CYBERPSYCHOLOGY, BEHAVIOR, AND SOCIAL NETWORKING

Volume 20, Number 12, 2017

ª Mary Ann Liebert, Inc.
DOI: 10.1089/cyber.2017.0364

For Video Games, Bad News Is Good News:

News Reporting of Violent Video Game Studies

Allen Copenhaver, PhD,1 Oana Mitrofan, MD,2 and Christopher J. Ferguson, PhD3

Abstract

News coverage of video game violence studies has been critiqued for focusing mainly on studies supporting
negative effects and failing to report studies that did not find evidence for such effects. These concerns were
tested in a sample of 68 published studies using child and adolescent samples. Contrary to our hypotheses, study
effect size was not a predictor of either newspaper coverage or publication in journals with a high-impact factor.
However, a relationship between poorer study quality and newspaper coverage approached significance. High-
impact journals were not found to publish studies with higher quality. Poorer quality studies, which tended to
highlight negative findings, also received more citations in scholarly sources. Our findings suggest that negative
effects of violent video games exposure in children and adolescents, rather than large effect size or high
methodological quality, increase the likelihood of a study being cited in other academic publications and
subsequently receiving news media coverage.

Keywords: video games, news media, publication bias, child development

Introduction specific predictors of the news media coverage of published

violent video game effects studies.
W hether violent video games cause aggression and
the extent to which this is so has been a sore spot of
contention in the academic literature for decades.1 Despite
Research on effects of violent video games
extensive research, there is no consensus among members of To better understand newer research on the social and
the academic community on the reality of the relationship political influences on the public’s perception of the link
between violent video games and aggression.2 The evidence between violent video games and aggression, it is worth
consists of a conflicting series of articles, some suggesting summarizing the academic debate. Among various potential
violent video games are likely to cause aggression3 while effects of exposure to violent video games, aggression has
others refute such claims.4 With such a conflicting evidence been the most studied.9 Both, some individual studies10 and
base, it is worth exploring avenues by which existing research meta-analyses,3 have linked violent games to aggression.
may inform public opinion. Some academics have made unsupported claims in linking
Researchers have more recently started to examine the violent video game play and criminal behavior. For example,
various social and political processes which could contrib- when commenting on how serial killers’ criminal urges
ute to the general public’s understanding and interpreta- influenced their future behavior, Hickey11 stated ‘‘using
tion of the potential link between violent video games and alcohol, pornography, or other such types of graphic litera-
aggression (as well as crime). Findings suggest that public’s ture may be useful in expediting the offender’s urge to
understanding is based on a variety of factors, including kill’’(p113) and that video game research has ‘‘yielded some
an individual’s subjective interpretation and own biases additional insights into aggressive behavior’’.(p133) He then
concerning the topic,5 ideological positions of professional cited several studies on the relationship between violent vi-
organizations,6 opportunistic actions of politicians,7 and deo games and aggression, none of which were conducted on
epistemological problems inherent to the field.8 This article samples of serial killers.
aims to add to this newly scrutinized area of video game A growing body of evidence contradicts these claims
research by examining several important contributors to, and and points toward important methodological flaws in these

1
Division of Social Science & Interdisciplinary Studies, Lindsey Wilson College, Columbia, Kentucky.
2
University of Exeter Medical School, Child Health Unit, Exeter, United Kingdom.
3
Department of Psychology, Stetson University, DeLand, Florida.

735
736 COPENHAVER ET AL.

studies. Ward12 noted that the association between violent literature through the interpretive sociological notion of so-
video game play and fighting in adolescents was only modest cial constructionism.25 According to this theory, human be-
and not statistically significant once additional demographic ings create their social realities through processes of social
variables were controlled. Ferguson and Kilburn13 pointed relationships and communication; this created reality often
out that the abovementioned findings by Anderson et al.3 diverges from objective reality.14 Potter and Kappeler26
relied heavily on bivariate correlations instead of more highlighted many instances where the news media were used
sophisticated multivariate analyses. Anderson et al.’s3 own to construct crime problems and shape the public’s under-
analysis produced considerably weaker effect sizes when standing of the reality of crime. Ferguson6 also pointed to-
controlling for one or two variables. Ferguson4 highlighted ward the consequences of the politicians’ use of research on
several pervasive problems in previous research, including violent video games’ effects that had actually been dis-
the use of mismatched games and failure to pretest the par- credited within the academic community.
ticipants’ levels of aggression in experimental studies, the An additional concern is that of accuracy in today’s
use of nonstandardized aggression measures and extrapo- American journalism. The financial pressures faced by
lating findings based on nonclinically relevant measures to newspaper companies and decreases in newspaper sales27
explain mass shooting incidents. Other problems included contribute to news media attempts to capture readers’ at-
failure to control for ‘‘third variables’’, selective interpreta- tention with intriguing constructions. News agencies are
tion of findings, selection bias in literature reviews, and lack responsible for determining what types of stories exhibit
of generalizability to real-world violence.14 Some studies ‘‘newsworthiness,’’ therefore, journalists would select those
even showed a decrease in real-life violence.15 elements more likely to generate public interest and lead to
Authors have argued that since homicide rates in countries increases in revenue.27 Arguably, it would be more difficult
such as the United States have declined with the rise of vi- to attract public interest in reading articles that did not im-
olent video game play, other societal factors should be mediately lead the reader to believe violent video games
considered.16 Adachi and Willoughby8 found that the com- have dramatic negative influences on players. The publica-
petitive nature rather than the violent content of video game tion bias in the relevant literature4 adds to this by potentially
play predicted aggression. Waddell and Peng17 similarly limiting the amount and quality of scientific evidence to
found that competitive game play was associated with ag- which journalists have access.
gression, whereas gaming that was cooperative by nature
increased cooperation among players. Some players begin The current study
playing violent video games with already higher-than-
average trait aggression.18 Research in potentially ‘‘at-risk’’ This study examined articles published in academic
populations such as individuals with neurodevelopmental or journals that reported on the effects of exposure to violent
psychiatric conditions did not support the hypothesis of video games in children and adolescents. The following
their vulnerability to video game violence effects.19,20 Any hypotheses were tested:
association between violent video game play and players’
aggression thus appears less straightforward than previ- H1: Studies with a large effect size were more likely to
receive newspaper coverage compared with studies with a
ously suggested; single explanations such as desensitization21 small or null effect size.
seem unlikely.
H2: Studies with a large effect size were published in high
impact compared with low-impact journals.
Politics and video games
H3: Studies with a large effect size were published in high-
Violent video games have become a source of political impact journals independently of study quality.
capital for politicians and professional agencies with an in-
terest in taking a stance against violent games.22 In other
words, politicians oppose violent video games because,
Methods
‘‘elected public officials are politicians who may capitalize
on the news to further their political agendas and to gain Included studies
support of voters’’.23(p3) Politicians often make use of sym-
The 68 studies included in the present analysis had been
bolic crimes14 by blending rhetoric and symbolic policy
previously included in a recent meta-analysis of the effects of
initiatives (which are also popular with the public) to ease
video game use on a range of outcome variables in child and
public fear of crime.24 Copenhaver’s7 qualitative document
adolescent samples.4 We selected those studies that specifi-
analysis of bills proposed since the early 1990s found poli-
cally examined exposure to violent video games as the pre-
ticians propose bills designed to regulate or address the
dictor variable, as opposed to general video game use. Initial
perceived problems associated with violent games, for ex-
screening was done by reading the abstract (initially 750
ample, taxation of violent video game sales and warning
hits), with confirmation coming once the entire article was
labels stating violent video games cause aggression; such
scrutinized for methods. One hundred one studies on a
proposals continued even after the Supreme Court’s 2011
variety of video game issues were reduced to the current
ruling in Brown vs. Entertainment Merchants Association.
sample once focused on the issue of violence in games.
Aggression was the outcome variable in the majority of
Media effects on public’s understanding
studies (86.76 percent); other outcome variables were pro-
The news media plays an important role in shaping the social behavior (11.76 percent) and depression (1.47 per-
public’s understanding of, and facilitating political stances cent). Studies were published between 1984 and 2014, with
on crime.23 This role is often understood in the academic the majority (57, 86.82 percent) being published in the 2000s
VIDEO GAME VIOLENCE RESEARCH IN NEWS 737

(c) In addition, lastly, the control of gender, trait aggres-

sion, and prior aggression in correlational/longitudinal
studies.4 Such controls are considered essential to
isolate the potential effects of violent video games
from other, third variables.
All included studies were examined for whether they had
received newspaper coverage using the LexisNexis database,
which includes coverage from over 1,600 newspapers. For
each study, we conducted searches by using the name of first
author, title of the journal, publication year, and the topic
‘‘video games.’’ Studies were coded as Yes/No for whether
they had received newspaper coverage. Data for the study
can be found at: christopherjferguson.com/NewspaperStudy
.xlsx. A full list of studies included is available on request.

Results
To test the first hypothesis, a binomial logistic regres-
FIG. 1. Frequency of violent video game studies pub- sion was employed using newspaper coverage as the de-
lished by year. pendent variable, with study effect size, publication year,
journal impact factor, and best practices as predictor vari-
and only a small number in the 1980s (seven studies) and ables. However, this model proved to be highly unstable
1990s (four studies). Figure 1 presents data on studies pub- given collinearity between effect size and the other predictor
lished by year. variables. As such, a comparison analysis of studies with and
Studies were coded for several variables, including out- without newspaper coverage was conducted using t test with
come, effect size, the presence of citation bias, year pub- effect size as the outcome. Among the studies considered, 18
lished, and the impact factor of the journal in which the study (26.5 percent) had received some newspaper coverage. Re-
was published. Citation bias occurs when authors only cite sults indicated that effect sizes were nearly identical for
prior studies that agree with their hypotheses, failing to in- studies with (M = 0.0879, standard deviation [SD] = 0.107)
form readers of controversies or inconsistencies in the field. and without (M = 0.0773, SD = 0.099) newspaper coverage
Impact factor was not available for five studies as they were [t(66) = 0.380; p = 0.707].
published in books, book chapters, or other similar outlets. Although it no longer included our main predictor vari-
For those studies with multiple outcomes, the effect sizes able (i.e., effect size), we nonetheless ran the binomial lo-
were combined into a single effect size (as per standard gistic regression with the remaining predictor variables,
meta-analytical practice). which did not result in an unstable model. This explor-
Study quality was assessed by using a ‘‘best practices’’ atory analysis resulted in a significant regression model
coding employed in the original meta-analysis. A full ac- (v2 = 17.86, p = 0.001; Nagelkerke R2 = 0.359). Only publi-
counting of this procedure is provided in the original article cation year (B = 0.208, standard error [SE] = 0.103; Wald =
and was not deviated from here. Of the studies included, 18 4.099 p = 0.043) was significantly associated with newspaper
(26.5 percent) met these criteria for best practices. The best coverage: recent studies were more likely to be covered by
practices criteria were designed to consider recent contro- newspapers. There was an inverse although nonsignificant
versies over measurement and internal validity in video relationship between best practices and newspaper coverage
game experiments. Studies had been assessed on the fol- (B = -1.401, SE = 0.751; Wald = 3.479 p = 0.062). As noted
lowing criteria: earlier, 18 studies (26.5 percent) met best practices criteria.
To examine the other two hypotheses, we employed
(a) The use of well-validated, standardized outcome mea- stepwise regression with journal impact factor as the out-
sures. Many studies employ unstandardized, poorly come and effect size, best practices and publication year
validated measures, increasing the potential for Type as predictors. The resultant model was significant [F(1,
I error. Measures, whose use change without expla- 61) = 4.77, p = 0.033; adjR2 = 0.057.] Again, only publica-
nation from one study to another, including within the tion year was significantly associated with journal impact
same research group, and which have not been vali- factor (b = 0.269, p = 0.033). Neither effect size nor best
dated against clinical measures of aggression, would practices were significantly associated with publication in
be examples of poor measures. high-impact journals.
(b) Careful matching of video games and careful control
of game content in experimental studies. Recent an-
Exploratory follow-up analysis
alyses8 have indicated that many experiments intro-
duced confounds and potential false positives by Although not part of our original hypotheses, we also
failing to match video games on variables other than obtained data on the number of PsycINFO citations for each
violence. Matching a highly competitive first-person study. Predictor variables included effect size, publication
shooter game with strong characters and narrative year, journal impact factor, and best practices, with results
with a relaxing puzzle game without these features run in a stepwise model. The resultant model was significant
would be an example of poor matching of conditions. [F(2, 58) = 5.53, p = 0.006; adjR2 = 0.131.] In this model, the
738 COPENHAVER ET AL.

number of citations was negatively associated with publi- The greater attention that poorly designed studies poten-
cation year (b = -0.323, p = 0.010). The number of citations tially receive from both academics and the public is a wor-
was also inversely related to best practices (b = -0.277, rying issue and definitely requires clarification. This is more
p = 0.026). so as previous research has indicated that poorly designed
studies were more likely to show negative effects of violent
video games’ exposure.4 This seems to suggest that findings
Discussion
of a negative effect of violent video games’ exposure, rather
This article aimed to add to the existing literature by ex- than a large effect size or high methodological quality, in-
amining a number of potential contributors to, and specific crease the chances for a study to receive attention in the
predictors of the news media coverage of published studies academic field as well as news media coverage. One could
on the potential effects of violent video games exposure in also infer that it is thus the negative effects that increase
children and adolescents. By controlling for a range of var- the likelihood of a study being published in a high-impact
iables we were able to identify factors predictive of whether journal. One potential explanation could be that negative
news media outlets would inform the public about particular effects feed into pre-existing beliefs among the general
studies published in academic journals. public as well as academics on the harmful effects of violent
Contrary to our hypotheses, study effect size was not a video game playing in children and young people. There is
predictor of either newspaper coverage or publication in also the question of the nature and quality of reporting of
high-impact journals. It is worth noting that all included results in academic articles, with authors sometimes over-
studies were conducted on children and adolescents and emphasizing the scientific or practical importance of weak
could therefore have smaller effect sizes compared with effects.
similar studies in a population of young adults such as Although publication bias on video game violence effects
college students.28,29 literature has previously been investigated, this is the first
Our findings of an association between study publication exploration of factors associated with the news media cov-
year and both newspaper coverage and journal impact factor erage of relevant research in children and adolescents. Our
indicate that research on the effects of violent video games findings are consistent with previous reports, highlighting
exposure in children and young people has gained increasing publication bias as a serious problem in this field.4,29 Our
attention from both the academic field and the news media results are limited by the number of included studies, therefore
coverage in recent years. It is worth noting that the great impacting on study power.
majority of the studies examined in this article were pub- High-impact journals are generally seen as gatekeepers
lished in the 2000s, suggesting that video game violence between the full range of scientific studies and news media.
exposure in children and adolescents became a ‘‘hot’’ topic Articles in high-impact journals may benefit from highly
for researchers at that time. High-impact journals appear to visible press releases, such as from the professional orga-
recently publish more studies in this field, suggesting a po- nizations which publish the journals in question (e.g., the
tential trend. It is therefore worrying that neither effect size American Psychological Association, the Association for
nor the overall methodological quality of such studies seems Psychological Science). Authors have previously pointed out
to predict publication in high-impact journals. These findings that journal impact factor and study quality are poorly as-
contradict the belief that high-impact journals publish better sociated,30,31 suggesting that the journals’ gatekeeping role
research articles in this field, and suggest that other factors is more likely linked to the ‘‘newsworthiness’’ nature of any
play a role in ensuring publication success. This would also scientific evidence. If high-impact journals are more inclined
warrant further exploration in future studies. The inverse to publish research showing negative effects of violent video
relationship between publication year and the number of game exposure, even if such effects are relatively weak and
PsycINFO citations could be explained by the fact that recent mostly come from poor rather than high-quality studies, and
publications were unlikely to have received a high number such research is most visible to journalists, this would further
of citations. distort the perspective of the scientific evidence among the
Although study quality did not predict publication in general public.
high-impact journals, poorly designed studies (those not
meeting best practices criteria) were more likely to be cited Author Disclosure Statement
by other academic publications and potentially receive
more news media coverage. The inverse relationship be- No competing financial interests exist.
tween best practices and newspaper coverage suggests that
newspapers were potentially more likely to cover poorer References
quality studies. We note that the interpretation of results
1. Cooper J, Mackie D. Video games and aggression in children.
with p-values above 0.05 is difficult and Type I error rates Journal of Applied Social Psychology 1986; 16:726–744.
become higher. However, the effect size of this result was 2. Quandt T, Van Looy J, Mäyrä F, et al. Digital games re-
fairly high and the reduced p-value was likely due to study search: a survey study on an emerging field and its preva-
power. Furthermore, we found a similar, inverse rela- lent debates. Journal of Communication 2015; 65:975–996.
tionship between best practices and number of PsycINFO 3. Anderson C, Shibuya A, Saleem M, et al. Violent video
citations that indicated that studies of poorer quality re- game effects on aggression, empathy, and prosocial be-
ceived more citations. Although the result of tentative havior in Eastern and Western countries: a meta-analytic
analysis and thus requiring cautious interpretation, these review. Psychological Bulletin 2010; 136:151–173.
findings would be worth further, more detailed exploration 4. Ferguson C. Do angry birds make for angry children? A
in future studies. meta-analysis of video game influences on children’s and
VIDEO GAME VIOLENCE RESEARCH IN NEWS 739

adolescents’ aggression, mental health, prosocial behavior, 19. Engelhardt C, Mazurek M, Hilgard J, et al. Effects of
and academic performance. Perspectives on Psychological violent-video-game exposure on aggressive behavior,
Science 2015; 10:646–666. aggressive-thought accessibility, and aggressive affect among
5. Ivory J, Kalyanaraman S. Video games make people violent— adults with and without autism spectrum disorder. Psy-
well, maybe not that game: effects of content and person chological Science 2015; 26:1187–1200.
abstraction on perceptions of violent video games’ effects 20. Ferguson C, Olson C. Video game violence use among
and support of censorship. Communication Reports 2009; ‘vulnerable’ populations: the impact of violent games on
22:1–12. delinquency and bullying among children with clinically
6. Ferguson C. Violent video games and the Supreme Court: elevated depression or attention deficit symptoms. Journal
lessons for the scientific community in the wake of Brown of Youth and Adolescence 2014; 43:127–136.
v. Entertainment Merchants Association. American Psy- 21. Brockmyer J. Playing violent video games and desensiti-
chologist 2013; 68:57–74. zation to violence. Child and Adolescent Psychiatric Clin-
7. Copenhaver A. Violent video game legislation as pseudo- ics of North America [serial online] 2015; 24:65–77.
agenda. Criminal Justice Studies: A Critical Journal of 22. O’Holleran J. Blood code: the history and future of video
Crime, Law and Society 2015; 28:170–185. game censorship. Journal on Telecommunications and High
8. Adachi P, Willoughby T. The effect of video game com- Technology Law 2010; 8:571–612.
petition and violence on aggressive behavior: which char- 23. Benekos PJ, Merlo AV. (2014) Crime control: politics and
acteristic has the greatest influence? Psychology of Violence policy. 2nd ed. New York: Routledge.
2011; 1:259–274. 24. Fairchild ES, Webb VJ, eds. (1985) The politics of crime
9. Hollingdale J, Greitemeyer T. The changing face of ag- and criminal justice. Thousand Oaks, CA: Sage.
gression: the effect of personalized avatars in a violent 25. Berger PL, Luckmann T. (1966) The social construction of
video game on levels of aggressive behavior. Journal of reality: a treatise in the sociology of knowledge. Garden
Applied Social Psychology 2013; 43:1862–1868. City, NY: Anchor Books.
10. Konijn E, Nije Bijvank M, Bushman B. I wish I were a 26. Potter GW, Kappeler VE, eds. (2006) Constructing crime:
warrior: the role of wishful identification in the effects of perspectives on making news and social problems. 2nd ed.
violent video games on aggression in adolescent boys. Long Grove, IL: Waveland Press.
Developmental Psychology 2007; 43:1038–1044. 27. Campbell R, Martin CR, Fabos B. (2013) Media and cul-
11. Hickey EW. (2010) Serial murderers and their victims. ture: an introduction to mass communication. 8th ed.
Belmont, CA: Wadsworth. Boston, MA: Bedford/St. Martin’s.
12. Ward MR. Video games and adolescent fighting. The 28. Sherry J. The effects of violent video games on aggression:
Journal of Law and Economics 2010; 53:611–628. a meta-analysis. Human Communication Research 2001;
13. Ferguson C, Kilburn J. Much ado about nothing: the mises- 27:409–431.
timation and overinterpretation of violent video game effects 29. Ferguson C. Evidence for publication bias in video game
in Eastern and Western nations: comment on Anderson et al. violence effects literature: a meta-analytic review. Ag-
(2010). Psychological Bulletin 2010; 136:174–178. gression and Violent Behavior 2007; 12:470–482.
14. Surette R, Maze A. Video game play and copycat crime: an 30. Brembs B, Button K, Munafò M. Deep impact: unintended
exploratory analysis of an inmate population. Psychology consequences of journal rank. Frontiers in Human Neu-
of Popular Media Culture 2015; 4:360–374. roscience 2013; 7:291
15. Markey P, Markey C, French J. Violent video games and 31. Tressoldi PE, Giofré D, Sella F, et al. High impact = high
real-world violence: rhetoric versus data. Psychology of statistical standards? Not necessarily so. PLoS One 2013;
Popular Media Culture 2015; 4:277–295. 8:e56180.
16. Ward MR. Video games and crime. Contemporary Eco-
nomic Policy 2011; 29:261–273.
17. Waddell J, Peng W. Does it matter with whom you slay? Address correspondence to:
The effects of competition, cooperation and relationship Dr. Christopher J. Ferguson
type among video game players. Computers in Human Department of Psychology
Behavior 2014; 38:331–338. Stetson University
18. Przybylski A, Deci E, Rigby C, et al. Competence- 421N. Woodland Boulevard
impeding electronic games and players’ aggressive feel- DeLand, FL 32729
ings, thoughts, and behaviors. Journal of Personality and
Social Psychology 2014; 106:441–457. E-mail: [email protected]
Copyright of CyberPsychology, Behavior & Social Networking is the property of Mary Ann
Liebert, Inc. and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.