Notes on: Burgess, M., Dill, K., Stermer, S., Burgess, S., Brown, B. (2011) 'Playing With Prejudice: The Prevalence and Consequences of Racial Stereotypes in Video Games'.  Media Psychology, 14: 289-311.  Doi: 10.1080/15213269.2011.596467.

Dave Harris

[A content analysis of best-selling video game magazines and of video game covers, then clips.  Both show racial stereotypes, with minority males more likely to be seen as athletes or as aggressive, and less likely to be seen in military combat or using technology.  Dangerous minority males also appeared.  Overall, minority males disproportionately used extreme guns, or were thugs and athletes.  In an experiment, players were exposed to both violent and nonviolent games with both black and white characters: participants were faster at recognizing violent stimuli {classic psychology test} with games with black characters and vice versa 'indicating the images of popular video game characters evoke racial stereotypes'{very controversial} (290)].

Video games are highly popular and sell very well.  The biggest group playing them are children between eight and 17 years old, with 90% being regular players.  There is lots of literature on effects.  And some 'recent brain research' showing different qualities of empathic responding following different game exposure.  Analysis of the characters has tended to focus on the portrayal of women, although there are some early investigations of race.  The latter showed infrequent appearance and stereotyped depictions when present.  Some work showed that minorities such as Latino women and native American men were never present, while black and Latino men rarely appeared as main characters.  The same studies found stereotyped characterization, with Latino characters are only found in sports, Asian characters largely in wrestling or fighting, black characters as immune to violence and suffering.  It was common even before 9/11 to see the targets of violence as Middle Eastern.  Other studies focused on aggression and noticed racial and gender dimension - although the majority of aggressive characters were white, the next largest groups were Asian or Pacific islander: they are also the most likely to be engaging in extreme violence.

The social effect of media images has also been studied for a long time.  One study by Dill and Brown exposed participants to stereotypical sexed types from video games, then asked participants to react to a real life account of sexual harassment: those exposed to video games images most were more tolerant of sexual harassment.  In a study on racial stereotypes by Dill and Burgess, video game images fitting aggressive criminal or dangerous stereotypes, and other black men including black leaders, were featured alongside white images, then, in  'a purportedly unrelated study' (291), participants were asked to evaluate the website of a political candidate depicted as either black or white.  Results showed interactive effects on  capability ratings of the candidate according to whether they were black or white, with those exposed to negative black video games rating a black candidate as less favourable, while those exposed to more positive images rated the black candidate as more favourable. This is sufficient 'to appreciate the damage inherent in repeated negative and stereotyped portrayals of minorities and women', but more specific research is needed on particular games such as Grand Theft Auto, in one version of which participants were invited to kill Haitians or Cubans.

Stereotypes have cognitive and affective components.  The frequent portrayals of groups in the media show that these become 'the relevant schema for processing members of that particular group' (292): 'the schema of the violent Black man teaches that it is appropriate to experience apprehension when approached by a Black male'.  Further exposure will produce additional thoughts and evaluations, 'ultimately predicting discriminatory behaviour'.  Stereotypes can also provoke unconscious impulsive reactions, which is important in hostility and aggression.  The key factors seem to be links to a negative experience, together with 'situational features associated with aggression'.

African-American males are commonly stereotyped as aggressive, hostile and criminal [citing a study by Berkowitz].  Those associated with aggression are also most likely to be the victims of aggression.  There need not be a conscious cognitive appraisal, rather aggression through 'an automatic impulsive route'.  People need not be conscious of their exposure to negative images and words.  Mass media images also associate African-American men with negative stereotypes, and this 'conditions viewers'.  The images in video games are not real but can still produce negative reactions to real black men [noted by an early 1987 study, which looks rather abstract, showing that stereotypes have elements in the imagination only - 'the failure of reality monitoring' (293).] Video images can confirm stereotypes.

It is to be expected that video games will portray violence and aggression, and the issue is whether this is differentiated by race.  There is a special interest in 'socially sanctioned' violence.  The first study involved imagery from video game magazines, analyzed according to variables such as 'race... hyper masculinity... aggression... war/military aggression...fighting... athletics.... and use of technology'.  The six top selling magazines were chosen, and the largest male and female images selected, producing a sample of 482, which were coded by one white male and one white female research assistants.

A number of codes were given to the races, including 'other, and humanoid' (294), with the latter also appearing to be of a non white race.  'Others' coded native American, Egyptian, and undeterminable.  Hyper masculinity meant 'exaggerated male characteristics such as unnaturally large muscles or expressions of dominance'.  Aggression was defined as 'being engaged in behaviour intended to harm another living being' [including verbal behaviour?].  They assume that war or military aggression were socially sanctioned, but not fighting, since military aggression is legal - the issue also depended on researchers judging whether there was or was not an 'identifiable military rationale'. Frequency data were calculated for each race and submitted to a Chi square test.  Some counts were too low for this procedure, so a collective category of minorities was assembled to be compared to white data, with some non ratio test of significance [one tailed z test].  The number of counts for females was also too low for anything other than counting frequencies.

A 'significant disparity' emerged between the frequencies of different races for males in the magazines as opposed to the actual U.S. population, with white males being over represented: they were also over represented when looking at the population of gamers themselves.  Humanoids were more frequent than minority characters.  Minority women were collected together, as above, and white women were overrepresented in gaining magazines compared to the U.S. population and compared to U.S. gamers: they were slightly more frequent than humanoids.

Both whites and minorities were represented as hyper masculine to the same extent, 21-29 percent.  69% of all images depicted aggression.  Minority males were slightly more likely to be portrayed as aggressive.  The distribution was similar across the different magazines.  Of the aggressive males, whites were more likely than minorities to be found in a war setting [only 8% overall, compared to zero].  Of the aggressive minorities, 51% were found to be in a fighting situation as opposed to 38% for whites.  Whites were more frequently to be found in the non aggression category, 34% as opposed to 23%.  Minority males were more likely to be seen as athletes, 19% to 8%.  More white males than minority males [14% to 6%] were seen using computers.  Minorities who were athletic, violent or both comprised 94% of the cases as opposed to 74% for whites, and all black Males 'were portrayed as either athletic or violent or both'.

Overall, the content analysis shows that not all minorities were portrayed stereotypically, but there were differences between the races 'consistent with stereotypes', whether counting simple frequencies, or comparing the groups.  Overall, 'humanoid characters were depicted more often than minority humans'[so is this good or bad?].  Not a single minority was portrayed in a military setting, which is quite inconsistent with statistics based on U.S. armed forces, where 16% were black, 11% Hispanic, and 6% other minorities.  51% of aggressive minorities were seen fighting, 'another relatively negative portrayal' (297).  Perhaps the most important finding 'from the perspective of pedagogy' is that non violent character portrayals also varied, 23% of all males for minorities, only 6% of total male images.  By contrast nonviolent white males'were as common as all minority males'.  Minority males were more than twice as likely as whites to be portrayed as athletic, and almost the reverse for computer and technology use.  However, it is necessary to 'be conservative when speculating on reasons for these patterns'.

False information is persuasive, a study shows.  Stereotypes include imaginary and actual elements.  Gaming magazines do show 'consistent stereotypes'[one way to read the data, but they are not 100% frequent].  As a result, 'players may experience failures in reality monitoring and may believe that they have had actual stereotype - confirming experiences'[note 'may'] and this 'could alter gamers thoughts, feelings and behaviours'.

The second study looked at game covers, which might be important when influencing the decision of parents to choose the game.  There also easily available to viewing, 'regardless of age appropriateness' (298), and even ' this single exposure is significant' [same study of reality monitoring].  Even non gamers 'are aware of social stereotypes prevalent in video games'.  The same categories were used as above, and data was also gained on 'game ratings, weapon type, and game genre'.  The focus was also on multiple characters and 'complex environmental representations'.

149 games were chosen, based on an earlier sample, including all the top 50 games for the different platforms.  Only five women of color [VII in the original sample of 225] were represented, showing that they [enjoy] 'marked absence'.  This sub sample excluded games without male humans', or ambiguous characters.  An expert panel's report was used to test that the sample was representative.  One male and one female rated each image, and there were intercoder reliability checks: out of 7130 judgments, there were 112 disagreements.

The same codes were developed from the first study, although aggression now included game genre and weaponry.  Ratings were also added.  Characters were also judged to be either primary or ancillary, in the foreground or the background.  Games genre included fantasy, war, cops and special OPS, provoked by regular people, illicit persons, stereotypes and race specific, again yielding socially sanctioned and unsanctioned categories, with some ambiguous types as in martial arts.  Weapons were categorised to include blades, sticks, hands, magic powers, guns and extreme guns.  Frequency data for race were calculated, although Hispanics [too few] and humanoids were  omitted.

[Usual results].  More minorities in games than in the U.S. population.  Whites and Asians exceeded the expected proportions.  White characters appeared in the foreground more often, whites and Asians were more likely to be primary characters, as solo primary, while minority characters shared their primary status more frequently.  With the less violent games, whites and minorities were more equally represented, but with the most violent ones, minorities nearly twice as likely as whites.  Hyper masculinity did not separate the races, but aggression did, with 65% of minority characters compared to 50% for white ones.  Fantasy settings were more common for white characters than minorities, socially sanctioned violence produced about the same percentages, but minorities were nearly five times more frequently represented in illicit violence than whites were.  Weapon characteristics did not differ by race, although minorities were twice as likely to use their hands as a weapon, especially in martial arts.  Minorities were nearly twice as likely to be represented as athletes, blacks appeared as athletes 32% of the time, all minorities 'were significantly over represented as athletic and/or violent' (302) [there could well be significant cultural differences between athleticism and violence, of course].

So the results were consistent with the first study, showing the appearance of stereotypes 'often'.  The overall message is that 'whites are heroic fighters, fighting to save and often romanticized world… or realistic war heroes...saving nothing less than Western Civilization itself', while black characters 'were too often the menace to society with oversized weapons and gang posturing'.  Asian characters are keen on martial arts and only threaten each other, but nor do they save anybody.  Asian and black characters were over represented in the less violent games, while 'black and other characters'(303) were overrepresented in the violent ones.  The only Hispanic characters on the most violent games appeared in things like GTA.  Games are 'rarely offered leadership roles to minority characters outside of the "model" (Asian) minority'.

Overall the two content analyses showed that images of minorities 'are consistent with a number of stereotypes', less frequently represented, rarely appearing as primary characters, more likely to appear at both in the least violent and the most violent games, more or likely to be engaged in unsanctioned violence.  However, some characters were not 'always negatively portrayed'.  A pedagogic function is suspected.

The third study looked at video games themselves, to see if there was a 'stereotype - consistent decision-making'(304).  A particular 'priming paradigm known as the weapons identification task' was used in a test [the idea seems to be to show student faces of different colours, and then ask them to classify stimuli as violent or not]. This was adjusted to fit video games in the real world.  Some game players think that stereotyping and the effects on decision-making is irrelevant, denying that they are affected in any significant way.  'This is common in media effects literature'.  The study set out to test this by asking participants to watch video clips of games being played [not playing themselves then, but apparently video gamers watch games being played quite frequently, so there is 'ecological validity': it also minimises the effects of different skill levels and progress when people actually play].  The race of characters was deliberately manipulated, as was the violence of the game and the use of violent or nonviolent stimuli to classify.

The primary hypothesis was 'that race would influence classification of violent stimuli': watching games with black characters would lead to 'faster classification of violent stimuli and slower classification of nonviolent stimuli'.  39 psychology students were chosen, 26 females and 13 males; 34 were whites, one black, to Hispanic, one from India, one of multiple races.  The ethics followed the usual APA guidelines.  Digital videos of violent and nonviolent games were created - there were fighting games involving one on one and gang violence [clips of commercial games].  Students were asked to classify the photographs of various weapons, for violent stimuli, or photographs of everyday objects for nonviolent.  Race of character, violent or nonviolent games, types of identification stimulus, and sex of subject were the variables [two options for each].  Students looked at clips on a laptop, and responded by pressing keys labelled as a violent or nonviolent.  They practiced.  They saw a clip of a nonviolent game with an unidentifiable character.  Then they were exposed to images of things to identify [for only 250 miliseconds!], and they had to identify them as violent or nonviolent by pressing the relevant key.  Then they saw two more presentations. Then the actual session began.  Students were offered 12 30-second video clips, randomly drawn from the four conditions of violence or race.  Then they were asked to classify three objects as violent or nonviolent.  The same 12 clips were shown again at random, and different images were assigned.  To examine 'mixed relationships between implicit and explicit stereotyping' (305), this test of implicit stereotyping was supplemented by a test of explicit stereotypes - the Modern Racism Scale, to be completed by each respondent.

Reaction times for correct responses were used to calculate an average reaction time for each game combination.  Outliers were excluded, and two students had outlying scores [for example slow reactions on every trial, or lots of incorrect identifications].  The other scores were tested by ANOVA.  There was a significant effects for the sex of subject, with males making faster identifications, but no interaction with any of the other variables.  There were significant correlations to support the main hypotheses, especially with violent stimuli being classified faster than non violent, irrespective of the violence of the video, but affected more by the race of the characters - black game characters with a violent stimulus produced the quickest identifications.  'Stereotype inconsistent' pairings such as a white game character and a violent stimulus, or  black game characters and non violence, produced slower responses.

Overall, this is 'the first empirical test of how video game portrayals can influence race related thinking'(306).  In particular the identification of weapons was faster after watching a video game with black characters.  These were significant results, and serve to refute 'the common assertion that exposure to (racial) game imagery does not have an effect on those exposed' (306-7).  This confirms other studies of the effects of priming, like those suggesting that rating of sexual harassment can also be affected after viewing images [reduced after watching images or professional men and women, apparently].

However, the level of violence of the game itself was not significant and did not seem to interact with the race of the character or the violence of the stimulus to be categorized [which seems contradictory].  However, it could be that physiological reactions are not affected after exposure to violent video games [and this might affect reflexes?].  It could also show 'cognitive prioritizing' where violence is to be expected, but black characters are more surprising because they are less frequent.  We cannot rule out that the violence level of the game affects players and their perceptions of race, but this needs further study.

Students did not express explicitly racist tendencies when tested, but their behaviour was 'certainly consistent with the negative stereotype associating blacks with violence in the United States'.  This is the claimed advantage of the reaction time variable, because it is 'relatively immune to intentional attempts by the participants to respond in a socially desirable manner...unlike the explicit attitudes as measured'.  There might be an unconscious process at work, as in implicit stereotyping, and students may not be aware of it.  It is also possible that students might have become desensitized - the more frequently we are exposed to demeaning images, 'the less seriously we may feel compelled to take them'.  Lots of work on implicit attitude show that bias can exist and influence behaviour, even if there is explicit denial and a study is cited exposing college students to exemplars, and measuring implicit and explicit racial attitudes - disliked black and admirable white exemplars did not show any particular difference, when compared to a control group given neutral exemplars.  [Instead of showing no effects, however], the results were interpreted as showing that '"normal" is the media portrayal of the disliked black and admirable white' (308) - the team tested this by exposing students to admired blacks and disliked whites, and this did reduce automatic pro white attitudes. Together then, dislike of all blacks in video games can have a stereotyped impact.  Reducing prejudice requires looking at the possibilities of doing 'something as simple as increasing the frequency of admirable black characters in games'.

In general, the studies addressed the stories being told about minorities presented in various ways, and some attempts to look at the consequences.  The stories  feature underrepresentation and stereotyping.  We have no pin down the issues for video games, gaming magazines and gain covers.  Future work is required, especially to see if there is a wider impact, for example on views of 'affirmative action, immigration, the death penalty'  (308).  Global diversity makes the issue more urgent, especially as video games are very popular with kids.  'Whereas schools are teaching children to tolerate and even celebrate diversity, this research demonstrates that some forms of popular media ascending opposing signals with troubling effects' (308-9).

back to key concepts page