Violence in E-Rated Video Games

MEDICINE AND THE MEDIA

Kimberly M. Thompson, ScD Context Children’s exposure to violence, alcohol, tobacco and other substances, and Kevin Haninger sexual messages in the media are a source of public health concern; however, content in video games commonly played by children has not been quantified. REATED IN 1994, THE ENTER- Objectives To quantify and characterize the depiction of violence, alcohol, tobacco tainment Software Rating Board and other substances, and sex in video games rated E (for “Everyone”), analogous to C(ESRB) rates video games ac- the G rating of films, which suggests suitability for all audiences. cording to the categories listed in the Design We created a database of all existing E-rated video games available for rent BOX and using content descriptors, or sale in the United States by April 1, 2001, to identify the distribution of games by which game manufacturers display on genre and to characterize the distribution of content descriptors associated with these the game box to inform consumer games. We played and assessed the content of a convenience sample of 55 E-rated choices.1 Analogous to the G rating of video games released for major home video game consoles between 1985 and 2000. films,2 the E rating (for “Everyone”) of Main Outcome Measures Game genre; duration of violence; number of fatali- video games suggests suitability for all ties; types of weapons used; whether injuring characters or destroying objects is re- audiences, but the E rating does not warded or is required to advance in the game; depiction of alcohol, tobacco and other mean violence-free. substances; and sexual content. Studies on children’s use of various Results Based on analysis of the 672 current E-rated video games played on home media document the popularity of video consoles, 77% were in sports, racing, or action genres and 57% did not receive any games as a major source of entertain- content descriptors. We found that 35 of the 55 games we played (64%) involved intentional violence for an average of 30.7% of game play (range, 1.5%-91.2%), and ment. A recent study found that 70% we noted significant differences in the amount of violence among game genres. In- of children (age, 2-18 years) live in juring characters was rewarded or required for advancement in 33 games (60%). The homes that have at least 1 video game presence of any content descriptor for violence (n=23 games) was significantly cor- console, 33% of children have video related with the presence of intentional violence in the game (at a 5% significance game consoles in their bedrooms, and level based on a 2-sided Wilcoxon rank-sum test, t53=2.59). Notably, 14 of 32 games 30% of children in the study played (44%) that did not receive a content descriptor for violence contained acts of vio- video games the previous day.3 Chil- lence. Action and shooting games led to the largest numbers of deaths from violent dren in the study reported playing video acts, and we found a significant correlation between the proportion of violent game games for 20 min/d on average, although play and the number of deaths per minute of play. We noted potentially objection- able sexual content in 2 games and the presence of alcohol in 1 game. older children (age, 8-18 years) accounted for most of this use (aver- Conclusions Content analysis suggests a significant amount of violence in some E- age, 27 min/d), with boys spending sig- rated video games. The content descriptors provide some information to parents and should be used along with the rating, but the game’s genre also appears to play a role nificantly more time playing video in the amount of violent play. Physicians and parents should understand that popular games than girls and white children E-rated video games may be a source of exposure to violence and other unexpected playing video games for significantly less content for children and that games may reward the players for violent actions. 3 time than black or Hispanic children. JAMA. 2001;286:591-598 www.jama.com Unfortunately, little information exists about the ratings and genres of the The health implications of expo- encing media violence exists within the games that children play as a function sure to video games and other media broad medical community.4,5 Al- of their age, sex, family income level, with violent content remain uncer- though several recent studies repeat and ethnicity, although some differ- tain, but considerable concern about the concerns about the content of video 3 ences in preferences exist. Overall, chil- potential impacts of children experi- games6-10 and the marketing of violent dren appear to play relatively more games in the action, adventure, and Author Affiliations: Harvard School of Public Health, Harvard Center for Risk Analysis, 718 Huntington Ave, 3 Harvard University, Boston, Mass. Boston, MA 02115 (e-mail: [email protected]). sports genres, but this may simply Corresponding Author and Reprints: Kimberly M. Medicine and the Media Section Editor: Annette reflect the types of games available. Thompson, ScD, Harvard School of Public Health, Flanagin, RN, MA, Managing Senior Editor.

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

selected the genre that was most com- Box. Entertainment Software Rating Board (ESRB) Ratings and Descriptions monly used to describe the game. Us- 1. Early Childhood (EC) ing the database, we then performed sta- Titles rated EC have content suitable for children ages 3 and older and do not tistical analyses to summarize the contain any material that parents would find inappropriate. distribution of games by genre and con- 2. Everyone (E)* tent descriptors. Titles rated E have content suitable for persons ages 6 and older. These titles To quantitatively assess the content will appeal to people of many ages and tastes. They may contain minimal vio- of games, we selected 55 E-rated video lence, some comic mischief (for example, slapstick comedy), or some crude games that represented the distribu- language. 3. Teen (T) tion of content descriptors and genres Titles rated T have content suitable for persons ages 13 and older. Titles in this and that were available for play on one category may contain violent content, mild or strong language, and/or sugges- of the current major home video game tive themes. consoles in the United States (DC, N64, 4. Mature (M) PS, or PS2). We designed the study to Titles rated M have content suitable for persons ages 17 and older. These prod- include several games on each console ucts may include more intense violence or language than products in the Teen and to play a mixture of both the highly category. In addition, these titles may also include mature sexual themes. popular games as well as ones that did 5. Adults Only (AO) not receive widespread consumer in- Titles rated AO have content suitable only for adults. These products may in- terest. clude graphic depictions of sex and/or violence. Adults Only products are not To explore the possibility of trends intended to be sold or rented to persons under the age of 18. 6. Rating Pending (RP) in series of video games, we also se- Product has been submitted to the ESRB and is awaiting final rating. lected 2 of the most popular series by sales for study: The Legend of Zelda se- *Prior to 1998, the “Everyone (E)” rating was named “Kids to Adults (K-A).” ries in the adventure genre and the Su- per Mario Bros. series in the action genre. We played all of the games in entertainment to children,11 more re- able for rent or sale in the United States these series, including games released search on the impact of violent inter- by April 1, 2001 (accessible at http: for older consoles like Nintendo En- active entertainment, including video //www.kidsrisk.harvard.edu) because tertainment System and Super Nin- games, is needed.4,5,9 Remarkably, no we expected that the level of violence tendo Entertainment System. Since the quantitative analysis exists on content in video games might depend on genre. 2 oldest games in The Legend of Zelda in E-rated video games or on the rela- The process involved using data from series were released prior to the cre- tionship between game content and the the ESRB1 and several Internet sites15-18 ation of the ESRB and have not been ESRB content descriptors. This study to identify all 672 E-rated console rated, we did not include them in our focuses on providing quantitative in- games, verify that each game was re- analysis of E-rated games even though formation to physicians and parents leased in the United States, determine we are confident that these games about the content of E-rated video each game’s content descriptor(s), and would receive E ratings. Finally, for his- games. classify each game by 1 of 11 primary torical comparison, we assessed the genres: action, adventure, casino, fight- content of 8 classic arcade games that METHODS ing, puzzle, racing, role-playing, shoot- have been rereleased as E-rated com- Video game console systems continue ing, simulation, sports, and strategy. A pilations or paired with E-rated re- to evolve12,13 with 3 manufacturers pres- small number of games (n=9) could not makes of the original games. Overall, ently dominating the market: Nin- be classified by these genres and were we played a total of 65 games. tendo, maker of Nintendo 64 (N64); labeled as “other.” Unfortunately, the For consistency, an undergraduate Sony, maker of PlayStation (PS) and subjective process of characterizing student with considerable video gam- PlayStation 2 (PS2); and Sega, maker game play, as well as the complexity in- ing experience played all of the games of Dreamcast (DC). Popular arcade troduced by the growing presence of and recorded all game play directly onto games featuring different types of game games of hybrid genres, have pre- videocassettes for later coding. The stu- play (eg, Space Invaders, Pole Position, vented a universal system for classify- dent played each game to its conclu- Donkey Kong) served as the first home ing video games by genre. For ex- sion or for at least 90 minutes, which- video games and gave rise to the mod- ample, one of the most popular games ever occurred first. Some action and ern video game market with its wide va- in our sample, The Legend of Zelda: Oca- adventure games that allow the player riety of games of different genres.14 We rina of Time, is classified as action, ad- to save game progress are designed for created a database of information about venture, or role-playing on different In- very long play times; consequently, not the universe of E-rated games avail- ternet sites.15,17,18 In such cases, we playing these games to their conclu-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

sion means that some content is missed. harm, the effects of natural disasters, or RESULTS In particular, some games may become the presence of dangerous obstacles that Our analysis of the universe of E-rated more difficult as the player advances and could not be attributed to the actions of video games led to a database of 672 they may offer additional weapons or a particular character. A violent incident games with 99% of these games avail- other more mature content. However, was defined as an uninterrupted dis- able for play on at least 1 of the major in our effort to strike a reasonable bal- play of violence by a character or a group home consoles in our study (DC, N64, ance between playing more games and of characters. We defined characters PS, or PS2). Of these 672 games, our playing individual games for longer broadly, including personified objects sample included 55 games (8.2%). Half times, we determined that playing the that attacked either the player or other (28 of 55 games) appeared on the game to its conclusion or for at least 90 characters. We did not code as violence monthly list that ranks the 25 best- minutes allowed us to obtain a reason- any intentional acts of physical force that selling games in the United States by ably good sample of game play for any represent normal play in a sports game units sold (regardless of ratings).20 In single game. Video games often start (eg, tackling in football or checking in the context of coding, we found good with an introduction and setup, which hockey), because the intention of the agreement between the author who the player may elect to bypass. Conse- player is technically to stop the other coded all of the games and the student quently, we did not include introduc- player without causing injury. We did who played them (␬=0.90). tions and game setup in our coding or code excessive physical contact in sports TABLE 1 shows the comparison of the calculations of the duration of game play, games, such as punching or otherwise at- 58 content descriptors for the 55 nonar- although we did generally observe them. tacking another player (eg, after the foot- cade games played in our sample vs the For consistency, we defined the begin- ball play was over). To quantify the 710 content descriptors for the uni- ning of game play as the first scene where amount of violence, we manually re- verse of 672 E-rated games. Overall, the autonomous movement occurred. corded the starting and ending times of sample has a similar distribution to the With the game play recorded on vid- each incident of violence toward other universe of content descriptors. Our eocassettes for consistency, one au- characters (hours, minutes, and sec- sample does not include any of the 20 thor (K.H.), who also has consider- onds from the beginning of the tape). games that received content ratings for able video gaming experience, reviewed In video games, characters often en- animated blood, realistic violence, vio- and coded all of the games using a stan- gage in a series of violent acts that are lence, mild violence, suggestive themes, dard coding instrument (available on punctuated by brief periods of time informational, strong language, suit- request) and entered the data into a da- spent running toward the next encoun- able for all users, or edutainment. Based tabase constructed with Microsoft Ac- ter. For consistency, we established a on analysis of the 672 E-rated video cess (version 2000, Microsoft Corp, rule that a series of violent acts would games released for home consoles, 384 Redmond, Wash). The first author and be coded as 1 violent incident only if games (57%) did not receive any con- the game player each independently individual acts of violence were sepa- tent descriptors. coded a subset of 10 games to assess in- rated by fewer than 10 seconds of non- TABLE 2 shows the comparison of tercoder reliability. We discussed all in- violent behavior. For each violent in- genres for the 55 games in our sample stances of games that presented diffi- cident, we recorded the type of weapons and the universe of E-rated games. culty in coding with verification of game used for violence, whether the violent Based on analysis of the 672 E-rated details from the undergraduate stu- incident resulted in injury, and the video games released for home con- dent who played the games. We per- number of character deaths attribut- soles, games in the sports (28%), rac- formed descriptive and statistical analy- able to the violent incident. In addi- ing (26%), and action (23%) genres ac- ses using Microsoft Access and Excel tion, for each game, we noted whether count for most of the games.3 Again, our (version 2000, Microsoft Corp) to com- injuring characters or destroying ob- sample has a similar distribution to the pare our sample to the universe of E- jects is rewarded or is necessary to ad- universe of E-rated games, although our rated games. We calculated intercoder vance in the game, whether the player effort to explore trends in 2 series of reliability using the ␬ statistic compar- could select weapons, and whether any video games contributed to our over- ing the duration of violence coded by of the following content was present: sampling of games from the action and both the game player and the primary alcohol, tobacco, illegal drugs, profan- adventure genres. coder (K.H.) in each 10-second inter- ity, and sex. Finally, we also looked for TABLE 3 summarizes the content of val for 30 minutes of game play (or un- the presence of music from explicit- violence in our sample of video games. til the end of the game) for 10 games.19 content–labeled recordings, because a We report the percentage of game play We defined violence as acts in which recent Federal Trade Commission re- depicting violence because compari- the aggressor causes or attempts to cause port found that 2 music companies had sons of the absolute durations of vio- physical injury or death to another char- approved the use of music with paren- lence for the games would be mean- acter. We did not include accidental ac- tal advisory labels in E-rated video ingless given the differences in game tions that led to unintentional physical games.10 play durations that resulted from the

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

We observed that each successive Table 1. Summary of Content Descriptors for Games in Study Sample Compared With All E-Rated Video Games game within The Legend of Zelda series Content Descriptors Games in Sample (%)* All E-Rated Games (%)* had progressively less violence (FIGURE) No descriptors 30 (54.5) 384 (57.1) and fewer deaths per minute; a less clear Animated violence 11 (20.0) 143 (21.3) trend was demonstrated for the Super Mild animated violence 12 (21.8) 82 (12.2) Mario Bros. series. One explanation that Comic mischief 1 (1.8) 45 (6.7) is consistent with our experience is that Mild language 2 (3.6) 24 (3.6) successive games in series may tend to Gaming 1 (1.8) 7 (1.0) involve more complexity in character Use of tobacco and alcohol 1 (1.8) 5 (0.7) development and engage the player in Animated blood 0 (0) 5 (0.7) more exploration and discovery activi- Realistic violence 0 (0) 4 (0.6) ties that will help him/her achieve a Violence 0 (0) 2 (0.3) goal. However, this trend of less vio- Mild violence 0 (0) 2 (0.3) lence may be offset by the tendency for Suggestive themes 0 (0) 2 (0.3) successive games to portray violence Informational 0 (0) 2 (0.3) more graphically and more realisti- Strong language 0 (0) 1 (0.1) cally as technology advances. The lim- Edutainment 0 (0) 1 (0.1) ited evidence of these 2 series should Suitable for all users 0 (0) 1 (0.1) not be overgeneralized. *Totals for these columns sum to 58 and 710, respectively, because some games received more than 1 content de- Although damage to objects was not scriptor. coded as violence in our analysis, we scriptor for violence, we found that the found that games rewarded characters for Table 2. Summary of Genres for Games in games with a violence descriptor con- destroying objects or required object de- Study Sample Compared With All E-Rated Video Games tained significantly more violence (at struction for advancement in 29 of 55 Content Games in All E-Rated a 5% significance level based on a games (53%). Table 3 indicates that 32 Descriptors Sample (%) Games (%) 2-sided Wilcoxon rank-sum test, of 55 games (58%) in our sample de- Action 22 (40.0) 154 (22.9) t53=2.59). Remarkably, we also found picted weapons other than the body and Sports 12 (21.8) 189 (28.1) Racing 9 (16.4) 176 (26.2) that 14 of the 32 games (44%) that re- that the player could select weapons in Adventure 3 (5.5) 19 (2.8) ceived no content descriptors con- 16 (50%) of these 32 games. A total of Puzzles 2 (3.6) 44 (6.5) tained acts of violence an average of 30 of 55 games (55%) used the body as Fighting 2 (3.6) 12 (1.8) Shooting 1 (1.8) 25 (3.7) 37% of the game duration (range, a weapon, 27 games (49%) used projec- Role-playing 1 (1.8) 20 (3.0) 3%-88%). All of the games we played tiles, 16 games (29%) used magic, 13 Strategy 1 (1.8) 13 (1.9) Simulation 1 (1.8) 6 (0.9) in the action (n=22), adventure (n=3), games (24%) used guns, 6 games (11%) Casino 1 (1.8) 5 (0.7) fighting (n=2), shooting (n=1), strat- used a knife or sword, 2 games (4%) used Other 0 (0) 9 (1.3) egy (n=1), and simulation (n=1) genres toxic substances (poisons), 17 games included violence, while only 2 of the (33%) used explosives, and 26 games student completing some games rela- sports games (17%) included violence (47%) used other weapons (eg, fire, ham- tively quickly and requiring at least 90 not associated with normal game play. mers, snowboard). This is not an ex- minutes (mean game play, 57 min- Given the relatively small sample size, haustive list of the weapons that might utes; range, 6.6-136 minutes) for other however, we caution against overgen- be encountered in the games because of games. In our sample of 55 games eralization of these particular results. the limited amount of time that each played, we found 20 games that did not In the sample of 55 games played, 27 game was played; consequently, it should include violent game play, and 35 games (49%) depicted deaths from vio- be viewed as a subset of the weapons de- games (64%) that involved inten- lence. Not surprisingly, the shooting picted in these games. tional violence, with an average of game showed the highest numbers of In addition to coding for violence, we 30.7% of the game duration represent- deaths per minute (23.8). In all 22 of also noted other content in the games ing violent game play for these games the action games, we found that injur- that might have led the ESRB to assign (range, 1.5%-91.2%). We found that in- ing characters was rewarded or was re- content descriptors to the game. For ex- juring characters was rewarded or required to advance in the game. Nearly ample, Goemon’s Great Adventure and quired for advancement in 33 (60%) all of the action games (21/22 [95%]) NFL Blitz 2000 received ESRB content games. Separating the 55 games into 2 depicted deaths from violence, with an descriptors for “mild language.” We groups, 1 group containing 23 games average (arithmetic mean) of 2.3 deaths found the word “damn” printed on the that received a content descriptor for per minute (range, 0 deaths per minute screen in Goemon’s Great Adventure and violence and 1 group containing 32 for Paperboy to 8.4 deaths per minute noted that the players taunt each other games that did not receive a content de- for Rat Attack). in NFL Blitz 2000. Although none of the

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

Table 3. Summary of Violent Content in Games Played by Genre* Length, Deaths Content Sec- Violence, Violence, Deaths, per Game Title Year System Descriptors onds Seconds % No. Minute Weapons Action 40 Winks†‡ 1999 PS AV 1559 194 12.4 34 1.31 B, P, O Donkey Kong 64†‡§ 1999 N64 MAV 4243 312 7.4 55 0.78 B, G, P, E, O Gex 3: Deep Cover Gecko†‡ 1999 N64 CM, MAV 6754 905 13.4 75 0.67 B, G, P, M, O Goemon’s Great Adventure†‡ 1999 N64 MAV, ML 5656 2325 41.1 739 7.84 B, K, G, P, M, O Mario Party†‡ 1999 N64 No descriptors 4619 484 10.5 4 0.05 B, G, P, E, O Monkey Magic†‡§ 1999 PS AV 7203 732 10.2 81 0.67 B, K, P, E, M, O Nuclear Strike 64†‡§ 1999 N64 AV 2007 1830 91.2 179 5.35 G, P, E Paperboy†‡ 1999 N64 MAV 2318 244 10.5 0 0 B, P, O Q*bert† 1999 PS No descriptors 1375 458 33.3 30 1.31 B Rat Attack† 1999 PS No descriptors 1465 1004 68.5 205 8.40 B, O Rayman 2: The Great Escape† 1999 N64 MAV 5113 547 10.7 25 0.29 B, G, P, E Sonic Adventure†‡ 1999 DC MAV 5454 181 3.3 1 0.01 B, P, E, O Spyro 2: Ripto’s Rage†‡ 1999 PS MAV 6137 932 15.2 80 0.78 B, M, O The Smurfs† 1999 PS No descriptors 2622 459 17.5 26 0.59 B, P, E, M Toy Story 2: Buzz Lightyear 1999 PS AV 6447 1254 19.5 100 0.93 B, G, P, E, O to the Rescue†‡§ Kirby 64: The Crystal Shards†‡§ 2000 N64 No descriptors 6173 4424 71.7 483 4.69 B, G, P, E, M, O Super Mario Bros.†‡࿣ 1985 NES No descriptors 2138 883 41.3 171 4.80 B, G, P, O Super Mario Bros. 2†‡࿣ 1988 NES No descriptors 2877 1203 41.8 160 3.34 B, K, P, E, O Super Mario Bros. 3†‡§࿣ 1990 NES No descriptors 4107 1229 29.9 147 2.15 B, G, P, M, O Super Mario World†‡§࿣ 1991 SNES No descriptors 4078 1629 39.9 177 2.60 B, G, P, M, O Yoshi’s Island: Super Mario World 2†‡§࿣ 1995 SNES No descriptors 5375 1427 26.5 265 2.96 B, P, M, O Super Mario 64†‡࿣ 1996 N64 No descriptors 5414 341 6.3 21 0.23 B, P, E Sports 1080° Snowboarding 1998 N64 No descriptors 1477 0 0 0 0 NHL ’99† 1998 N64 AV 1467 22 1.5 0 0 B FIFA 2000 1999 PS No descriptors 1806 0 0 0 0 Mario Golf 64 1999 N64 No descriptors 636 0 0 0 0 NBA2K 1999 DC No descriptors 1693 0 0 0 0 NFL Blitz 2000 1999 N64 AV, ML 1033 115 11.1 0 0 B NFL2K 1999 DC No descriptors 3138 0 0 0 0 Sega Bass Fishing 1999 DC No descriptors 2053 0 0 0 0 All Star Baseball 2001 2000 N64 No descriptors 2605 0 0 0 0 Tee Off Golf 2000 DC No descriptors 4257 0 0 0 0 Tony Hawk’s Pro Skater‡ 2000 N64 AV 3384 0 0 0 0 Triple Play 2001 2000 PS No descriptors 2390 0 0 0 0 Racing Gran Turismo 1998 PS No descriptors 6833 0 0 0 0 Barbie Race and Ride 1999 PS No descriptors 3174 0 0 0 0 Crash Team Racing†‡§ 1999 PS MAV 1467 518 35.3 0 0 P, E, M, T, O Demolition Racer†‡ 1999 PS No descriptors 1772 824 46.5 0 0 E, O Hydro Thunder 1999 DC MAV 1802 0 0 0 0 Lego Racers†‡§ 1999 N64 No descriptors 1777 849 47.8 0 0 P, E, O Speed Devils 1999 DC No descriptors 4939 0 0 0 0 Supercross 2000 1999 N64 No descriptors 3552 0 0 0 0 Ridge Racer V 2000 PS2 No descriptors 1566 0 0 0 0 Adventure The Legend of Zelda†‡§ 1987 NES NA¶ 5477 3748 68.4 869 9.52 B, K, P, E, O Zelda II: The Adventure of Link†‡§ 1988 NES NA¶ 5374 2470 46.0 300 3.35 B, K, P, E, M, O The Legend of Zelda: A Link to the Past†‡§࿣ 1992 SNES MAV 5832 2074 35.6 286 2.94 B, K, P, E, M, O The Legend of Zelda: Ocarina of Time†‡§ 1998 N64 AV 8133 945 11.6 61 0.45 B, K, P, M, O The Legend of Zelda: Majora’s Mask†‡§ 2000 N64 AV 5598 181 3.2 15 0.16 B, K, P, M, O Continued

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

Table 3. Summary of Violent Content in Games Played by Genre* (cont) Length, Deaths Content Sec- Violence, Violence, Deaths, Per Game Title Year System Descriptors onds Seconds % No. Minute Weapons Fighting Monster Rancher 2†‡§ 1999 PS AV 3831 806 21.0 0 0 B, P, M, O Pokemon Stadium†§ 2000 N64 MAV 1726 1276 73.9 0 0 B, P, M, T, O Puzzle Fantavision 2000 PS2 No descriptors 905 0 0 0 0 Who Wants to Be a Millionaire? 2000 PS No descriptors 1202 0 0 0 0 Shooting Centipede†‡ 1999 DC MAV 394 335 85.0 156 23.76 B, G Role-Playing Harvest Moon 64‡ 1999 N64 TA 7141 0 0 0 0 Strategy Worms Armageddon†‡§ 2000 N64 AV 2564 1970 76.8 22 0.51 G, P, E, O Simulation Pokemon Snap 1999 N64 No descriptors 1689 51 3.0 2 0.07 B, E, M Casino Caesars Palace 2000 2000 PS GA 1765 0 0 0 0 Arcade Centipede (1981)†‡ 1999 DC MAV 203 188 92.6 124 36.65 B, G Dig Dug (1982)† 1999 N64 No descriptors 228 153 67.1 30 7.89 B, O Galaga (1982)†‡ 1999 N64 No descriptors 85 85 100 88 62.12 B, G, O Galaxian (1979)†‡ 1999 N64 No descriptors 160 160 100 92 34.50 B, G Ms. Pac-Man (1981)† 1999 N64 No descriptors 276 150 54.3 6 1.30 B Pac-Man (1980)† 1999 N64 No descriptors 358 221 61.7 17 2.85 B Pole Position (1982) 1999 N64 No descriptors 143 0 0 0 0 Q*bert (1982)† 1999 PS No descriptors 179 60 33.5 6 2.01 B *PS indicates PlayStation; N64, Nintendo 64; DC, Dreamcast; NES, Nintendo Entertainment System; SNES, Super Nintendo Entertainment System; PS2, PlayStation 2. Enter- tainment Software Rating Board (ESRB) content descriptors are AV, animated violence; MAV, mild animated violence; CM, comic mischief; ML, mild language; GA, gaming; and TA, use of tobacco and alcohol. Weapons are B, body; K, knife; G, gun; P, projectile; E, explosive; M, magic; T, toxic substance; and O, other. †Injuring other characters is rewarded or is necessary to advance in the game. ‡Destroying objects is rewarded or is necessary to advance in the game. §Players can select their weapons. ࿣This game is rated Kids to Adults (K-A) by the ESRB. ¶This game was released prior to the creation of the ESRB and has not been rated (NA indicates not applicable).

also noted sexual innuendo in Gex 3: games in military battle training. Con- Figure. Trend in the Percent of Violence in 2 Video Game Series Deep Cover Gecko. Finally, in Harvest troversy and concern about the effects Moon 64, which received a content rat- of video games on children con- 80 ing “use of tobacco and alcohol,” the tinue,4-10 although much remains to be Super Mario Bros. Series 70 The Legend of Zelda Series player can choose to purchase and con- learned. 60 sume beer, wine, or liquor resulting in With all of the questions about the

50 a red face and a fall to the floor. We did impact of violence in video games on not find any depiction of tobacco in our children, this is the first study to our 40 play of that game or find any music with knowledge to quantify the amount of 30 Violence, % parental advisories in any of the games violence in E-rated video games and to 20 played in our sample. show that many E-rated games do in- 10 volve violence, killing, and the use of 0 COMMENT weapons in the course of normal play. 19851990 1995 2000 Year of Game Release The first public outcry over violence in No games provide messages about not video games occurred in 1976, when using violence, and some games re- Exidy Games withdrew from the mar- ward or require violence and the de- games received a content descriptor for ket Death Race 2000, a game that struction of objects. “suggestive themes,” we noted the pro- awarded players points for running over The video game genre, ESRB rating, vocative leather outfit worn by Ai stick figures.12,13 In the 1980s, the US and ESRB content descriptors provide Fukami in Ridge Racer V, the screen government began using video games important information about the con- shot between her thighs, and the for military training purposes, and re- tent of the game, and overall illustrate phrases “curb your desire” and “push cently Grossman and DeGaetano22 pub- the considerable variability that exists it to the limit” in the introduction.21 We licized the use of off-the-shelf video in the universe of E-rated games. One

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

implication of this finding is that stud- Remarkably, we found some nearly be aware of games’ ratings, content de- ies that assess video game content with identical games that received different scriptors, and genres, and parents whose a mix of games of different ratings and ratings on different consoles (eg, children play games should actively par- genres might produce very different re- Nuclear Strike 64 and Gex 3: Deep Cover ticipate in game selection and engage sults than studies that focus on a single Gecko are E-rated on N64 but rated T their children in discussion of the game rating and genre; future researchers will for “Teen” on PS), which may make content. Several Internet sites also pro- need to carefully consider the process game selection more confusing. We be- vide helpful information for parents who of selecting the games for their samples. lieve that the ESRB should avoid as- want to better understand the content Clearly, efforts to standardize defini- signing different ratings of the same of video games. In addition to the ESRB tions for genres would be both chal- game on different platforms and should Web site, the National Institute on Me- lenging and helpful. assign the highest rating to all of the dia and the Family’s KidScore media The content descriptors appear to games of the same title to avoid inad- evaluation system offers information to provide limited information about vio- vertently misleading consumers who parents about many types of content in lence. We found that receiving any con- may not appreciate the differences be- video games.23 We compared the 24 tent descriptor for violence (animated tween platforms. games in our sample that also appear in violence, mild animated violence, etc) A few important limitations exist in the KidScore database and noted a few provided a good indication of vio- this study. First, the sample of games differences. In particular, 2 games (Crash lence in the game, but the absence of a represents only a small subset of the Team Racing and Kirby 64: The Crystal descriptor did not mean violence-free. available E-rated games. Second, the re- Shards) that received “green lights” in The definition for the E rating states that sults depend on the actual game play the KidScore database contained signifi- the game “may contain minimal vio- that we recorded and the methods we cant amounts of violence in our sample. lence,” yet our experience shows that used for coding information, which in- Also, we noted that the KidScore data- many E-games contain a significant clude subjective judgment in the defi- base did not note any sexual content or amount of violence and demonstrates nitions and their implementation. nudity in Ridge Racer V. Collaborative ambiguity in what constitutes “mini- While our approach of having a single efforts among industry, parents, physi- mal violence.” We did not see how the person play all of the games and an- cians, advocacy groups, and the re- ESRB distinguishes between different other independent person code all of the search community to standardize me- content descriptors for violence and we games reduces variability in both the dia rating systems and to develop a believe that efforts to standardize the game play and the coding, it may have universal rating system should be pur- definitions of content descriptors would limited the representativeness of the sued.24 be helpful. Another approach to con- game play for different types and skill Our content analysis suggests that sider would be to have content descrip- levels of players and accuracy in the many E-rated video games contain a sig- tors that provide information about the coding. However, our method of re- nificant amount of violence and that an amount of violence using a scale in- cording the game play on videocas- “E” rating does not automatically sig- stead of noting simply its presence. This sette and recording actual screen times nify a level of violence acceptable for very might help consumers distinguish for incidents provided opportunity for young game players. Physicians and par- among games that receive the same de- validation of the coding, which showed ents should understand that popular E- scriptors but contain very different excellent reliability between coders. rated video games may be a source of ex- amounts of violence (eg, Nuclear Strike Third, our use of a broad definition of posure to violence for children that 64 vs 40 Winks or Rat Attack vs The violence focused on the intention of the rewards them for violent actions and that Smurfs). We also noted some inconsis- character may differ from other simi- they may contain other content that is tencies between games that received a larly legitimate definitions. For ex- not expected given the E rating. We be- content descriptor and games that did ample, the sports and racing games, es- lieve that physicians, particularly pedia- not. sentially all of which receive an E rating, tricians, should consider asking pa- Currently, the ESRB rates games provided the greatest challenges to cod- tients about their experience with video based on information and excerpts sub- ing because they contain intended acts games and the medical and public health mitted by the game manufacturer, but of physical contact like checking in communities should play an active role does not play the game before assign- hockey and tackling in football that are in informing parents about the content ing the rating. While giving the same not intended to cause injury, al- in video games. materials to raters may promote con- though other studies or coders might sistency, our experience playing the deem these acts to be “violent.” Author Contributions: Study concept and design, ob- tained funding, administrative, technical, or mate- games leads us to believe that the ESRB Despite these limitations, this study rial support, and study supervision: Thompson. raters should play the finished game, provides important and useful informa- Acquisition of data, analysis and interpretation of data, drafting of the manuscript, critical revision of the including the introduction, before as- tion to physicians and parents about the manuscript for important intellectual content, and sta- signing a rating. content of E-rated games. Parents should tistical expertise: Thompson, Haninger.

Downloaded From: https://jamanetwork.com/ on 09/29/2021 VIOLENCE IN E-RATED VIDEO GAMES

Funding/Support: This research was partially sup- 5. American Academy of Pediatrics, Committee on Available at: http://www.emuunlim.com/doteaters. ported by a gift to the Harvard School of Public Health Communications. Media violence. Pediatrics. 1995; Accessed March 16, 2001. from Mitchell Dong, BA, and Robin LaFoley Dong. 95:949-951. 14. Wolf MJP. The Medium of the Video Game. Aus- Acknowledgment: We thank Seamus Ryan, BA, for 6. Robinson TN, Wilde ML, Navracruz LC, et al. Ef- tin: University of Texas Press; 2000. playing the video games and coding a subset of them fects of reducing children’s television and video game 15. IGN Games. Available at: http://games.ign as part of a summer internship and Fumie Yokota, MS, use on aggressive behavior: a randomized controlled .com. Accessed April 1, 2001. for helpful comments and assistance in organizing and trial. Arch Pediatr Adolesc Med. 2001;155:17-23. 16. CNET GameSpot. Available at: http://gamespot managing the data. 7. Anderson CA, Dill KE. Video games and aggressive .com/gamespot. Accessed April 1, 2001. thoughts, feelings, and behavior in the laboratory and 17. GamePro. Available at: http://www.gamepro in life. J Pers Soc Psychol. 2000;78:772-790. .com. Accessed April 1, 2001. REFERENCES 8. Dill KE, Dill JC. Video game violence: a review of 18. Daily Radar. Available at: http://www.dailyradar 1. Entertainment Software Rating Board. Video and the empirical literature. Aggression Violent Behav. .com. Accessed May 30, 2001. computer game ratings database. Available at: http: 1998;3:407-428. 19. Cohen J. A coefficient of agreement for nominal //www.esrb.org/esrb.asp. Accessed April 1, 2001. 9. Singer DG, Singer JL. Handbook of Children and the scales. Educ Psychological Meas. 1960;20:37-46. 2. Yokota F, Thompson KM. Violence in G-rated ani- Media. Thousand Oaks, Calif: Sage Publications; 2001. 20. NPD data. NPD group essential market informa- mated feature films. JAMA. 2000;283:2716-2720. 10. Villani S. Impact of media on children and ado- tion. Available at: http://www.npd.com. Accessed June 3. Roberts DF, Foehr UG, Rideout VJ, Brodie M. Kids lescents: a 10-year review of the research. J Am Acad 1, 2001. & Media @ the New Millennium. Menlo Park, Calif: Child Adolesc Psychiatry. 2001;40:392-401. 21. Daily Radar. New Ridge Racer V babe revealed. Kaiser Family Foundation; 1999. 11. Federal Trade Commission. Marketing violent en- Available at: http://www.dailyradar.com/features 4. American Academy of Pediatrics, American Medi- tertainment to children: a review of self-regulation and /game_feature_page_180_1.html. Accessed March 16, cal Association, American Academy of Child & Ado- industry practices in the motion picture, music record- 2001. lescent Psychiatry, American Psychological Associa- ing, & electronic game industries. Available at: http: 22. Grossman D, DeGaetano G. Stop Teaching Our tion, American Academy of Family Physicians, and the //www.ftc.gov/opa/2000/09/youthviol.htm. Ac- Kids to Kill. New York, NY: Crown Publishers; 1999. American Psychiatric Association. Joint Statement on cessed March 1, 2001. 23. National Institute on Media and the Family. Kid- the Impact of Entertainment Violence on Children, 12. Herman L, Horwitz J, Kent S. The history of video Score. Available at: http://www.mediafamily.org Congressional Public Health Summit, July 26, 2000. games. Available at: http://www.videogames.com /kidscore/index.shtml. Accessed June 1, 2001. Available at: http://www.aap.org/advocacy/releases /features/universal/hov/index.html. Accessed March 24. Walsh DA, Gentile DA. A validity test of movie, /jstmtevc.htm. Accessed March 1, 2001. Accessibil- 16, 2001. television, and video-game ratings. Pediatrics. 2001; ity verified June 27, 2001. 13. Hunter W. The dot eaters: videogame history 101. 107:1302-1308.

We are an intelligent species and the use of our in- telligence quite properly gives us pleasure. In this re- spect the brain is like a muscle. When it is in use we feel very good. Understanding is joyous. —Carl Sagan (1934-1996)

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LETTERS

and 7.9% (95% CI, 7.5%-8.3%) of women reported having Table. Relationship Between Income Adequacy and the Prevalence of Asthma Among Canadians Aged 12 or More Years* asthma. The prevalence of asthma was higher for the age group Income Adequacy No. (%) Asthma Cases AOR† (95%CI) 12 through 24 years (10.5%) compared with other age groups Men (25-39 years, 6.6%; 40-54 years, 5.1%; 55-69 years, 5.6%; Ն70 Low 7586 (8.0) 529 1.30 (1.00-1.68) years, 6.0%). The prevalence of asthma increased with decreas- Middle 19 241 (6.0) 1241 Reference ing household income in both men and women (TABLE). Af- High 38 431 (5.1) 2242 0.74 (0.63-0.87) ter adjusting for sex, age, history of allergy, household size, and Unknown 19 053 (5.2) 1111 0.89 (0.76-1.05) number of bedrooms, individuals with low incomes had a higher Women risk of asthma while those with high incomes had a lower risk, Low 10 761 (10.4) 1178 1.26 (1.08-1.47) compared with those having middle incomes. The results were Middle 20 938 (8.0) 1702 Reference consistent for men and women. High 36 495 (7.1) 2765 0.79 (0.70-0.91) Comment. This analysis indicates that the risk of asthma in- Unknown 20 527 (7.4) 1579 0.97 (0.83-1.11) creases with decreasing income adequacy in both sexes. We Total Low 18 347 (9.4) 1707 1.27 (1.11-1.45) consider income adequacy, as measured at the household level, 5 Middle 40 179 (7.1) 2943 Reference to be an indicator of familial resources and standard of living. High 74 926 (6.1) 5007 0.77 (0.69-0.86) There exist several mechanisms whereby income adequacy may Unknown 39 580 (6.4) 2690 0.93 (0.84-1.04) influence asthma. Poor inner city housing may increase expo- *AOR indicates adjusted odds ratio; CI, confidence interval. sure to cockroach and mouse antigens. The prevalence of ciga- †Adjusted for sex, age, history of allergy, household size, and number of bedrooms. rette smoking is inversely related to income. Other unex- plored possibilities include living in more polluted terparts with high incomes; however, there was no significant neighborhoods and differences in dietary habits. Our results difference observed between middle- and high-income catego- suggest that asthma control and prevention programs should ries.2 A much larger sample size of the second cycle of NPHS target lower income families to a greater extent than higher in- allowed us to further explore whether the prevalence of asthma come families. However, it is not yet known if income is re- increases consistently with decreasing income. lated to new development of asthma, exacerbations of preex- Methods. We analyzed the cross-sectional data of the NPHS, isting asthma, or both. conducted by Statistics Canada in the period 1996 through 1997. Yue Chen, MD, PhD The design and execution of the survey have been detailed else- Mei Tang, MSc where.3 A total of 173032 respondents aged 12 years or older Daniel Krewski, PhD who responded to the question about asthma were included Department of Epidemiology and Community Medicine in this analysis. Respondents who answered the following ques- University of Ottawa tion affirmatively were considered as having asthma: “Do you Ottawa, Ontario have asthma diagnosed by a health professional?” Robert Dales, MD, MSc Department of Medicine Based on total household income adjusted for the number University of Ottawa of household members, subjects were classified into 3 income categories: low (Ͻ$15000/y for 1 or 2 people; $10000- Funding/Support: This work was supported by the Canadian Population Health Initiative. Dr Chen is the holder of a Canadian Institutes of Health Research in- $14999 for 3 or 4 people; $15000-$29999 for 5 or more vestigator award. people), middle ($15000-$29999 for 1 or 2 people; $20000- 1. Krahn MD, Berka C, Langlois P, Detsky AS. Direct and indirect costs of asthma $39999 for 3 or 4 people; $30000-$59999 for 5 or more in Canada, 1990. CMAJ. 1996;154:821-831. people), and high ($30000-$59999 for 1 or 2 people; 2. Chen Y, Dales R, Krewski D, Breithaupt K. Increased effects of smoking and Ն obesity on asthma among female Canadians. Am J Epidemiol. 1999;150:255- $40000-$79999 for 3 or 4 people; $60000 for 5 or more 262. people) (all currencies expressed in Canadian dollars). Other 3. Peaudet MP, Chen J, Pe´ rez C, Ross N, Wikins K. National Population Health variables included in the analysis were sex, age, history of Survey Overview, 1996/1997. Ottawa, Ontario: Health Statistics Division, Sta- tistics Canada; 1998:1-14. Catalogue 82-567. allergy, household size, and number of bedrooms. Point esti- 4. Rao JNK, Wu CFJ. Resampling inference with complex survey data. J Am Stat mates were weighted according to the demographic profile of Assoc. 1998;83:231-241. 5. Krieger N, Williams DR, Moss NE. Measuring social class in US public health the Canadian population, and the Rao-Wu bootstrap method research. Annu Rev Public Health. 1997;18:341-378. was used to estimate the standard errors of these estimates to take into account the complex survey design.4 Logistic regres- sion models were used to evaluate the association between income adequacy and the prevalence of asthma after adjusting for covariates. Model parameters were estimated by the CORRECTION method of maximum likelihood, and were tested for signifi- Incorrect Wording: In the Medicine and the Media article entitled “Violence in cance using the Wald statistic. E-Rated Video Games” published in the August 1, 2001, issue of THE JOURNAL (2001; 286:591-598), there was an incorrect word in a sentence. On page 596, at the Results. Of 84311 men and 88721 women 12 years of age bottom of the first column, the part of the sentence that read “curb your desire” or older, 5.7% (95% confidence interval[CI], 5.4%-6.0%) of men should have read “control your desire.”

Downloaded From: https://jamanetwork.com/ on 09/29/2021