An analysis of the research used, and argument for, the view that violent video games result in an increased level of anti-social behaviour.
Created for PSYA4 A2-Level Psychology homework.
2. Facts, Theories & Ideas
In an age where more kids stay inside than go out, where 38% of American households own a video-games console [ESA,
2008] and where 95% of teenagers play video-games [PI&ALP, 2008] it is hardly surprising that people have began to
seriously question the possible negative effects that violent video-games are having on us, and on children in particular.
Initially statistics seem to suggest a strong correlational link between video-games and anti-social behaviour in children,
such as increased aggression, increased peer confrontation and decreased academic achievement [Gentile et al, 2004],
and there are also undeniable physiological effects such as an increased heart rate and blood pressure [Tafalla, 2007].
From these facts, many explanations have be offered to suggest a direct link between the two.
Zillman’s excitation transfer theory [1979] argues that the arousal produced by engaging in violent media may be transferred
to real-life situations involving conflict, due to either a misinterpretation of the arousal’s origin or an uncontrollably high
arousal level. [Berkowitz, 1993].
Exposure to violent video-games has also been shown to have a desensitizing effect on brain activity; when viewing violent
images, habitual game players had a reduced response in the P300 component of the brain, which is an ERP used in
decision making, compared to the responses of non-regular game players.
However, despite the biological evidence there is a lack of non-subjective and methodologically sound research into the
psychological effects of violent video-games, due mostly to the difficulty in effectively measuring and operationalising both
aggression within video games and long-term changes within individuals as a result of video-games.
The General Aggression Model [Anderson et al] is currently the most complete and influential model used for suggesting a
connection between violent video games and increased anti-social behaviour. However it makes note that the occurrence
and scale of any possible negative effects depends first and foremost on the individual themselves. It argues that input
variables, such as personality and gender, and situational factors, such as provocation, influence or mediate an individual’s
reaction to playing violent video-games. From this, exposure is said to increase aggression in three possible ways;
increased arousal, changes in cognitions and affect (emotional) changes.
There are also theories and evidence which suggest the opposite effect. For example the catharsis theory disputes the
claim that violent video game content encourages aggression. This theory suggests that the emotional drive evoked by
violent video game play reduces the chance of a child actually exhibiting violent behaviour; the child’s fantasy play and
imagined actions causes the child to have reduced urges to act out aggression in actual behaviour [Ivory, 2001].
3. Research
Anderson & Dill (2000) - Examined video games and their Guimetti & Markey (2007) – Examined violent video games and
relation to aggressive thoughts, feelings and behaviour. anger as predictors of aggression.
•210 psychology undergraduates had their gender and irritability level •167 undergraduates were given two questionnaires to complete – one to
measured and were then asked to play one of two games, selected via a record demographics and one to assess their dispositional anger trait,
pilot study, that differed only on amount of violence (Wolfenstein 3D & using the AQ. [Buss & Perry, 1992]
Myst) for three 15 minute sessions over the period of a number of weeks. •Participants were then randomly assigned one of 6 games, 3 violent & 3
•After each session one of 3 aspects were measured; aggressive thought, non-violent to play for 15 minutes.Afterwards they were given 3 story
affect or behaviour eg. Aggressive thought was measured via response stems, such as ‘The Car Accident’, with negative and ambiguous endings
times to different words, some violent, some not, and aggressive to read, and asked to write down 20 unique things the main character
behaviour was measured via a ‘competitive reaction time’ task, where may feel, think or do for each one, making a total of 60 responses, which
participants were allowed to punish their opponents with a noise blast of the researchers coded as ‘aggressive’ or ‘non-aggressive’.
self-decided duration and intensity. •Results showed that participants who played a violent video game gave
•Whilst there was some variation depending on gender, results showed more aggressive response to the story stems than those who played a
that violent video games seemed to increase aggressive behaviour and non-violent video game but also that those who rated highly for anger
accessibility of aggressive thought. However state hostility was only gave more aggressive responses after a violent video game than after a
influenced by trait hostility and gender – the video games had no non-violent video-game whereas those with low anger remained
discernable effect. relatively unaffected.
Anderson & Bushman (2009) – Examined the effect of violent Freedman (2002) – A meta-analysis evaluation of the research
computer games on later response to real life violence. on violent video games.
•320 college students of equal gender distribution, were individually asked •Partially a response to the Anderson & Bushman (2001) meta-analysis,
to play a randomly assigned video game from a selection of violent and which only contained 9 studies that directly dealt with aggressive
non-violent games, under the guise of a video-game/person-type test with behaviour as a result of video games.
the reward of extra course credit. •Pointed out a plethora of research shortcomings such as:
•After playing their game for 20 minutes, participants were given a lengthy •Few studies had come close to and none at all had solved the problem
questionnaire. 3 minutes into the questionnaire the researcher began of making violent and non-violent video-games comparable on all
playing a 6 minute audio recording of a staged fight outside of the room, variables other than the level of violence, meaning that the cause of any
altered based on the participants gender, and designed for realism using a resultant aggression is open to interpretation.
pilot study. •Those studying the impact of video games have not dealt with demand
•The ‘fight’ ended with the aggressor ‘exiting’ the building and the ‘victim’ characteristics effectively leaving almost all of the results open to the
groaning in pain. The researcher recorded how long it took for the alternative interpretation that they are caused by demand factors rather
participant to respond, if at all, for up to 3 minutes. than the variable of interest.
•Participants who played a violent video game took significantly longer to •Almost all of the measures used by research involved questionable
respond compared to non-violent game players, were more likely to rate analogues of aggression rather than the real thing.
the fight as less severe and more likely to report not hearing the fight. •Concluded that there was no evidence that violent video games cause
any long-term or lasting changes in aggression.
4. Initial Thoughts
What is interesting about the studies examined here is their similarity in most aspects except choice of
measure for aggression, with each one taking a slightly different route. Whilst this helps to get a more
thorough picture of the different effects of video games it does make direct comparison between results a
little more difficult.
Another problem that arises with these studies, and most likely with all studies attempting to ascertain a
connection between anti-social behaviour and violent video games, is the ethical issues involved. A study
that attempts to record changes in anti-social behaviour by misleading participants into undergoing these
changes has a wide-range of implications. Not only are short-term changes recorded but also there are
other possible long-term changes that may occur, and participants may suffer further psychological harm,
for example, feeling a lack of self-control if told that their behaviour is not completely their own. With this
knowledge the studies must be very careful in minimising any possible damage and offer a full debriefing
once over.
Essentially it seems as though conclusions drawn from, or even conclusions about, the research
performed on this topic must be done so cautiously and critically.
5. Positives
Both the Anderson & Dill (2000) and Anderson & Bushman (2009) studies gain significant increases in validity and
reliability due to their use of pilot studies. The implications of these pilot studies is important to consider and was well
noted by the researchers. In the former the pilot study attempted to minimize the possible effects of variables, other than
level of violence, within the games, meaning that the results can be more reliably attributed to the area of study.
In the latter the pilot study significantly increased the validity as it ensured that the recording used was believable,
meaning that participants were unlikely to have acted with the knowledge that the ‘fight’ was not real.
Anderson & Bushman (2009) and Guimetti & Markey (2005) also gain validity due to their use of random allocation and
greater range of games. This helps to ensure that participants remain unaware of the nature of the study, for example
deducing that being asked to play a specific violent video game was attempting to elicit a particular response, thus
reducing the effect of demand characteristics.
Whilst the Freedman (2002) meta-analysis is very different from other studies being examined, this is by no means a
boon. In particular it gains validity by taking into account every piece of research done in the English language on the
subject of video-games and their possible negative effects, allowing a complete and detailed evaluation. Furthermore, this
removed perspective on the research allowed a critical and, more importantly, objective view of the methodology and
conclusions provided. This enables it to effectively note the key weakness surrounding much of the research, such as
lack of evidence and interpretive conclusions.
Obviously the laboratory settings in the 3 studies improved reliability due to their replicability, highly controlled nature and
reliability of raw results. However the Guimetti & Markey (2005) study dealt with this particularly well as measures of
aggression within the participants statements were collected and compared by 3 undergraduate research assistants, and
so contained inter-rater reliability, and were also close enough to be able to average the scores out, increasing precision.
Some of the studies also have the possibility of application to real life. Using the conclusions of Anderson & Dill (2000)
recommendations could be made for individuals to partake in a non-violent game or activity after playing a violent video-
game in order to combat the short-term effects of any subsequent aggressive behaviour changes. The Guimetti &
Markey (2005) study, on the other hand, offers the advice that parents simply ‘know’ their child, specifically their
temperament, and apply this knowledge when allowing their child to play video-games, as an aggressive child is likely to
experience a more exaggerated negative outcome in response to a violent video-game.
6. Negatives
There are numerous, clear shortcomings that the studies by Anderson & Dill (2000), Anderson & Bushman (2009)
and Guimetti & Markey (2005) share. These include; lack of generalisability, due to the small number of participants all
of whom had similar backgrounds; lack of realism, due to the laboratory settings, although this can be argued to have
minimal effect due to the fact that most games are played in isolation; and lack of conclusive evidence, due to the short-
term nature of the studies. However each one also possesses their own specific issues.
Anderson & Dill (2000):
Although the games were chosen due to their similarity in all variables except level of violence it is clear that Wolfenstein 3D and Myst are two
very different games, with different gameplay mechanics and goals. This means that the results are open to interpretation. The decision to use
only two games also increases the risk of demand characteristics becoming involved. The weak cover-story of a ‘learning in developing skills’
activity exacerbates this issue as it is likely participants would question the nature of the study and desired responses. Furthermore the measures
used, and conclusions drawn from such, are questionable as they are merely analogues of aggression, not aggression itself. For example the use
of reaction times to a variety of words and the conclusion that ‘violent video-games prime aggressive thoughts’ does not demonstrate actual
aggression or intent to act more aggressively.
Anderson & Bushman (2009):
One issue involved in this study is that it drew from the same participant pool and same researchers as used in a published study two years prior
[Carnegy et al, 2007] which investigated the physiological desensitizing effects of violent video-games. As this informational was available,
particularly for that participant pool, this may have effected the validity of the student’s responses. Also, whilst using a selection of 6 games is
good in terms of producing more generalisable results, the games used were vastly different, ranging from fighting to the death (Mortal Kombat) to
pinball (3D Pinball). This means a large lack of control over the variables within the games and a subjective conclusion based around the
assumption that it was the difference in violence levels that was the key factor in influencing participant’s responses.
Guimetti & Markey (2005):
In the Guimetti & Markey (2005) study the participant’s responses were coded as aggressive or non-aggressive. However, using this information
as a measure of aggression as a result of playing the violent video-game is incredibly weak, for the same reasons and perhaps to a greater extent
than in the Anderson & Dill (2000) study, as that study used three different analogues.
It is also worth noting that Freedman’s (2002) evaluation does not escape scrutiny, at least when used as evidence of
an absence of link between long-term changes in aggression and video-games. It seems only logical that when played
daily, as many games are, that short-term negative effects essentially become long-term due to the addictive nature of
video-games.
7. Conclusions
It seems evident that from a critical viewpoint, demonstrated out by Freedman (2002), that evidence for violent video-
games resulting in increased anti-social behaviour is inconclusive at best, due to the various methodological issues.
However it is important to recognise that even with these issues, nearly all studies have demonstrated a positive link or
correlation between the two, whether that be an equal and consistent effect for all participants, such as in Anderson &
Dill (2000) and Anderson & Bushman (2009) or one that depends mostly on each individuals personal traits, as in
Guimetti & Markey (2005). This means that, although as of this moment one can argue there is no conclusive evidence
that video-games have a negative effect on behaviour, further research is clearly needed to fully investigate whether
those links have any significant validity.
Ideally, future research would have to attempt to solve the issue of variables within video-game comparisons.
Unfortunately, with out purposely building two or more video games with carefully controlled variables and then going to
the lengths of marketing these video-games so that they retain ecological validity, this seems unlikely to ever be truly
solved.
Furthermore the measurement of aggression needs to be either refined or expanded; perhaps Anderson & Dill (2000)
were on the right track by using 3 analogues of aggression rather than 1. Fully measuring aggression with over-stepping
the boundaries of ethical limits also seems unlikely but a wider range of measures would provide a more complete
picture of possible effects.
Whilst it seems logical to also suggest attempting research into the long-term effects of violent video-games these ethical
limits also prevent that from effectively achieved. The possibility of a long-term negative alteration of someone's life in an
experiment is definitely not looked well upon, but performing an long-term observational study of children’s behaviour in
conjunction with video-games provides only correlation, not causation, leaving the long-term effects still open for debate.
Finally, as Freedman (2002) summarises, it may well be that further research will indicate that playing violent video
games is harmful. For the moment, however, there is no such work and no scientific reason to believe that violent video
games have bad effects on children or on adults, and certainly none to indicate that such games constitute a public
health risk, as was suggested by Anderson & Bushman (2001).