The Effect of Bias in Video Game Violence Research: Spurious Conclusions on Physiological Desensitization to Real-Life Violence by Caranagey et al.

Abstract 

This is a review article for “The effect of video game violence on physiological desensitization to real-life violence” (2006).  In their article, Carnagey, Anderson, and Bushman pursue the negative effects of violent video games in an ostensibly unprecedented examination of the physiologically desensitizing effects manifested by violent video game exposure, specifically to real-life violence.  Following a literary review of research supporting the claim that violent media has negative effects to those exposed, the article then shifts to a brief synopsis of GAM and its relevance to the study.  The experiment itself appears to have been skillfully conducted, however the conclusions reached are by no means justified by the yielded results that are themselves quite misleading.

The Effect of Bias in Video Game Violence Research

The ongoing public debate on the consequences of exposure, both limited and extended, to violent video games has been known to extend to the academic realm for some time now.  The “abundance of research” demonstrating the harmful effects of violent media, combined with violence being a predominant aspect of video games, surpassing 85% of the market, are a few of the reasons Carnagey et al. cite for researching further.  Though the connection of desensitization and violent media have been explored before, they point out that no previous works prior to their study has specifically addressed whether or not the desensitization extends to real-life violence.

Aptly and immediately, the paper examines the term, “desensitization”.  The plurality of definitions used by crowds ranging from academia to the senate are noted, listed, and then dismissed for a working definition to be used for the paper’s purposes: “a reduction in emotion-related physiological reactivity to real violence” (p. 490).  This is followed by examples in which systematic desensitization was utilized to reduce frenetic responses to items ranging from spiders to combat.  The effectiveness of systematic desensitization is then compared to the effectiveness of unintentional desensitization, in which the adaptation of surgeons to horrific stimuli, often present when operating, is used as an illustration.  Having legitimized the efficacy of both intentional and unintentional modes of desensitization, the paper then issues an admonition against the negative effects desensitization can produce.

A literary review, consisting of research having already documented such negative effects is then presented.  All studies reviewed reached similar conclusions, mainly that exposure to violent media decreases participants’ arousal response to subsequent horror clips or stages of real-life violence.  These studies, Carnagey et al. point out, though important and revealing, lack both the focus and criteria that the current paper purports to possess; the focus being “real-life” violence, not of the theatrical variety, and the criteria as follows: 

(1) Random assignment to violent or nonviolent media exposure groups; (2) use of violent and nonviolent entertainment media that are equivalent (or statistically controlled) on various nonviolent aspects… (3) use of emotion-related physiological indicators as the dependent variable (e.g., heart rate, GSR); and (4) use of real violence as the emotion-provoking stimulus in the dependent variable assessment… (p. 490)

While the paper notes that previous studies may contain one or most of these, it insists that no previous work has made use of all of them, and that in order to establish causality, all four are requisite. Indeed, the authors do seem to have implemented these four into their research methodology, while extending the General Aggression Model (GAM) as their framework.  Condition 1 is reported to have been followed, with 257 college students—124 men and 133 women—serving as randomly assigned participants.  Condition 2 was met in their selection of games (4 violent, 4 nonviolent).  Heart rate and GSR monitoring were also used, satiating the third condition.  And the fourth condition was approached by videotapes of “real violence,” in the form of outbursts, police confrontations, shootings, and prison altercations.  Fig. 1 and the textual contents of page 491 explain the GAM’s application to the study: through the use of rewarding emotional stimuli (e.g., ‘epic’ background music, rewards for killing, etc.), for violent behavior, GAM theory suggests that a gradual reduction of fear will occur, causing a desensitization to hostile action, resulting in a retardation of appropriate response to “real-life” violence.  Using the GAM as their guide, the authors’ hypothesis was as follows: “We predicted that violent game players would show less physiological arousal to real-life violence than would nonviolent game players.

Their procedure for testing their hypothesis is as follows: participants, tested individually, were first measured for base HR and GSR.  During this 5 min procedure, the subjects were asked to respond to a series of questions, some pertaining to video game preferences, and others meant to gauge trait aggression.  The electrodes were then removed from the participants, and what followed was 20 min of either violent or nonviolent gameplay.  Following gameplay, the HR/GSR measurements were taken again for 5 minutes.  The participants then watched a 10 min recording of real violence whilst HR and GSR were concurrently monitored.  No following measurements were taken.

The yielded results presented largely in a 2x2x3 ANCOVA for both HR and GSR, with adjusted averages, means and figures possessing a slightly more diminished presence, are presented as mostly supportive of the authors’ hypothesis.  Significant HR contrasts between violent and nonviolent groups do not occur until film viewing.  And, as expected, HR for nonviolent gamers increased, F(1, 131)=16.60, p<.05, d =.72, while Violent gamers HR displayed no change, F(1, 116)=.41, p > .05, d=.11, with the mean heart rates for violent and nonviolent HR being 68.5 and 70.7 respectively (p. 493).  GSR measurements revealed significant differences in gender readings.  Women were higher at baseline, post gameplay, and during film viewing, however readers are not informed of gender inequality from group to group.  Like that of HR, significant GSR contrasts did not result until during viewing, with the nonviolent group once again expressing higher arousal than the violent group: F(1, 244)=4.67, p<.05, d=0.28, with mean GSR scores at 242.9 and 286.8 for violent and nonviolent groups respectively (p. 493).

Upon stating the significance of the results—“that playing a violent video game, even for just 20 min, can cause people to become less physiologically aroused by real violence,”—the discussion moves to its crux, GAM.  The connection of the paper’s findings to GAM is made by suggesting that the findings can explain aggressive and lack of prosocial behavior through media-induced desensitization to violence.  The infamous witnesses of the Kitty Genovese murder are used as model individuals, lacking the impetus to assist.  Studies indirectly supporting GAM’s link between desensitization and ‘bad Samaritans’ are then cited, followed by a recapitulation.  Future research is then encouraged to answer questions such as, what the long-term effects of desensitization are, and whether or not the effects of violent video games considerably worse than other violent media.  The study-relevant conclusions drawn that were essentially a recapitulation of the introduction to the discussion: that exposure to violent video games cause desensitization to real-life violence.

Criticism of this study could, frankly, go on for some time.  Opening remarks of the paper even prove problematic.  In its third paragraph, the paper states, “Youth exposed to violent media…become more aggressive adults” (p. 489).  It is difficult to comprehend what the necessary conditions would need to be to solve for the multitudinous variables in such a long-range study.  As Dr. Michael Ward, of University of Texas, Arlington remarked, concerning measuring long-term effects of violent video game exposure, “I don’t know that a psychological study can ever answer that question definitively” (Carey, “Shooting,” 2013).  Indeed, how could causation possibly be established?  Unsurprisingly, findings of this kind, this particular one being a work by Anderson et al.—Anderson being one of the authors of the study being reviewed here—have been critiqued and repudiated by others (criticisms of any such aggressive link include Bryce & Kaye (2011) Ferguson & Kilburn (2010); Ferguson & Rueda (2010); Jenkins (2006); Kutner & Olson (2008); Pinker (2002); Tedeschi & Quigley (1996).).  Positive correlation (though not causation) between increased sales of violent video games and a decrease in violent crime has been established (Ward, “Understanding,” 2011), and some scholars even claim that children who do not play video games are at the most risk for aggressive behavior (Kutner & Olson, 2008).   The absence of findings contrary to the effect of the study under review go unannounced within it, an unfortunately less than odd phenomenon in academic debates.

The “abundance of research demonstrating negative effects of violent media exposure,” as mentioned earlier, is one reason Carnagey et al cite for further discussion, but virtually no mention of contrary studies (such as Copper & Mackie (1986); Graybill, Kirsch, & Esselman  (1985); Graybill, Strawniak, & O’Leary (1987); Scott (1995); Winkel, Novak, & Hopson (1987)) are included in any portion of the article.  Research analysis by Dr. of Psychology, Jonathan Freedman (2002), revealing that approximately half of available research has found no correlation between violent media and subsequent aggressive behavior, makes this an even more striking omission.  Not only is this misleading to readers who may be unfamiliar with the debate within academic circles, it also not an exception to the authors’ works.  Both Anderson and Bushman have been criticized in the Supreme Court and elsewhere for failing to cite findings antithetical to their own (ESA, VSDA, and IRMA v. Blagojevich, Madigan, and Devine, (2005); Ferguson (2013); Ferguson & Rueda (2010); Carnagey, however, seems to be an exception.  It goes without saying that publication bias is an important to note, but the issue with this paper extends beyond misrepresentation and shortchanging conflicting findings because such findings are not even acknowledged, leaving the appearance of a consensus. 

The presentation of results is equally troubling.  While it is understandable for ANCOVA findings to take priority over mean HR and GSR scores, there seems no reason that the mean HR and GSR’s be tucked away in footnotes.  That is, unless the results are not as “significant” as the authors claim them to be.  One could argue that the measurements have an unavoidable presence in Figures 2 and 3 (p. 493), but that argument would be based on a thorough observation of the charts. Figures 2 and 3 do indeed present bar graphs of the mean HR and GSR measurements respectively, but neither vertical axis is grounded at zero—a pedestrian, age-old deception, most commonly seen in the public sphere, certainly not befitting the annals of academia.  The HR chart, for example, grounded at 60, capped at 75, is well within the range for normal resting HR.  Mean HR for violent and nonviolent respectively, are footnoted as 68.5 and 70.7; a sizable difference on a graph spanning 15 BPM, to be sure, but it boggles the mind to ascertain by what standards a mean difference of two BPM is considered significant.

Discussion and conclusion of the results withdraw even further from the realm of credibility.  The results do not demonstrate that violent video game exposure causes people “to become less physiologically aroused by real violence,” yet the authors purport the results do so.  Aside from the deceptive display of meaningful results, there is also a methodological facet contesting the alleged demonstration: participants were not exposed to real-life violence.  However horrific or unsettling recordings of real-life violence may be, the differences in responses to recorded and actual real-life violence would unquestionably differ dramatically.  It is ergo not surprising that the paper attempts to relate findings to cases of real-life violence (e.g., the Kitty Genovese murder).  What is surprising is its approach.  Upon addressing the failure of the “several” witnesses to assist Genovese, the paper immediately, almost slyly, shifts to linking GAM to bystanders in general.  It almost as if the authors are attempting to infer a possible replacement of the bystander effect with desensitization, though indirectly, and with enough distance to avoid accusations of the frivolous attempt.  The paper temporarily redeems itself by skillfully linking the results of other studies with GAM and its own findings, but fault is inevitably found in the conclusion when the paper’s demonstrations and implications are reiterated.  

There exist both positive and negative elements in the Carnagey et al (2006) study, unfortunately the cons envelop the pros.  While it is good to see a disambiguation of ‘desensitization,’ as the failure to define the increasingly enigmatic term has been a source of criticism by those that contest any such product of violent media exists.  It is also pleasing to see the continuation of a somewhat healthy debate, and hopefully, the future will yield a decrease in bias on both sides of the fence.  The articles less admirable qualities include publication bias, study methods (real-life violence was not measured as it is not feasible, nevertheless the title and article content portray recorded violence as “real-life” violence), specious presentation of results, and false implications of so called, “significant” findings.

References

Carey, B. (2013, February 11). Shooting in the Dark. The New York Times, pp. 1-2. Retrieved from http://www.nytimes.com/2013/02/12/science/studying-the-effects-of-playing-violent-video-games.html?pagewanted=1&_r=2&adxnnlx=1372307110-ilsrIQSdXJ9zjrecuDkGVQ

Carnegey, N., Anderson, C., & Bushman, B. (2007). The effect of video game violence on physiological desensitization to real-life violence. Journal of Experimental Social Psychology, 43(3), 489-496. Retrieved from http://dx.doi.org/10.1016/j.jesp.2006.05.003

Cooper, J., & Mackie, D. (1986). Video Games and Aggression in Children. Journal of Applied Social Psychology, 16(8), 726-744. Retrieved from http://dx.doi.org/10.1111/j.1559-1816.1986.tb01755.x

Fanti, K. A., Vanman, E., Henrich, C. C., & Avraamides, M. N. (2009). Desensitization to Media Violence Over A Short Period of Time. Aggressive Behavior, 35(2), 179-187. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/19172659

Ferguson, C. (2013). Violent Video Games and the Supreme Court: Lessons for the Scientific Community in the Wake of Brown v. Entertainment Merchants Association. American Psychological Association, 68(2), 57-74.

Ferguson, C. J., & Rueda, S. M. (2010). The Hitman Study. European Psychologist, 15(2), 99-108. Retrieved from http://www.tamiu.edu/~cferguson/hitman.pdf

Findings suggest that playing violent video games may reduce stress and depression, however, "results are correlational," and additional research is recommended.

Freedman, J. L. (2002). Media violence and its effect on aggression: assessing the scientific evidence. Toronto: University of Toronto Press.

Jenkins, H. (n.d.). Reality Bytes: Eight Myths About Video Games Debunked. Impact of Gaming Essays - PBS. Retrieved from http://www.pbs.org/kcts/videogamerevolution/impact/myths.html

Kutner, L., & Olson, C. K. (2008). Grand theft childhood: the surprising truth about violent video games and what parents can do. New York: Simon & Schuster.

Ramos, R., Ferguson, C., Frailing, K, & Ramirez, M. (2013). Comfortably Numb or Just Yet Another Movie? Media Violence Exposure Does Not Reduce Viewer Empathy for Victims of Real Violence Among Primarily Hispanic Viewers. Psychology of Popular Media Culture, 2(1), 2-10. Retrieved from http://www.tamiu.edu/~cferguson/desensitization.pdf

Sherry, J. (2001). The Effects of Violent Video Games on Aggression. A Meta-analysis. Human Communication Research, 27(3), 409-431. Retrieved from http://www.icagames.comm.msu.edu/vgma.pdf

Ward, M. (2011, April 21). Understanding the Effects of Violent Video Games on Violent Crime by Scott Cunningham, Benjamin Engelstatter, Michael R. Ward  SSRN. Social Science Research Network. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1804959