is an important hallmark of healthy child development. Playful competition facilitates moral learning, rewards perspective-taking skills, and challenges children to healthily regulate
unpleasant emotions such as frustration, anger, and jealousy. Despite this, research on the effects of
competitive video gaming has focused on antisocial outcomes, such as declines in prosocial behavior.
Moreover, methodological shortcomings such as experimental studies using designs with poor generalizability, and a lack of longitudinal studies, leave open the influence of competitive gaming on social
development among preadolescent children. This longitudinal study therefore investigated the relation
between competitive gaming and changes in children’s social development across 3 measures: conduct
problems, peer relations, and prosocial behavior. At 2 timepoints, 1 year apart, 184 Dutch children
(8.31–12.68 years old) reported their gaming frequency and listed their favorite games to play, and their
parents reported their children’s psychosocial health. Children’s nominations were coded as including or
not including a competitive video game. Children who nominated a competitive game at the first time
point were more likely to show a decrease in conduct problems and an improvement in peer relations.
No interactions were observed between competitive gaming and gaming frequency. These results
encourage future research to investigate the social benefits of playful competitive gaming among peers,
and for future studies to take other variables such as violent content, cooperative play, and real world
competitive play into account.
Public Policy Relevance Statement
Video games have become a cultural fixture and a staple of child development. This research
describes the potential benefits which may come with children playing competitive video games,
particularly for boys.
Keywords: gaming, competitive play, child development, prosocial behavior, peer relations
capacities of their own bodies and objects in their immediate
environment (Piaget, 1962). Sensorimotor play remains popular
throughout childhood; it is seen in hand– eye coordination games,
such as catch, and in forms of pretend play, such as when children
play by building. In pretend play, children’s imaginations allow
them to fantasize about nonexistent entities, construct narratives
for inanimate objects, and assume the roles of adults and
professionals. Because these behaviors help children develop
perspective-taking skills, and learn how to cooperate with others
(Fein, 1981; Lillard et al., 2013), pretend play is important for
children’s socialization (Denzin, 1975). At the same time that
children enact pretend play with others, they also begin to engage
in games with rules. Due to their fixed structure, these rule-based
games align with children’s interest in better understanding the
world (Whitebread, Basilio, Kuvalja, & Verma, 2012) while also
enabling children to playfully compete with peers.
Recognized as a child’s right by the United Nations General
Assembly (UNGA), play is essential for social development
(Frost, Wortham, & Reifel, 2008; UNGA, 1959). Play comes in
many forms, each with developmental benefits. As infants, children engage in sensorimotor play, toying with and discovering the
This article was published Online First August 24, 2017.
Adam Lobel, Swiss Center for Affective Sciences, University of Geneva; Rutger C. M. E. Engels, Trimbos Instituut, Netherlands institute of
Mental Health and Addiction, Utrecht, the Netherlands; Lisanne L. Stone,
Overwaal, Centre for Anxiety Disorders, Pro Persona, Nijmegen, the
Netherlands; Isabela Granic, Behavioural Science Institute, Radboud University.
Correspondence concerning this article should be addressed to Adam
Lobel, Swiss Center for Affective Sciences, University of Geneva, Chemin
des Mines 9, 1202 Geneva, Switzerland. E-mail: adam@adamlobel.com
76
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
GAINING A COMPETITIVE EDGE
Competitive play is crucial for social development. Playful
competition is a hallmark of play in primates and mammals, whose
young almost invariably engage in rough-and-tumble play (Power,
2000). In humans, rough-and-tumble play emerges early in child
development, and often persists through adolescence and into early
adulthood. Because of its intimate nature, and the necessity for
mutual trust between players, this form of play facilitates emotion
recognition and cultivates bonding between children and their
peers and family members (Jarvis, 2010). But, as children’s working memory and executive function skills improve, competitive
play in games with rules also emerges. These are games generally
predicated on clear win- and loss-states, and they prescribe specific
actions that may or may not be performed during play. Common
examples are board games such as Checkers, Chess, and Monopoly, and physical games such as Hide and Seek and sports. Much
like rough-and-tumble play, playing competitive games with rules
also provides a valuable context for social development. For example, because it encourages players to predict their opponent’s
strategies, competitive play may promote perspective-taking and
the development of children’s theory of mind (Goodie, Doshi, &
Young, 2012). Moreover, competitive play is highly relevant for
both children’s moral development and peer relationships. Competitive play forces children to cooperate and take turns, abiding by
the game’s rules and finding a common ethical ground. The
pressure of competition may also elicit unpleasant emotional experiences, such as frustration, disappointment, and embarrassment.
Sharing and working through these experiences with peers may
promote bonding and prepare children to regulate these emotions
with more facility outside of play contexts (Erikson, 1993; Russ,
2003; Wagner et al., 2014).
Modern Video Gaming and Competitive Play
Here, we apply this developmental lens to one of the most
common “playgrounds” where children today are commonly
found, video games. Indeed, video games have become a virtually
universal aspect of child development, with over 90% of children
and adolescents dedicating at least an hour per week to gaming
(Lenhart et al., 2008). As modern video games have become
increasingly social in nature (Olson, 2010). The impact of competitive video game play on social development seems particularly
relevant. Today’s video games can be played alone, in person with
small groups, or online with up to hundreds of people simultaneously. Video games therefore seem to represent a modern playground, inviting children to play in a myriad of ways. Reflecting
the increased prevalence of gaming as a form of social play, nine
of the 10 best-selling video games in the United States in 2016
extensively featured multiplayer functionality (NPD Group, as
cited by Tassi in Forbes, 2017), with competitive game modes
being central to the game’s design in eight of these releases (e.g.,
Battlefield 1, Overwatch, and Fifa 17).
Competitive Gaming and Social Competencies
Despite a body of literature supporting the socioemotional benefits of competitive play, research into the effects of competitive
video game play has predominantly focused on the potentially
deleterious effects of competition. Under a dichotomy of
cooperative-versus-competitive gaming (Ewoldsen et al., 2012;
77
Greitemeyer, Traut-Mattausch, & Osswald, 2012; Schmierbach,
2010; Velez, Mahood, Ewoldsen, & Moyer-Gusé, 2014), competitive gaming has been widely studied as an antecedent to increased
aggression and decreases in prosocial behavior. These hypotheses
have their theoretical underpinnings in the General Learning
Model (Buckley & Anderson, 2006; originally formulated as the
General Aggression Model in Bushman & Anderson, 2002), a
model developed within the more widely researched violent gaming field (see: Ferguson & Konijn, 2015). Under the General
Learning Model, gaming fosters scripts about how to manage real
world interactions. Under this model, violent video game playing
biases players into more likely perceiving aggressive intentions in
others, and to also perceive aggressive behaviors as a more viable
solution to conflict. Importantly, however, not all competitive
video games are violent in nature; for example, although many
first-person shooter titles feature competitive game modes, racing
and sports games center around competition without being violent
(Adachi & Willoughby, 2016). Thus, when applied to competitive
gaming research, the General Learning Model predicts that competitive video game playing biases players into perceiving more
social situations as adversarial and as calling for aggressive behavior.
However, research on the deleterious effects of competitive
gaming has almost exclusively been conducted among adolescents
or adults, and in short-term, lab-based settings (for a longitudinal
study among adolescents and adults, see Adachi & Willoughby,
2016). This raises a several issues. First, findings regarding adolescents and adults may not generalize to children. This is because
children are very much in the process of developing the cognitive
and socioemotional skills needed to create and maintain relationships (Bigelow, 1977; Newcomb & Bagwell, 1995). Second, the
observed effects of competitive gaming in lab-based studies may
only operate in the short-term. Longitudinal designs are needed to
demonstrate the potential lasting influences of competitive gaming. Third, the assignment procedures in lab-based studies precluded participants from playing competitively against their
known peers. Competitive play against strangers—and against
individuals that one may never meet in person—may have different consequences than competitive play against friends. For example, competition among friends seems more prone to instilling
a playful spirit, and may also be based on feelings of mutual trust
and respect. Competitive play among friends may also lack the
sense of finality that play against random strangers might; when
playing with friends, losses and victories can be contextualized
within a history of competitive play where each player’s skills and
tactics develop.
There are therefore several gaps in the literature. Given the lack
of research conducted on competitive gaming among children,
longitudinally, and in naturalized environments, the developmental
impact of competitive gaming in children remains largely unclear.
Similarly, the relative focus on competitive gaming as an antisocial
activity leaves open whether competitive gaming among peers
could foster healthy relationships. This study therefore employed a
longitudinal design to investigate the potential influence of competitive gaming on children’s conduct problems, peer relationships, and prosocial behavior.
As an example of traditional competitive games with rules,
sports provide an ideal point of comparison for illustrating these
potential benefits. Sports are widely considered a valuable domain
LOBEL, ENGELS, STONE, AND GRANIC
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
78
for moral development (Bailey, 2006; Fraser-Thomas, Côte, &
Deakin, 2005; Kleiber & Roberts, 1981), and numerous sportbased interventions have been implemented in order to manifest
psychosocial improvements (Gould & Carson, 2008; Hellison,
1998; Romance, Weiss, & Bockoven, 1986). A recent review
indicated that among children and adolescents, playing sports was
associated with psychosocial benefits, and that such associations
were more pronounced in team-based sports (Eime, Young, Harvey, Charity, & Payne, 2013). This is due to several features. First,
sports and other games with rules emphasize notions of fair play.
Competitive play is known to heighten affective arousal, riling up
children to become potentially more (physically) aggressive (Ensor, Hart, Jacobs, & Hughes, 2011). Second, children’s emotion
regulation skills are challenged when experiencing the heightened
emotional arousal that accompanies winning or losing a game. In
victory, children must learn how to experience pride without
overly bragging, and in defeat, they must learn how to overcome
disappointment without unfairly disparaging their play partner.
Sport and competitive play therefore help instantiate the importance of moral behavior, and encourage children to exercise their
perspective-taking skills and to be gracious to their peers even in
emotionally charged instances (Denzin, 1975).
It is important to determine whether competitive gaming can
afford similar benefits. Competitive gaming seems a particularly
important domain to investigate in this regard for several reasons.
First, the last years have seen children migrate from outdoor
activities and sports to video game play (Hofferth, 2010). Second,
perhaps more so than sports and other traditional games with rules,
competitive gaming is a potential hotbed for aggressive interactions. Derogatory banter and “rage quitting” (when a player angrily
and abruptly quits a game; Linderoth, Björk, & Olsson, 2012)
commonly occur in competitive games. These behaviors are socially alienating; indeed online games often give rise to selfpolicing communities which reject players who are notorious for
being a poor sport (Williams, Caplan, & Xiong, 2007). Because
preadolescent children typically play (competitive) video games
with their friends, children may be learning how to work through
these aggressive urges, maintain composure, and respond in a
normative, playful manner. In sum, competitive gaming may foster
an environment that is ideal for teaching children to suppress
antisocial urges and to resolve social conflicts.
The Present Study
This study was designed to address the gaps of past research by
(a) using a longitudinal design, to (b) investigate competitive
gaming and the development of social competencies among (c)
preadolescent children. Children between the ages of 8 and 11
were interviewed twice, one year apart. Children described their
gaming behavior and nominated their favorite video games, and
parents were asked to report on children’s social competencies. We
hypothesized that children who played competitive video games
would show improvements in conduct problems, peer relationships, and prosocial behavior. For exploratory analyses, we investigated a potential dosage effect whereby those children who
played competitive video games with greater frequency may show
additional benefits.
Method
Participants
Data were collected during home visits one year apart (days
between visits: range: 194 – 462, M ⫽ 347.65). For recruitment, we
invited participants in Stone and colleagues (2013) to participate in
a three-wave longitudinal study; the present study used data collected in its second and third waves. We left out the first wave of
data from our analyses because there was an insufficient degree of
variation in the sample with regards to children playing competitive video games; whereas over 50 children reported preference for
a competitive video gaming in the study’s second and third waves,
only 26 children did so at Year 1 of the longitudinal study. This
may largely be attributed to the age of study’s sample, a quarter of
which was younger than eight years old, and none of which was 12
years or older at the study’s first wave.1 For hypothesis testing, we
segmented out nongamer children because our hypotheses specifically concerned differences in gaming behavior. In these tests,
therefore, competitive gamers were compared against other gamers
from our sample (see Planned Analyses, below).
The study’s procedures were approved by the Behavioral Science Institute’s Ethical Review Board under the Radboud University, and informed-consent forms were collected at all timepoints.
Descriptive statistics for the sample at Year 2 and Year 3 (Y2 and
Y3, respectively) are reported in Table 1. Ten participants from Y2
(n ⫽ 184; male ⫽ 48.9%) declined to participate at Y3 (n ⫽ 174;
male ⫽ 47.78%). Data from 10 parent reports were missing at Y2
because their data were not properly saved by the recording
software, and three parents failed to complete their online questionnaires at Y3.
Procedure
Children provided self-reports during a private, face-to-face
interview with an experimenter. Data were collected by the first
author and a team of senior Bachelor’s students enrolled in the
Radboud University’s Pedagogical Sciences program. Mandatory
for their studies, these students were formally trained for conducting interviews with children; under supervision of the first author
and prior to data collection, students were further trained in the
study’s protocols and interview procedure. During each interview,
the experimenter hand-recorded the participant’s responses. To
ensure that these data were properly transferred to a digital dataset,
hand-written data were twice annotated to a computer. Parents
provided their survey responses via an online questionnaire. Families were rewarded a 30 and 50 Euro voucher check (per child) for
their participation at Y2 and Y3, respectively.
Measures
Social competency measures. Three social competencies
were measured by parent’s reports on subscales of the Dutch
version of the Strengths and Difficulties Questionnaire (SDQ
(Goodman, 1997); Dutch version (van Widenfelt, Goedhart, Treffers, & Goodman, 2003)). The SDQ uses a 3-point Likert scale
1
The characteristics of the sample at Year 1 are described in Lobel and
colleagues (2014).
GAINING A COMPETITIVE EDGE
79
Table 1
Child and Parent Demographics at Y2 and Y3
Children
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Sex
Y2
Y3
Age
Y2
Y3
Parents
n
Male
Female
n
Male
Female
184
174
90 (48.9%)
83 (47.7%)
94 (51.1%)
91 (52.3%)
174
174
24 (13.8%)
19 (11%)
150 (86.2%)
153 (89%)
Range
M
SD
Range
M
SD
8.31–12.68
9.30–13.53
10.23
11.16
1.14
1.14
30.68–52.42
31.7–53.58
42.83
43.72
3.76
3.68
(0 –2 Not true to Very true). Of the SDQ’s five subscales, the three
social competency subscales used were: (a) conduct problems, (b)
peer problems, and (c) prosocial behavior. Consistent with Stone
and colleagues (2013) reliability was calculated using ; this
reliability index has repeatedly been shown to yield more accurate
estimates than ␣, particularly so when data are skewed, as is the
case with the SDQ (Stone et al., 2015; Zinbarg, Revelle, Yovel, &
Li, 2005). All subscales showed acceptable to good reliability at
Y2 and Y3: (a) conduct problems (sample: Often fights with other
children or bullies them; Y2: M ⫽ 0.84, SD ⫽ 1.46, Y2 ⫽ .89; Y3:
M ⫽ 0.78, SD ⫽ 1.22, Y3 ⫽ .81); (b) peer problems (sample:
Rather solitary, tends to play alone; Y2: M ⫽ 0.97; SD ⫽ 1.27,
Y2 ⫽ .68; Y3: M ⫽ 0.92; SD ⫽ 1.33, Y3 ⫽ .78); and (c) prosocial
behavior (sample: Shares readily with other children; Y2: M ⫽ 6.9;
SD ⫽ 1.31, Y2 ⫽ .78; Y3: M ⫽ 6.82; SD ⫽ 1.49, Y3 ⫽ .86).
Gaming frequency. Gaming frequency was measured with
child reports for the number of hours they had played video games
during the past week. Given the potential difficulty of children
recalling their gaming hours across an entire week, this measure
was scaffolded by an additional measure of gaming frequency: In
interviews, children looked over a calendar with the experimenter
and indicated for each day over the past full week whether or not
they had played a video game in the morning, afternoon, and
evening. Parents separately reported via an online questionnaire
regarding the number of hours their child played on average per
week. Moderate correlations were observed across the three frequency measures at each time point (Y2: r ⱖ .44, p ⬍ .001; Y3: r ⱖ
.56, p ⬍ .001). “Video games” were explicitly described to parents
and children as any game that can be played on an electronic
device, and several example games were listed.
We specifically used child reports of gaming frequency. This
was done so that our analyses would rely on different reporters for
the predictor and predicted variables; again, the social competency
outcome variables were reported by parents. This cross-reporter
analysis avoids the potential single source bias that is introduced
by relying on a single reporter (Burk & Laursen, 2010; Lobel,
Granic, Stone, & Engels, 2014). Both parent and children’s reported hours of gaming were Windorized with a cut-off at 3 SD
above the mean; at Y2, two outliers were present based on both
child and reports, and at Y3 four outliers were present in child
reports and two in the parent report (Y2: M ⫽ 5.92, SD ⫽ 5.9; Y3:
M ⫽ 5.59, SD ⫽ 5.46).
Competitive gaming. Similar to previous studies (Anderson
& Dill, 2000; Prot et al., 2014), children were asked to report their
favorite video game(s) from the past several weeks. Competitive
gaming was therefore computed as a dichotomous variable; children who listed a competitive video game among their favorite
games were assigned a 1, and those who did not were assigned a
0. Children listed over 140 games as favorites2, with Minecraft
being the sample’s most popular game (n ⫽ 46) and Fifa the most
popular competitive game (n ⫽ 21). As in Adachi and Willoughby
(2015), games were deemed competitive if their design was predominantly built around competition. Games were assigned jointly
in consultation between the study’s first and final author (see
Appendix for full listing and coding). Competitive games came
from a diverse array of genres, such as puzzle games (e.g., Ruzzle,
n ⫽ 1), strategy games (e.g., Clash of Clans, n ⫽ 12), sports games
(e.g., games from the Fifa series, n ⫽ 21), racing games (e.g.,
games from the Mario Kart series, n ⫽ 16), and (violent) firstperson shooter games (e.g., games from the Call of Duty series,
n ⫽ 8). In all such games, the primary form of interaction involves
players trying to perform better than their opponents, whereas
popular noncompetitive games were Minecraft (n ⫽ 46)—a game
where players (cooperatively) build structures—Flappy Bird (n ⫽
18) and Subway Surfer (n ⫽ 15)—reaction time (RT) games
playable by only one player at a time—and games from the Mario
Brothers series (n ⫽ 17)— bright, fantastical games about collecting items while avoiding physical contact with enemies and environmental hazards. Sixty-one children identified a competitive
game among their favorite games at Y2 (31.4%). To check the
validity of this coding scheme, children were also asked to report
how often their gaming sessions involves them “playing against
others; that the game is competitive” (5-point Likert scale, Never
to Every time or almost every time). Children who identified a
competitive game among their favorite games reported playing
competitively more often (competitive M ⫽ 3.07, SD ⫽ 1.28
noncompetitive M ⫽ 2.47, SD ⫽ 1.18; t(177) ⫽ 2.83, p ⫽ .002).
Planned Analyses
All analyses were performed in Statistical Package for the
Social Sciences (SPSS; Version 23). As stated above, we segmented out nongamer children we intended to determine whether
gaming competitively could be beneficial or detrimental compared
2
This figure collapses games within the same series as referring to one
game. For example, Call of Duty Black Ops 2 and Call of Duty: Ghost are
counted as the same title.
LOBEL, ENGELS, STONE, AND GRANIC
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
80
( ⫽ ⫺.20, t ⫽ ⫺2.55, p ⫽ .012). Contrary to expectations,
competitive gaming was not associated with changes in prosocial
behavior ( ⫽ ⫺.09, t ⫽ ⫺1.27, p ⫽ .205). Outside the purview
of our hypotheses, we also observed a positive association between
gaming frequency and prosocial behavior ( ⫽ .13, t ⫽ 1.99, p ⫽
.049). Violent gaming was not associated with changes in any of
the social competencies ( range: ⫺.07 to .06; t range: ⫺.833 to
.82; p range: 0.406 to .645). Finally, no interaction effects were
observed (conduct:  ⫽ ⫺.01, t ⫽ ⫺0.15, p ⫽ .881; peer:  ⫽
.066, t ⫽ 0.77, p ⫽ .442; prosocial:  ⫽ .059, t ⫽ 0.12, p ⫽ .902),
suggesting no dosage effect of competitive gaming.5
with gaming in noncompetitive ways. Children who regularly
played video games were defined as children who played for more
than one hour per week (98.27% of children at Y2, n ⫽ 171).3 For
preliminary analyses, independent t tests and a chi-square test
were used to determine whether there were gender differences on
all variables at both timepoints.
Multiple linear regression analysis was used to investigate our
hypotheses. Competitive gaming among children who regularly
played video games was tested as a main predictor of changes in
three separate social competencies, conduct problems, peer problems, and prosocial behavior. As a follow-up, we next explored
whether the amount of competitive gaming may have a dose effect.
These subsequent models therefore added an interaction term to
each of the previous models. This interaction term was derived
from centering children’s gaming frequency and multiplying this
value by the competitive gaming variable. Age, gaming frequency,
and violent gaming4 were included as direct predictors as control
variables. As a backup check, all models were also run using
parental reports of children’s gaming frequency.
When running these models on the boys in the sample, competitive gaming showed the same relationship to changes in social
competencies; among girls, however, no associations were observed (see Table 4). Violent gaming did not predict changes in
social competencies for boys nor girls. Also, among boys, gaming
at Y2 was not associated with an increase in prosocial behavior.
Results
Discussion
Preliminary Analyses
No age differences were observed between competitive gamers
and the remainder of the sample (Y2: t(166) ⫽ ⫺.32, p ⫽ .750; Y3:
t(155) ⫽ ⫺.48, p ⫽ .632). Likewise, no differences were observed
in the time between visits, t(158) ⫽ ⫺1.19, p ⫽ .236. Gender
differences were observed at both time points for social competencies, gaming frequency, and for competitive gaming (see Table
2). Regarding social competencies, boys showed less prosocial
behavior than girls at both timepoints, t(160) ⫽ ⫺2.33, p ⫽ .02;
Y3: t(151.12 ⫽ ⫺2.41, p ⫽ .017), and at Y3, boys showed more
peer problems, t(157) ⫽ 1.98, p ⫽ .050. (Despite the emergence of
a gender difference in peer problems at Y3, paired-samples t tests
indicated that peer problems remained constant among boys and
girls from Y2 to Y3; boys t(76) ⫽ ⫺.655, p ⫽ .514, girls t(75) ⫽
1.35, p ⫽ .181.) Boys also reported gaming more hours per week
at both Y2, t(165.13) ⫽ 3.64, p ⬍ .001 and Y3, t(138.69) ⫽ 4.81,
p ⬍ .001. As a result of these gender differences, we added gender
as a control variable.
Boys were also more likely to nominate a competitive game,
with 52.8% of boys compared with 16.5% of girls listing a competitive game among their favorites (2(1) ⫽ 24.09, p ⬍ .001).
Because of this disparity, and the fact that so few girls nominated
a competitive game, we chose to additionally run our main analyses separately among boys and girls.
Gaming, Competitive Gaming, and Social
Competencies
Two regression models were run for each of the three social
competencies: one model with the dichotomous competitive gaming variable as main predictor, and a second which added the
interaction between competitive gaming and gaming frequency.
Table 3 contains the correlations between the predictor and predicted variables used in the first models. In line with our predictions, competitive gaming was associated with decreases in conduct ( ⫽ ⫺.20, t ⫽ ⫺2.66, p ⫽ .009) and peer problems
Gender-Specific Outcomes
This study investigated the relationship between playing competitive video games and changes in children’s social competencies. Children who reported playing a competitive video game
showed improvements over one year in conduct problems and peer
relationships. No associations were observed between competitive
gaming and changes in prosocial behavior. Gaming frequency did
not moderate any of these findings. The observed main effects of
competitive gaming support the notion that, like more traditional
forms of competitive play, competitive gaming may provide a
context for the development of adaptive social competencies.
Changes in prosocial behavior were not associated with competitive gaming. This is in line with neither past findings that
competitive gaming negatively predicts prosocial behavior
(Ewoldsen et al., 2012) nor our hypotheses to the contrary. Past
research describes competitive gaming as a domain that promotes
antisocial cognitions and behaviors (Schmierbach, 2010). In this
light, competitive gaming encourages children to view relationships as being adversarial, and helping behaviors as being costly.
However, as our null findings indicate, competitive gaming is
likely more complex. For one, competitive gaming requires a
certain fundamental level of cooperation; players must collectively
abide by the rules of the game. Second, competitive gaming
sometimes allows for cooperation and prosocial goals (Adachi &
Willoughby, 2013). Team-based competitive play requires coop3
The observed pattern of results remained the same when only including
children who played for more than two (89.65%, n ⫽ 156) or more than
three hours (77.58%, n ⫽ 135) per week.
4
Violent gaming was computed as a dichotomous variable using the
same approach as in Lobel, Engels, Stone, Burk, and Granic (2017). As
described in Lobel et al. (2017), there was some debate whether Minecraft
be considered a violent game. In the reported analyses, Minecraft was not
classified as a violent video game; however, the pattern of results was
identical when Minecraft was coded as a violent video game.
5
With one exception, these patterns of results were identical when using
parent reports for children’s gaming frequency. The only divergent finding
was that parental reports of children’s gaming hours per week was not
associated with changes in children’s prosocial behavior ( ⫽ ⫺.01,
t ⫽ ⫺0.22, p ⫽ .825).
GAINING A COMPETITIVE EDGE
81
Table 2
Gender Differences in Gaming Frequency and Social Competencies at Y2 and Y3
Y2
Boys
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Conduct problems
Peer problems
Prosocial behavior
Gaming frequency
Y3
Girls
Boys
Girls
M
SD
M
SD
t
p
M
SD
M
SD
t
p
1.08
1.06
6.65
7.62
1.56
1.25
1.41
5.84
0.68
0.91
7.13
4.71
1.41
1.30
1.17
4.60
1.72
0.74
⫺2.33
3.64
.089
.460
.021
⬍.001
0.96
1.12
6.51
7.76
1.24
1.39
1.66
5.95
0.58
0.71
7.08
3.98
1.24
1.27
1.31
3.81
1.96
1.98
⫺2.41
4.81
.052
.050
.017
⬍.001
eration despite players pursuing competitive, arguably antisocial
goals. This allows for prosocial behaviors amid competition, for
example, in games that specifically enable players to heal or
protect their teammates. Similarly, competitive games may vary
greatly in terms of violent content. The competitive games that
were popular among children in this study were generally nonviolent. This may account for the discrepancy between our findings
and past research, although we also observed that violent gaming
was unrelated to changes in social competencies. Future studies
should investigate the conditions under which competitive games
may positively or negatively influence prosocial behavior in the
long-term; cooperation and violent content may be relevant mediators.
No dosage effects were observed as the frequency of competitive gaming seemed unrelated to changes in social competencies.
This may also speak to the complex nature of competition. For
example, the relationship between competitive gaming and social
competencies may be nonlinear such that competitive play may be
beneficial in small doses, whereas high levels of competitive
gaming may be detrimental. Competitive gaming may therefore be
best in moderation; a little amount may provide valuable contexts
for moral development and bonding, but excessive competition
may foster an unhealthy lens through which children perceive their
social environment.
The relative psychosocial impact of violence and competition in
gaming remain a source of debate in the literature. In one series of
lab experiments, competitive video games were compared with
competitive video games high in violence, and playing the latter
led to greater levels of aggressive cognition, affect, and behavior
(Anderson & Carnagey, 2009). On the other hand, studies conducted by Adachi and colleagues have indicated the opposite
(Adachi, 2015); when systematically controlling for violent content, competitive games led to more aggressive outcomes than
noncompetitive games (Adachi & Willoughby, 2011). The present
study adds to the complexity of these outcomes. First, violent
gaming in this study was not associated with increases in aggressive or antisocial tendencies. Moreover, competitive gaming was
associated with positive outcomes.
Table 3
Correlations Between Social Competency, Gender, Age, and Gaming Measures at Y2 and Y3
Y2
Y2
Peer
r
Prosocial
r
Gender
r
Age
r
Frequency
r
Violent
r
Competitive
r
⫺.13
⫺.06
.18ⴱ
—
—
—
—
.04
.09
⫺.01
.19ⴱ
—
—
—
.21ⴱⴱ
.06
⫺.09
⫺.27ⴱⴱ
.04
—
—
.07
.11
⫺.12
⫺.48ⴱⴱ
.03
.19ⴱ
—
.15
⫺.08
⫺.20ⴱ
⫺.38ⴱⴱ
.03
.05
.41ⴱⴱ
.31ⴱⴱ
—
—
—
—
—
—
⫺.11
⫺.11
—
—
—
—
—
Y2
Conduct
r
Peer
r
Prosocial
r
Frequency
r
Conduct
Peer
Prosocial
Gender
Age
Frequency
Violent (0, 1)
Competitive (0, 1)
.61ⴱⴱ
.36ⴱⴱ
⫺.12
⫺.16
.04
.16ⴱ
.00
⫺.21
.30ⴱⴱ
.54ⴱⴱ
⫺.19ⴱ
⫺.16ⴱ
.07
.15
⫺.00
⫺.15
⫺.05
⫺.24ⴱⴱ
.68ⴱⴱ
.19ⴱ
.05
.07
⫺.12
⫺.22ⴱⴱ
.23ⴱⴱ
.15
⫺.21ⴱⴱ
⫺.40ⴱⴱ
⫺.09
.39ⴱⴱ
.23ⴱⴱ
.17ⴱ
Conduct
Peer
Prosocial
Gender
Age
Frequency
Violent (0, 1)
Y3
Note. Gender coded as boys ⫽ 1; girls ⫽ 2. Correlations do not control for gender.
ⴱ
p ⱕ .05. ⴱⴱ p ⱕ .01.
LOBEL, ENGELS, STONE, AND GRANIC
82
Table 4
Gender-Specific Outcomes for the Associations Between Competitive Gaming and Changes in Social Competencies
Boys
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Conduct
Peer
Prosocial
Girls
Competitive ⫻ Frequency
Competitive
Competitive ⫻ Frequency
Competitive

t
p

t
p

t
p

t
p
⫺.29
⫺.28
⫺.05
⫺2.49
⫺2.46
⫺0.58
.015
.016
.566
.15
.16
⫺.03
1.04
1.17
⫺0.23
.301
.248
.818
⫺.04
⫺.12
⫺.08
⫺0.45
0.05
⫺0.73
.654
.961
.470
⫺.07
⫺.04
⫺.08
⫺0.74
⫺0.36
0.59
.461
.718
.557
One explanation for this discrepancy could lie in the measures
used in these studies. Utilizing the SDQ, the present study measured conduct problems, which may reflect a broader array of
psychological processes than simply aggressive tendencies. Only
one item on the SDQ’s conduct problems scale concerns interpersonal aggression (“fights with other children or bullies them”),
whereas the others may be more broadly related to children’s
ability to control impulsive urges or to obey social rules (e.g.,
“generally obedient . . . ” and “often has temper tantrums”).
Another explanation could be due to study design. The experimental designs like those used by Anderson and Carnagey (2009) and
Adachi and Willoughby (2011), may lack ecological validity. As
raised in the introduction, competitive gaming in the lab may miss
the social function that competitive gaming has when voluntarily
played in one’s home. In the latter, competition can be framed
within a broader context of self-improvement and comradery with
fellow players.
Our sample’s age may also be highly relevant. First, to our
knowledge, no longitudinal studies have been conducted regarding
the effects of competitive gaming among preadolescent children.
Second, given their age, children in this study were likely restricted by their parents in the video games they could play.
Preadolescent children do not typically have money to buy their
own games, and parents may have the final say in what games their
children play. This allows parents to socialize their children, with
some perhaps allowing or even preferring that their children game
with others.
Parental influence is also important to consider with regards to
gender. Compared with girls, boys generally showed greater deficiencies in their social competencies. Boys also spent more time
gaming than girls, and were far more likely to play competitively
than girls. These gender differences in social development are
commonly observed (Zimmer-Gembeck, Geiger, & Crick, 2005).
Moreover, boys seem to generally prefer competitive play more
than girls (Lever, 1976; Hartmann & Klimmt, 2006; Greenberg,
Sherry, Lachlan, Lucas, & Holmstrom, 2010; Olson, 2010). This
could indicate that competitive play is of particular relevance for
male social development. For example, boys are more likely to
become aggressive during competitive play (Ensor et al., 2011),
which may indicate that competitive play is a better testing ground
for them to develop their emotion regulation skills. Likewise,
parents may be socializing their children along gender stereotypical lines, giving girls more cooperative games, and boys more
competitive ones.
In line with past research, while about half of the boys in this
sample played at least one competitive video game, only a small
minority of girls did (Hartmann & Klimmt, 2006; Olson, 2010).
Whereas nearly 50% of boy gamers listed a competitive game
among their favorites, just 13, compared with 66 girl gamers, listed
a competitive game. As a result, it is difficult to interpret the null
finding among girls that competitive gaming was unrelated to
changes in social competencies. Notably, competitive gaming was
associated with improvements in conduct and peer problems both
among boys and across our entire sample (when controlling for
gender). Paradoxically, despite improvements in peer problems
among competitive boy gamers, however, boys showed more peer
problems than girls at this study’s final measurement point. Although difficult to interpret, this may indicate a greater need to
buffer male children’s ability to relate to their peers. Given the
observed findings and boys’ proclivity toward video games, it is
possible that competitive gaming helps meet this demand.
Although competitive gaming may be uniquely beneficial to
boys, it is also possible that competitive video games are disproportionately designed for male audiences. For example, the most
popular competitive game in our sample was a soccer video game
(Fifa); in the Netherlands, soccer is a predominantly male sport. It
may therefore be important for studies to identify the competitive
games favored by girls and to investigate their potential influence
on girls’ social competencies. Similarly, it may be worthwhile for
game designers to develop competitive games that target female
audiences.
Limitations and Future Directions
This study had a number of limitations. First, our competitive
gaming variable allowed some ambiguity. We chose to determine
competitive gaming based on children’s nominations, a method
used similarly in other studies (Adachi & Willoughby, 2016). This
is a more naturalistic and likely less subjective method than using
a Likert scale. However, children were able to nominate more than
one video game among their favorites. Thus, although all children
who nominated a competitive video game were considered competitive gamers, some of these children were likely more inclined
to play competitively than others. Similarly, for our moder...
Purchase answer to see full
attachment