Who plays violent video games? An exploratory analysis of predictors of playing violent games

Abstract

For roughly two decades, academics, politicians, and the media have debated the relationship between playing violent video games and engaging in violent acts. Despite the extensive attention paid to this possible outcome, no such spotlight has been placed on what leads to youth playing violent video games, which would provide the necessary context for potential later effects. The present study uses five datasets that include over 19,000 American youth in fifth through twelfth grades to provide an exploratory, inductive investigation into the predictors of playing violent video games. The results identify several themes of predictors of violent game play, including gender, family, health and nutrition, and various other social factors. These findings provide a foundation for future research to investigate and test these possible relationships.

Full text

Who Plays Violent Video Games?
An Exploratory Analysis of Predictors of Playing Violent Games
Whitney DeCamp
Western Michigan University
Abstract
For roughly two decades, academics, politicians, and the media have debated the relationship
between playing violent video games and engaging in violent acts. Despite the extensive
attention paid to this possible outcome, no such spotlight has been placed on what leads to youth
playing violent video games, which would provide the necessary context for potential later
effects. The present study uses five datasets that include over 19,000 American youth in fifth
through twelfth grades to provide an exploratory, inductive investigation into the predictors of
playing violent video games. The results identify several themes of predictors of violent game
play, including gender, family, health and nutrition, and various other social factors. These
findings provide a foundation for future research to investigate and test these possible
relationships.
DeCamp, Whitney. (2017). Who Plays Violent Video Games? An Exploratory Analysis of
Predictors of Playing Violent Games. Personality and Individual Differences, 117, 260-266.
doi:10.1016/j.paid.2017.06.027
The final publication is available at:
http://dx.doi.org/10.1016/j.paid.2017.06.027
The data used in this research were collected by the University of Delaware Center for Drug and
Health Studies as part of studies supported by the Delaware Health Fund, by the Delaware
Division of Substance Abuse and Mental Health, and by the Centers for Disease Control and
Prevention. The author wishes to thank the anonymous peer reviewers for their valuable insights
and suggestions.
1

Introduction
Violent video games have been a controversial subject for decades. Although violent and
gruesome acts in video games date back to the early days of the medium, the concern became
more public and widespread following the release of Mortal Kombat in 1992. Moreover, that
particular game is often credited (e.g., Crossley, 2014) as having started a moral panic over
violence in video games and launching political interest in the subject, as well as leading to the
creation of ratings groups, such as the Entertainment Software Rating Board, for video games.
More recently, state laws designed to limit minors’ ability to purchase violent games resulted in a
Supreme Court of the United States case that concluded that there was not convincing evidence
that video games cause violence (Brown v. Entertainment Merchants Association, 2011). Today,
the controversy continues unabated, with an increasingly common focus on perpetrators of mass
shooting having played violent video games. Several major media outlets, for example,
published news articles stating that the 2016 Munich shooter had played a game in the Counter-
Strike series. Many of these stories (e.g., Reuters, 2016) included a quote from a high ranking
police authority stating that the game is “played by nearly every known rampage killer.” To date,
the Counter-Strike franchise has sold over 60 million copies (SteamSpy, n.d.), so even if the
claim from the quote is factual (no evidence is available to suggest that it is), it omits the context
that such perpetrators would represent less than one-hundredth of one percent of this one game’s
players. Similarly, following the 2012 mass shooting in Newtown, Connecticut, President Obama
called for funding for research into the connection between violent video games and gun
violence (Molina, 2013). In contrast, connections between violent games and mass shootings has
been described as a myth in scholarly research (Fox & DeLateur, 2014).
During the same time period that video games have been the focus of media and political
attention, much research on the subject has been conducted and published. A sizable amount of
this research has focused on aggression, showing that, for example, playing violent video games
results in short-term rises in aggressive behavior in comparison to a non-violent control
(Bartholow, Sestir, & Davis, 2005), though some studies also find no difference (McCarthy,
Coley, Wagner, Zengel, & Basham, 2016). A meta-analysis of hundreds of such studies supported
the presence of such a link (Anderson et al., 2010), though later re-analysis argued that bias led
to an overestimation of the effect (Hilgard, Engelhardt, & Rouder, in press). The American
Academy of Pediatrics (2016) found this evidence strong enough to issue a policy statement,
noting that there is “proven scientific connection between virtual violence and real-world
aggression” (p. 5) and lamented that “unfortunately, media reports frequently present ‘both sides’
of the... issue” by consulting “a contrarian academic” (p. 3). Despite this bold stance, it remains
quite debatable whether the relationship found in research extends to real-world violence.
Specifically, a growing number of studies examining actual violence rather than aggression
within an experiment find that the correlation that exists is largely or entirely lost after
controlling for other influences (Breuer, Vogelgesang, Quandt, & Festl, 2015; DeCamp, 2015;
DeCamp & Ferguson, 2017; Gunter & Daly, 2012; Przybylski & Mishkin, 2016; von Salisch,
Vogelgesang, Kristen, & Oppl, 2011; Wallenius, & Punamäki, 2008; Ward, 2010). Recent meta-
analyses examining the connection with violence found null to trivial effects from video games
(Ferguson, 2015; Furuya-Kanamori & Doi, 2016). Thus, the effects of violent video games on
violence or long-term aggression remain unclear.
One possible reason for differing conclusions, even when examining the same outcome,
is that there may be a spurious relationship between playing violent video games and engaging in
violence (DeCamp, 2015; Gunter & Daly, 2012). That is, a youth who chooses to play violent
2

games or is permitted by his/her parents to play such games may already be more prone to
violence irrespective of and prior to playing such games. The present study examines the
predictors of playing violent games in order to begin building towards an understanding of the
mechanisms that might lead some youth toward violent games more than others.
Predictors of Violent Video Game Play
There have been hundreds of studies that have examined violent video games, yet nearly all of
them have focused on the potential outcome of playing violent video games. In contrast, there
has been limited investigation into the matter of who plays violent video games. Although
research into this area is rare, a few studies have examined this topic previously.
Demographical and media consumption behaviors have been investigated as predictors of
violent game play. Not surprisingly, boys have been found to be significantly more likely than
girls to play violent video games in many studies (e.g., Kasumovic, Blake, Dixson, & Denson,
2015; Olson et al., 2007), and the effect from gender is notable as being a substantively powerful
predictor as well (Olson et al., 2007). Additionally, indicators of greater prevalence of media
consumption (including playing in the bedroom and playing more hours per day) and social
gaming (playing with friends and playing with strangers online) also positively correlated with
playing violent video games (Olson et al., 2007).
In more personality-focused research, one study found that time spent playing violent
video games is positively correlated with being more open and with being less agreeable (Chory
& Goodboy, 2011). In addition to observing these same effects again, examining whether
someone’s favorite game was violent also identified positive relationships between violent video
games and extroverted and neurotic personality types (Chory & Goodboy, 2011). Violent game
play has also been connected to sexuality, with individuals (adults) who have a greater interest in
sexual intercourse (Kasumovic et al., 2015). Research also finds that women who play violent
video games rate themselves as more desirable to men than those who do not play violent games
(Kasumovic et al., 2015).
What has more rarely been examined are social factors and behaviors not directly
connected to media. For example, although it has been found that there is no significant
correlation (positive or negative) between playing video games with parents and playing violent
video games (Olson et al., 2007), the parental relationship in general has not been explored as an
influence on violent media consumption. This is particularly at odds with the justification for
legislation over violent video games relating to parental roles in deciding what media is
acceptable for children (see Brown v. Entertainment Merchants Association, 2011, or Justice
Thomas’s dissent thereof). The parental and social context for playing violent video games is a
meaningful part of understanding this issue. Such a context for an individual might include
parents who encourage or are permissive of consuming violent media, peer pressure to play the
same types of games as friends, norms and expectations based on gender or social groups, etc.
These social forces can influence both desires relating to media and other aspects of personality.
If, for example, research were to one day conclude that there is a causal relationship between
playing violent video games and engaging in violent behavior, then understanding the antecedent
causes of violent game play would be invaluable in offering a full-range of implications. If, on
the other hand, it were determined that the relationship between playing violent video games and
violent actions were spurious, then understanding the mutual predictors would again be
beneficial as they relate to an underlying desire toward violence. The present study investigates
3

this context for playing violent video games through the use of datasets encompassing many
social and familial variables with large numbers of youth participants at various ages.
Material and Methods
The data used for these analyses come from a variety of surveys conducted in public and public-
charter schools in the state of Delaware between January and June, 2015. These surveys include
the fifth, eighth, and eleventh grade Delaware School Survey (DSS), as well as the middle school
and high school Delaware Youth Risk Behavior Survey (YRBS). These surveys were selected for
this study because they include large samples and are omnibus surveys that cover many areas
suitable for predicting behaviors. The DSS surveys are designed to be a census of students
present on the day of administration in their respective grades, minus students who refused to
participate, who were denied parental consent, or who were in a classroom randomly selected to
receive a YRBS survey instead (eighth and eleventh grades only). The YRBS surveys are
designed to include a random sample of classrooms in each school, again minus students who
refused to participate or who were denied parental consent. Under the institutional review board
approved protocol, parental consent is obtained passively (i.e., parents must notify the school if
they object), and refusal to participate (by parents or students) is rare. All surveys were
administered in classrooms during a normal class period. The response rates ranged from a low
of 97% to a high of 99%. The sample sizes were 6,934 for fifth grade, 5,133 for eighth grade,
3,886 for eleventh grade, 3,102 for middle school, and 2,777 for high school.
Variables
The main variable of interest for this study is playing violent video games. This was
measured using the question, “how often on average do you play violent video games, such as
games that are rated M?” This question was includes on all surveys at all grade levels.
Participants were given a series of responses ranging from never to more than ten hours per
week. Because this study is interested in the decision to play violent video games rather than the
amount of game play (which may be affected by other factors than merely the desire to play
them), a dichotomous recode indicating whether the participant never or ever plays violent video
games is used for these analyses.1 Descriptive statistics for playing video games by grade are
displayed in Table 1.
The independent variables used in these analyses are too numerous to be listed
individually (see below for a discussion of this design choice and why it is appropriate).
However, question numbers are provided in the Appendix to allow for replication and for further
understanding of this study’s exact design. Rather than select variables to include, the
approached used here was to use all variables in the datasets except for those specifically chosen
to be excluded. The variables selected to be withheld from analysis were those associated with
possible outcomes, including violent or deviant acts and substance use, as well as variables that
would be methodologically challenging to use in a regression, such as ZIP code.
1 The second response category was “very rarely.” This answer could, depending on interpretation by a
participant, be inclusive of individual who tried, but did not like or continue, playing violent games. The
analyses performed here were alternative performed using a recode that compared never/very rarely to the other
responses (i.e., switching which category very rarely responses were coded as). The results were markedly
similar and would lead to the same conclusions.
4

Table 1: Prevalence Rates by Grade, Gender, Regularity, and Dataset
Ever†5th 6th 7th 8th 9th 10th 11th 12th
DSS 58.3 --- --- 67.4 --- --- 62.2 ---
Males 79.2 --- --- 89.8 --- --- 88.1 ---
Females 37.9 --- --- 45.5 --- --- 37.8 ---
YRBS --- 62.1 64.4 69.7 65.6 64.6 59.5 60.1
Males --- 84.7 87.7 91.8 90.6 90.9 89.6 83.3
Females --- 43.0 43.7 47.1 41.9 41.4 33.1 36.8
Weekly 5th 6th 7th 8th 9th 10th 11th 12th
DSS 32.7 --- --- 42.4 --- --- 37.4 ---
Males 52.3 --- --- 69.0 --- --- 63.7 ---
Females 13.6 --- --- 16.8 --- --- 12.6 ---
YRBS --- 36.9 39.1 45.7 40.7 39.3 34.2 39.0
Males --- 62.0 66.4 70.4 69.0 69.7 61.4 63.3
Females --- 15.6 14.7 20.0 14.1 12.1 10.2 14.2
† “Ever” refers to any response other than “never” and does not necessarily reflect “lifetime” prevalence.
Analyses
The dependent variable used in these analyses is dichotomous, thus logistic regression is
used for estimation. In order to achieve an ideal model that is as reasonably parsimonious as this
design allows, backward elimination stepwise regression is used. In this approach, all variables
are entered into the model initially and are eliminated beginning with the least significant until
only significant (p < .05, in this case) effects remain in the model. Because different questions
were asked on different surveys, each dataset is analyzed separately. This also provides some
added reliability through a split sample-like approach, allowing common themes to emerge
across datasets in addition to within each.
Missing data were handled differently depending on the variable in question. Cases
missing on the dependent variable were simply removed from the datasets entirely, resulting in
final sample sizes of 6,734 for fifth grade, 4,181 for eighth grade, 3,179 for eleventh grade, 2,981
for middle school, and 2,566 for high school. The larger number of eliminations from this seen
for eighth and eleventh grades relative to the other datasets is due to some students being unable
to finish the survey within one class period, as it was the longest survey and the video game
question appeared in the final quarter of the survey (page 13 of 16). For the independent
variables, missing data were imputed using SAS’s multiple imputation procedure (PROC MI).
The variance inflation factor statistics for the final regression models did not indicate any
concerns with multicollinearity in the models.
The final analysis is effectively a content analysis of the variables found to significantly
predict playing violent video games. These variables are assessed for common themes within and
across models to identify possible antecedents of violent game play.
A Note on Study Design
An important consideration is that this study is both exploratory and quantitative. In some
respects, this can be considered “data mining,” though not in the derogatory sense that the term is
sometimes used. The goal of this research is not to test hypotheses or to reach conclusions about
5

what causes or even predicts violent game play. Although the models will indeed predict whether
participants play violent games or not, these findings cannot be the end result of this study given
the atheoretical design. Rather, such model results are used to generate an understanding of the
possible causes or predictors of violent game play. In essence, this is an exercise in theory
building, not theory testing. The thematic results that follow the statistical analysis thus become
the foundation for theory to be tested and refined in future research. Because there is a one-in-
twenty chance of an effect being erroneously found significant and 646 relationships being
tested, it is virtually certain that some of these relationships are due to chance alone. Thus, a
finding that a specific variable significantly predicts violent video game play should not be
interpreted as evidence of such a relationship, but rather as evidence that future research should
investigate such a possibility.
Model Estimation
The fifth grade model initially included 122 predictors with an R2 of .385.2 After backward
stepwise elimination of non-significant effects, the final model included 33 predictors with an R2
of .374. The eighth grade model initially included 171 predictors with an R2 of .465. After
backward stepwise elimination of non-significant effects, the final model included 30 predictors
with an R2 of .443. The eleventh grade model initially included 171 predictors with an R2 of .475.
After backward stepwise elimination of non-significant effects, the final model included 25
predictors with an R2 of .434. The middle school model initially included 123 predictors with an
R2 of .537. After backward stepwise elimination of non-significant effects, the final model
included 20 predictors with an R2 of .503. The high school model initially included 59 predictors
with an R2 of .420. After backward stepwise elimination of non-significant effects, the final
model included 13 predictors with an R2 of .409. In each of these estimations, the majority of
predictors were deemed non-significant and removed without a more than trivial impact on the
model’s predictive power. The results from each model are presented in Tables 2-6.3
Thematic Results
The final models indicated that there were 121 significant effects across the five datasets.
After analyzing the content of the questions asked, seven major (not counting tautological
effects) and a few minor themes emerged.
2 All R2 values presented for these estimations are Nagelkerke pseudo-R2 values.
3 All significant effects are shown, even when the effect size is trivial. Instituting a minimum effect size was
considered, but not ultimately used because of the specificity of the indicators. For example, there are multiple
indicators about food consumption. Although their individual effects are sometimes weak, they collectively may
suggest a relationship worth investigating.
6

Table 2: Fifth Grade Logistic Regression Results Predicting Violent Game Play
Variable Std. β Q#
Female -.548** 2
Time online or playing games .255** 22
Amount of soda consumed .145** 67
School grades -.075** 12
Time physically playing/exercising .070** 24
Father smokes .066** 75c
Teachers provide support/encouragement -.063** 102d
Amount of sleep -.057** 104
Time reading for pleasure -.056** 25
Parents said to not drink -.056** 80b
Hispanic/Latino .055** 4
No one at home smokes -.055** 75a
Kids obey teachers at school -.054** 31
Friends provide support/encouragement .053** 102g
Amount of exercise .052** 101
Times moved since kindergarten .049** 13
Parents talk about things that matter -.048** 60
Feel afraid -.047* 19
Family member recently in military .047** 7b
Mother’s age -.046* 8
Feel really worried .046* 18
Amount of fruit juice consumed .046* 64
Friends smoke .043* 39
Parents enforce rules -.042* 54
Amount wears seat belt -.042* 103
Sibling(s) work(s) to pay the bills .041* 11d
Mother works to pay the bills .041* 11a
Could get cigarettes from vending machine .040* 74g
Feel safe at school -.037* 28
Family member recently in jail/prison .037* 7a
Lives with mother -.036* 10a
Parents know what I do -.035* 37
Could get cigarettes from store clerk .034* 74h
** p < .01, * p < .05
7

Table 3: Eighth Grade Logistic Regression Results Predicting Violent Game Play
Variable Std. β Q#
Female -.714** 2
Time online or playing games .210** 182
Perceived risk from occasional marijuana -.114** 147
Perceived risk from regular marijuana .096* 148
Family member recently in military .083** 7
Amount of non-diet soda consumed .077** 190
Amount of exercise .075** 196
Parents take interest in my activities -.073** 172
Parents monitor Internet or phone use -.072** 199k
Race: Asian -.070** 6b
Amount of sleep -.070** 193
In school club or activity -.068** 170a
Amount hit by another teen to hurt .068** 162
Not okay to do the wrong thing -.068** 169i
In other club -.068** 170e
Important to not hurt people -.064* 169e
Mother smokes .062** 65b
Race: Black -.061** 6c
Amount of fruit juice consumed .059* 186
Feel nervous -.057* 181
In boxing or martial arts .056** 170m
Feel safe in neighborhood -.053* 28
Amount of milk consumed .053* 191
Student respect teachers at school -.052* 32
Could get cigarettes from adults .049* 71e
Amount see/hear media message against teen drinking .048* 62
Father smokes .048* 65c
Taking art lessons .048* 170j
Father lost job .047* 18a
Taking bipolar prescription .046* 93f
** p < .01, * p < .05
8

Table 4: Eleventh Grade Logistic Regression Results Predicting Violent Game Play
Variable Std. β Q#
Female -.747** 2
Time online or playing games .215** 182
Time reading for pleasure .115** 184
Amount of non-diet soda consumed .108** 190
Important to help friends .106** 169c
Cares about doing well in school -.096** 169a
Not okay to do the wrong thing -.092** 169i
Mother smokes .078** 65b
Witness crime in neighborhood .077** 164
Father works to pay the bills .076** 10b
Parents provide support/encouragement .076** 168b
Employed -.075** 170o
Race: White .074** 6d
I get along well with my parents .074* 31
Time doing school work -.072* 183
Step-parent(s) work(s) to pay the bills .072** 10e
Important to not hurt people -.070* 169e
School rules are fair -.068* 35
Amount of fruit juice consumed .067* 186
Taking allergy prescription .061* 93e
Bullied in neighborhood .061* 166
Race: Native American .059* 6a
Taking asthma prescription .059* 93c
Perceived risk from trying marijuana -.057* 146
Friends’ parents provide support/encouragement -.056* 168e
** p < .01, * p < .05
9

Table 5: Middle School Logistic Regression Results Predicting Violent Game Play
Variable Std. β Q#
Female -.663** 3
Time online or playing games .293** 76
Amount of non-diet soda consumed .101** 69
Amount of sleep -.095** 81
School grades -.092** 82
Amount wears seat belt -.083** 26
Self-injury (non-suicidal) .082** 35
Diagnosed as having asthma .068** 79
Amount of milk consumed .064* 70
Amount of caffeinated drinks .062* 72
Thought about suicide -.060* 36
Diagnosed as having difficulty concentrating -.054* 90a
Time watching TV .053* 75
** p < .01, * p < .05
Table 6: High School Logistic Regression Results Predicting Violent Game Play
Variable Std. β Q#
Female -.754** 3
Time online or playing games .383** 110
Bet on video games .129** 113h
Not diagnosed with a condition -.123** 20h
Not taking medication for a condition .111* 21h
Days ate breakfast -.105** 107
Parents listen -.103** 143
Amount of milk consumed .098** 104
Time watching TV .093** 109
Difficulty waling or climbing stairs .093** 18
In church group -.082** 115g
Could talk to parents about problems .082* 123b
Amount of fruit juice consumed .081** 103
Difficulty concentrating due to condition .076* 17
Feel sad -.075* 147
In school club or activity -.074* 115a
Diagnosed as having difficulty concentrating -.073* 19c
Race: Black -.071* 6c
Diagnosed as having asthma .067* 119
Family member recently in military .065* 13
** p < .01, * p < .05
10

Gender
Above all other variables, one alone stands as a consistent and very powerful predictor of
violent game play: gender.4 In all five models, gender was the strongest predictor and noticeably
more powerful than any other effects. Specifically, the effect from gender was two to three times
stronger than the effect from the time spent playing video games. Ignoring tautological effects
(see below), the difference is three to six times stronger than the most powerful other effect.
Tautological Effects
Several of the variables included in the models are very closely connected with playing
violent video games. These include the time spent on the Internet or playing video games, the
time spent watching television (which, depending on participant interpretation, may include
playing video games because watching a television or other video monitor is necessary), and
betting on video games. Although these variables could be excluded from the model, they were
included so as to act as control variables. Obviously, playing video games and using a television
(or other video monitor) is a prerequisite for playing violent video games, and therefore the
causal/necessary effect from these variables is not in question.
Family Context
A number of variables about the type of family one has emerged as significant in the
models. The most straightforward of these is whether anyone in the family “recently” served in
the military. Additionally, several predictors relating to various family members working to
support the family were also significant, with a positive relationship with playing violent games.
Moreover, youth whose father lost his job, who experienced a family member recently being in
jail/prison, or who moved frequently also have an increased probability of playing violent video
games. This suggests that those lacking support networks, role models, etc., may be more
inclined to play violent games. Conversely, those who reporting living with their mothers had
lower odds of playing violent games, and those who reported an older age for their mother were
also less likely to play. A maternal role model, especially older ones, is thus associated with
lower propensity towards violent game play.
Family Relationships and Support Relationships
Within this category, there are mixed findings. Youth who report that their parents take an
interest in their activities, talk to them about things that matter, monitor their Internet or phone
use, listen to them, enforce rules, or know what the youth does are all less likely to play violent
games. Conversely, those who report that their parents provide support or encouragement, that
they could talk to their parents about anything, or that they get along with their parents are more
likely to play violent games. The difference between these sets of variables may relate to
differing parenting styles. Specifically, measures that relate to parents acting authoritatively
correspond to lower odds of violent game play, whereas measures relating to more permissive,
parent-as-friend styles correspond to higher odds. Further research in this area is needed to
determine this potentially complex relationship. With regard to support relationships, those
whose friends provide support/encouragement are more likely and those whose friends’ parents
4 Two surveys used the word “sex,” two surveys used the word “gender,” and one used “boy” or “girl” in the
question itself in place of the categorical type. Regardless, it is unlikely that many youth would make a
distinction between sex and gender without an explanation of the difference (none was provided). For simplicity,
the results presented here simply refer to gender.
11

provide support/encouragement are less likely to play violent games, suggesting an effect from
having adult, rather than peer, support system.
Social Environment
Youth who feel safe at their school, feel safe in their neighborhood and find support from
teachers were all less likely to play violent games. Those who report attending a school where
students obey teachers and treat teachers with respect were also at lower odds. Conversely, those
who reported having witnessed a crime in their neighborhood were more likely to play violent
games. Together, these findings suggest that a safe and nurturing environment decreases the odds
of youth playing violent games.
Health and Nutrition
Although each predictor within this theme is individually weak, this category contains the
largest number of significant predictors and therefore has a substantial gross influence on the
odds of playing violent games. Youth who reported having been diagnosed with asthma were
more likely to play violent games, as were youth who reported difficulty concentrating or
walking, or who were taking medication for allergies, asthma, or bipolar disorder. Conversely,
those who reported having not been diagnosed with a condition and those who were diagnosed
with difficulty concentrating were less likely. With regard to the latter, it is possible that those
who were diagnosed were getting treatment to correct the problem, whereas those who simply
reported experiencing the problem might not all have been diagnosed, thus explaining these
seemingly contradictory findings. Additionally, those who reported non-suicidal self-injury were
more likely to play violent games and those who reported suicidal thoughts were less likely.5
Across several models, youth who reported greater frequency of fruit juice, milk, and
non-diet soda consumption were more likely to play violent video games. Each of these
beverages contains high amounts of sugar. Youth who reported greater consumption of drinks
with caffeine were also at higher odds. It is possible that these relationships are biochemical in
nature, are spurious relationships stemming from differing approaches to parenting with regard to
nutrition, or co-occurring recreational choices (sugary and caffeinated drinks are stereotypically
associated with video games), though the present data are unable to provide insights into the
nature of this relationship to this level of detail, so determining which of these explanations
apply is beyond the scope of this study. Youth who reported greater amounts of sleep and
regularly wearing their seat-belts in cars were less likely to play violent games. The latter
especially suggests influence from parenting style rather than a direct relationship.
Academics and Activities
Another very straightforward relationship is present between academics and game play:
youth who care about doing well in school, who spend more time doing schoolwork, and who get
better grades are less likely to play violent video games. The relationship between non-school
activities and game play is less clear from these analyses. Being in a group/club or being
employed, for example, has a negative relationship with violent games, whereas physical
exercise, boxing/martial-arts, and taking art lessons has a positive relationship. Reading for
5 Due to the omnibus nature of the survey (i.e., limited space for any one topic) and the age of the participants,
only simple, easy to understand questions were used, even for complex issues such as these. The single-item,
yes/no questions used for these behaviors were: “During the past 12 months, did you do something to purposely
hurt yourself without wanting to die, such as cutting, scraping, or burning yourself on purpose?” and “Have you
ever seriously thought about killing yourself?” These effects are difficult to interpret in light of the many known
correlates of self-injury and suicide ideation (Bakken, & Gunter, 2012; DeCamp & Bakken, 2016).
12

pleasure related to decreased odds of violent games for fifth grade youth, but increased odds for
eleventh grade youth.
Substance Use Messages
Although substance use itself was not analyzed in this study, messages and beliefs about
substance use were included. Youth who reported being around people who smoke or having
access to cigarettes were more likely to play violent games. Those who saw a message against
substance use in some form of media (which may or may not be video games) were also at
higher risk, whereas those who were told to not drink by their parents were at lower risk. The
effect from the perceived risk from smoking marijuana varies by the degree of marijuana use in
question, with perceived risk from trying or occasional use corresponding to decreased odds and
regular use corresponding to higher odds.
Other Factors: Feelings, Beliefs, Victimization, and Race
Youth who reported feeling afraid, nervous, or sad were less likely to play violent games,
whereas those who felt worried were more likely. Those who reported believing that it is
important to help friends were more likely, while those who reported believing that it is
important to not hurt people and that it is not acceptable to do “the wrong thing,” or that school
rules are fair were less likely. White, native American, and Hispanic students had higher odds,
whereas black and Asian students had lower odds. Finally, those who have been hit by another
youth or who were bullied in their neighborhood were also more likely to play violent video
games.
Discussion
Although much research has investigated whether violent video games have an effect on
aggression or violence (see Ferguson, 2015; Furuya-Kanamori & Doi, 2016), there has been only
limited interest in what leads youth to playing violent video games (for exceptions, see Chory &
Goodboy, 2011; Kasumovic et al., 2015; Olson et al., 2007). This is particularly important, as the
context for choosing to play violent games may help to explain the relationship with violence
and/or be useful in trying to prevent such game play should it be proven harmful. The present
study investigated this issue using data from youth at varying ages to identify possible predictors
of violent game play through an exploratory design.
The themes that emerged from the significant effects include: gender, family context,
family relationships and support relationships, social environment, health and nutrition,
academics and activities, substance use messages, and other factors. These themes are, in some
cases, quite nuanced, so complete elaboration on each here would be duplicative of the results
themselves. However, it is worth nothing that some of the consistent effects across models
included gender, sugar/caffeine consumption, physical activity, having family in the military, and
having a family member who smokes. Because some variables were present in all datasets and
others were not, this brief list should not be considered exhaustive of effects that could be
consistent had such testing been possible.
Although prior research in this area has been limited, there are a few comparisons that
can be drawn between the present study and those that preceded it. First among such
comparisons is the effect of gender, which was the strongest effect by a notable magnitude. The
strength of this effect is not surprising, and is undoubtedly consistent with prior research
(Kasumovic et al., 2015; Olson et al., 2007). There has been a large gender-gap in most computer
13

technologies for several decades, arguably stemming from the switch to marketing computer
technology, including video games, to boys and young men (Bennon, 2014). Modern video
games are typically marketed at a masculine audience, with few female characters and highly
sexualized female characters when present (Ivory, 2006; Near, 2013). More generally, violence,
whether fictional or real, is generally considered a masculine area, and so this cultural norm
would logically extend to violent video games as well.
Another consistency comes from the connection between violent games and families with
military backgrounds. The U.S. Army previously published a first person shooter (FPS) game as
a recruitment tool (White, 2005) and continues to consider and use video games in recruitment
strategies (Democracy Now, 2015). On the other hand, the findings of this study at least partially
diverge with prior research in that there was no significant connection with sexuality. Although
sexual intercourse itself was not among the effects tested, the high school dataset did include
indicators relating to sexting, which were not significant and thus did not survive for the final
model. This non-significance runs counter to prior findings that greater interest in sex is
associated with playing violent games (Kasumovic et al., 2015). Beyond these few comparisons
that can be made with prior research, most of the themes that emerged were previously not
examined as possible influences on violent game play and thus represent new areas for further
study.
As noted previously, the findings in this study should not be interpreted as evidence that
these relationships can be generalized beyond the samples used here. The atheoretical
exploratory design of these analyses presents too great a risk of erroneously finding significance
given the large number of variables used. Moreover, some variables might be indicators of
multiple underlying factors, and this research is unable to distinguish the true one that connects
to violent video games (e.g., sugary beverage consumption probably does not cause people to
play violent video games, but it may be indicative of more permissive parenting decisions or of
culturally learned behaviors relating to the gaming community). Rather, these findings provide
the foundation for future research, both qualitative and quantitative, to investigate and test these
potential relationships. Ideally, this would be done theoretically – building factors/constructs
from the newly-identified concepts that may be connected to choosing to or being allowed to
play violent games and then testing these relationships and corresponding hypotheses – or with
richer qualitative designs to better understand the mixed effects in some areas. Thus, this
research is designed not for policy implication or to reach generalizable conclusions, but as
implications for future research.
In addition to the potential lack of generalizability already discussed, a few other
limitations should also be noted. First, these data are cross-sectional and therefore time-order is
unclear in some cases. Possible outcomes from playing video games, such as deviance, violence,
substance use, etc., were not used as predictors, but other variables from gray areas (e.g.,
academic grades) were included and could be just as much an outcome as a predictor. Second,
these data do not include any indicators for whether the participants desire to play violent games
and whether their parents allow such game play. Theoretically, the measured outcome of playing
violent games would be observed only when both of these effects are present, although this is
potentially muddled by peer-pressure to do something not desired and disobedience that ignored
parental rules. A specific measure for whether parents allow youth to play violent games would
be a logical addition in future research.
14

Conclusion
Although the design used for this study does not permit conclusions about the causes or
predictors of playing violent video games (nevertheless, it is clear that gender is a major
influence), the findings here do permit conclusions about directions for future research. This
inductive, exploratory study identified several themes in the data that appear to connect to
violent game play. In order to determine whether these relationships truly exist, future research
should use the concepts identified within these themes and collect and test data better designed to
measure these concepts. Whether violent video games have a spurious, weak, or meaningful
impact on violent behavior remains a debated issue. Developing a better understanding of the
context in which violent game play occurs will help us to better understand the relationship with
real-life violence as well.
References
American Academy of Pediatrics Council on Communications and Media. (2016). Virtual
violence. Pediatrics, 138(2), 1-4.
Anderson, C. A., Shibuya, A., Ihori, N., Swing, E. L., Bushman, B. J., Sakamoto, A., & ...
Saleem, M. (2010). Violent video game effects on aggression, empathy, and prosocial
behavior in Eastern and Western countries: A meta-analytic review. Psychological
Bulletin, 136(2), 151-173. doi:10.1037/a0018251
Bakken, N. W., & Gunter, W. D. (2012). Self-Cutting and Suicidal Ideation among Adolescents:
Gender Differences in the Causes and Correlates of Self-Injury. Deviant Behavior, 33,
339-356. doi:10.1080/01639625.2011.584054
Bartholow, B. D., Sestir, M. A., & Davis, E. B. (2005). Correlates and consequences of exposure
to video game violence: Hostile personality, empathy, and aggressive behavior.
Personality and Social Psychology Bulletin, 31, 1573-1586.
Bennon, P. (2014, October 17). Episode 576: When women stopped coding. Planet Money.
Retrieved from http://www.npr.org/templates/transcript/transcript.php?
storyId=487069271
Breuer, J., Vogelgesang, J., Quandt, T., & Festl, R. (2015). Violent video games and physical
aggression: Evidence for a selection effect among adolescents. Psychology of Popular
Media Culture, 4(4), 305-328. doi:10.1037/ppm0000035
Brown v. Entertainment Merchants Association, 131 S. Ct. 2729 (2011). Retrieved from:
http://www.supremecourt.gov/opinions/10pdf/08-1448.pdf
Chory, R. M., & Goodboy, A. K. (2011). Is basic personality related to violent and non-violent
video game play and preferences?. Cyberpsychology, Behavior, and Social Networking,
14(4), 191-198.
Crossley, R. (2014, June 2). Mortal Kombat: Violent game that changed video games industry.
BBC News. Retrieved from: http://www.bbc.com/news/technology-27620071
DeCamp, W. (2015). Impersonal agencies of communication: comparing the effects of video
games and other risk factors on violence. Psychology of Popular Media Culture, 4, 296-
304. doi: 10.1037/ppm0000037
DeCamp, W., & Bakken, N. W. (2016). Self-Injury, Suicide Ideation, and Sexual Orientation:
Differences in Causes and Correlates among High School Students. Journal of Injury and
Violence Research, 8, 15-24. doi:10.5249/jivr.v8i1.545
15

DeCamp, W., & Ferguson, C. J. (2017). The Impact of Degree of Exposure to Violent Video
Games, Family Background, and Other Factors on Youth Violence. Journal of Youth and
Adolescence, 46, 388-400. doi:10.1007/s10964-016-0561-8
Democracy Now. (2015, November 20). From console to trigger: How Pentagon "exploits" video
game culture to wire youth for war. Democracy Now. Retrieved from
http://www.democracynow.org/2015/11/20/from_console_to_trigger_how_pentagon
Ferguson, C. J. (2015). Do angry birds make for angry children? A meta-analysis of video game
influences on children’s and adolescents’ aggression, mental health, prosocial behavior
and academic performance. Perspectives on Psychological Science, 10, 646-666.
Fox, J. A., & DeLateur, M. J. (2014). Mass shootings in America: moving beyond Newtown.
Homicide Studies, 18, 125-145.
Furuya-Kanamori, L., & Doi, S. A. (2016). Angry birds, angry children, and angry meta-analysts:
a reanalysis. Perspectives on Psychological Science, 11, 408-414.
Gunter, W. D., & Daly, K. (2012). Causal or spurious: using propensity score matching to
detangle the relationship between violent video games and violent behavior. Computers
in Human Behavior, 28, 1348-1355. doi: 10.1016/j.chb.2012.02.020
Hilgard J., Engelhardt C., Rouder J. (in press). Overestimated effects of violent games on
aggressive outcomes in Anderson et al. (2010). Psychological Bulletin.
Ivory, J. D. (2006). Still a man’s game: Gender representation in online reviews of video games.
Mass Communication and Society, 9, 103-114.
Kasumovic, M. M., Blake, K., Dixson, B. J., & Denson, T. F. (2015). Why do people play violent
video games? Demographic, status-related, and mating-related correlates in men and
women. Personality and Individual Differences, 86, 204-211.
McCarthy, R. J., Coley, S. L., Wagner, M. F., Zengel, B., & Basham, A. (2016). Does playing
video games with violent content temporarily increase aggressive inclinations? A pre-
registered experimental study. Journal of Experimental Social Psychology, 67, 13-19.
Molina, B. (2013, January 16). Obama seeks research into violent video games. USA Today.
Retrieved from: http://www.usatoday.com/story/tech/gaming/2013/01/16/obama-gun-
violence-video-games/1839879/
Near, C. E. (2013). Selling gender: Associations of box art representation of female characters
with sales for teen- and mature-rated video games. Sex Roles, 68, 252-269.
Olson, C. K., Kutner, L. A., Warner, D. E., Almerigi, J. B., Baer, L., Nicholi, A. M., & Beresin,
E. V. (2007). Factors correlated with violent video game use by adolescent boys and girls.
Journal of Adolescent Health, 41(1), 77-83.
Przybylski, A. K., & Mishkin, A. F. (2016). How the quantity and quality of electronic gaming
relates to adolescents’ academic engagement and psychosocial adjustment. Psychology of
Popular Media Culture, 5(2), 145-156. doi:10.1037/ppm0000070
Reuters. (2016, July 24). Munich gunman, a fan of violent video games, rampage killers, had
planned attack for a year. CNBC. Retrieved from: http://www.cnbc.com/2016/07/24/
munich-gunman-a-fan-of-violent-video-games-rampage-killers-had-planned-attack-for-a-
year.html
Steam Spy. (n.d.). Search Results: Counter-Strike. Retrieved August 13, 2016, from:
https://steamspy.com/search.php?s=counter-strike
von Salisch, M., Vogelgesang, J., Kristen, A., & Oppl, C. (2011). Preference for violent
electronic games and aggressive behavior among children: The beginning of the
downward spiral? Media Psychology, 14(3), 233-258.
doi:10.1080/15213269.2011.596468
16

Wallenius, M., & Punamäki, R. (2008). Digital game violence and direct aggression in
adolescence: A longitudinal study of the roles of sex, age, and parent-child
communication. Journal of Applied Developmental Psychology, 29(4), 286-294.
doi:10.1016/j.appdev.2008.04.010
Ward, M. R. (2010). Video games and adolescent fighting. Journal of Law & Economics, 53,
611-628.
White, J. (2005, May 27). It's a video game, and an army recruiter. Washington Post. Retrieved
from http://www.washingtonpost.com/wp-dyn/content/article/2005/05/26/
AR2005052601505.html
Appendix: Questionnaire Indicators
Due to the extensive number of unique questions used and limited space to present information
here, a full list of questions used will not be presented. There is undoubtedly information to be
gained from knowing what variables were not found to be significant as well, so the list below
reflects what question numbers were used. Copies of the questionnaires are available on the
survey projects’ webpages† or from the author upon request.
5th Grade: q2 to q14 q17 to q25 q28 to q40 q43 to q47 q50 q52 q54 to q62 q64 to q68 q74a to
q75f q79a to q80e q92 to q104
8th and 11th Grades: q2 to q6d q7 q9a to q10f q11 to q14 q17b to q18f q20a to q20c q21a to q37
q56 q57 q62 q65b to q65f q71a to q71g q92 to q93i q113a to q113d q142 to q151 q161 to q162
q164 to q167 q168b to q170o q171 to q184 q186 to q196 q199a to q199k
Middle School: q2 to q8 q11 to q23 q26 q27 q33 to q38 q65 to q83 q86 to q90d
High School: q2 to q8 q11 q13 to q21h q23 q30 q31 q35 to q48 q99 to q113i q115a to q117 q119
q120 q122 to q123i q125 to q131 q140 to q150
† https://www.cdhs.udel.edu/seow/school-surveys/delaware-school-survey-(dss)
https://www.cdhs.udel.edu/seow/school-surveys/youth-risk-behavior-survey-(yrbs)
17