Introduction

Like all mammals, we learn and grown through play, both alone and with others. The role of play and leisure in modern, Western lives has massively expanded amongst the young and old alike. Perhaps the most radical and highly mediated form of popular play is the online video game. The digital game industry has grown from its humble roots in the 1960’s to an industry with sales of $6.35 billion for 2001 in the US alone [1] ; just behind total box office sales at $8.41B [2] . Video gaming is a major pastime, with children and adults playing increasingly in multiplayer games via the Internet. Teachers, psychologists and even corporate teamwork advisors have long known that social play experiences have formative effects on how the people contextualize their social interactions and relations, and online games are provide these social play experiences to growing millions worldwide. The online game, simply put, is a rapidly expanding and uncharted new form of social medium.

This is why we at the Media Analysis Laboratory of Simon Fraser University implemented the Online Gaming as Emergent Social Media survey. In the preliminary first wave of this audit, we surveyed 569 online gamers from around the world to explore several key issues and concerns relating to the social practices of video gaming. The audit was designed to look at the total context of gamers’ digital game use, both online and offline. Additional focus was afforded to the two particular online titles ­ EverQuest, a massively multiplayer online role-playing game, and Half-Life: Counterstrike, an online first-person shooter. In this report we start to look at what the world’s online gamers have to say about their play.



Executive Summary

Almost all the survey respondents are net-savvy online gamers, with only 7.3% of them playing an average of less than two hours per week online gaming and fully a quarter of respondents playing online more than 25 hours each week. Many of them are somehow socially involved in online gaming, whether through reading online game forums and news sites, chatting with other players, or just going to their local ‘net café for a game. In fact, most respondents came across this survey in one such manner or another, as it was not administered to a random sample of the world’s populace, but publicized in these places that devoted online gamers are likely to see.

And many respondents are devoted gamers indeed. More than 87% of respondents feel that people become addicted to the games, yet less than 19% feel that they themselves are addicted. Nearly half the respondents report that they have been in conflict with family or friends over their online gaming, yet continue to play.

Of course, not all gamers are the same. In fact, we have found that online gamers can be statistically categorized into four archetypal components; the warrior, the narrator, the strategist, and the interactor. Statistical analysis suggest that these four archetypes account for more than two-thirds of an online gamer’s general gameplay preferences.

 

Methodology

Over the run of this audit, 709 respondents took part in the first section of the survey and an additional 469 took the entire survey in two parts, making a total of 1178 respondents to the first and more general section of the survey. These results were gathered from August 21st, 2002 through November 9th, 2002. The two sections of the questionnaire are available for reading here and here, respectively. The draw of the survey, besides the chance for participants to offer their opinions about their hobby, was the chance to be selected for one of three thank you gift certificates, good for $60 CAN each.

Recruitment for the survey was accomplished in several various ways. The first and simplest was by "word of mouth". We in the Media Laboratory asked, in person or by e-mail, our friends and acquaintances, resident local gaming dens and Internet cafes, and fellow students and staff to spread the word to online game players they know, particularly those that play EverQuest of Half-Life: Counterstrike. Several Greater Vancouver gaming cafes posted our flyers on their doors, while posters were delivered to comic book and game stores.

More rigorous recruiting was performed online, where word of the questionnaire was submitted directly to general video game 'news' web sites such as Gamers.com and Gamespy. Web sites focused on EQ and CS such as Everlore.com and Counterstrikecenter.com were particularly targeted for these submissions. Online comic strips such as GUComics.com and Penny-Arcade.com were also notified. Further, whenever a CS or EQ Internet forum was discovered notice of the survey was posted in an appropriate area. Finally, notice was posted in various non-video game related, yet still-receptive forums where gamers might be found. Examples include the paintball site PBReview.com and Slashdot.net. All in all, more than 100 news and forums submissions were made. The greatest period of submission occurred within a week of the questionnaire going live.

Sample bias, in this case, is fairly evident. Internationally, there is an obvious bias towards English-speaking nations, and results show that Americans and particularly Canadians are strongly represented in this data set. Given the origins of some responses, an inordinate number of respondents are students at Simon Fraser University (estimated: 10), yet combined this accounts for less than 0.1% - a figure which we have decided is unlikely to skew our results noticeably. Age bias is difficult to identify given the lack of data from other sources, but the intimidating nature of the survey itself and the personal savvy needed to access the web forums where much of the survey's publicity originate from suggest that very young players are not fairly represented in our results. In terms of game choice, the representation of CS and EQ players is fortuitously similar.

Another bias of the ways the survey was publicized is that the participants are typically more dedicated or "hardcore" gamers; the ones to find our notices in news and forums are the ones that seek out extra information or interaction with the communities of online gaming. Such gamers might be more likely to take gaming seriously, be more informed of social issues surrounding gameplay, take higher interest in out-of-game socialization, and so on. Gender representation in the survey rests at 10% female, which some figures suggest is high in comparison to computer games on the whole, yet low for EQ. Because no effort was made to target either men or women specifically we feel that any sampling error here is simply related to gender and access to web forums and news sites. In order to avoid any skewing of the data caused by the gender break, many questions are analyzed for both genders comparatively, when appropriate.

In hindsight, we see now how valuable a "How did you hear of this questionnaire?" query would be in identifying not only the most effective ways to contact online gamers, but also to address these very issues of representativeness and sampling. Nevertheless, The goal of this audit was never to find a perfect cross-section of one in 25000 people that have ever played online. Instead, it is more of a foray into the realm of sociality as it is found in online gaming communities and environments. Strategically, we chose to simply reach as many people that were willing to take a lengthy, 20 minute survey as possible. We are working to refine our sampling process in order to avoid data contamination or sampling bias as much as feasibly possible for future online surveys, and if you have any suggestions we would be most happy to hear from you.

Each response was checked to ensure that it was not submitted in error or duplicate by checking the time it was sent and from which Internet service provider address (for lack of a better term; not the participants IP Address, which was not recorded). Further, each response was checked to ensure that no missing responses went unmarked, which could cause later questions to be mislabeled. Approximately 150 responses (either a first or second half of a given survey) were discarded, with 9 out of 10 of these being clearly the result of somebody accidentally submitting twice in rapid succession, or accidentally submitting when only partially finished that page and going back to redo the parts of the page they skipped. Some other responses were garbled by participants deleting default data from their survey (which was repaired when we could be absolutely certain where the errors were - 3 cases), or someone clearly submitting blank responses to see the second page of the questionnaire. In some cases the last few responses on a given page were untransmitted, presumably by someone closing their window immediately after submitting, resulting in only a partial transmission of data. In these cases (approx. 25) we simply entered a "Don't know / n.a." or "Null Data" value for these missing responses.

Results were collected by e-mail and individually transferred into a tab-delimited format for entry into the analysis program. Answers that were left "Don't know / n.a." were marked as excluded from calculations. Every precaution was made to back up data and ensure that there was no manual character transcription, in order to avoid clerical errors. In future, we hope to use a purpose-built SQL server we are designing to streamline the process and help further avoid human error.

One final note on how this data has been analyzed thus far - what you see below is a preliminary report of the most surface-level findings and some interesting things that have come up out of our work, but this is only the tip of the iceberg. Currently, we at the Media Laboratory are working on several concurrent pieces we hope to produce, and are confident that this 327 question survey hasn't revealed all of its treasures yet, so, please, pop in on this site once in a while...

General Findings

Most respondents have been playing online games for 3 to 6 years. In terms of what is important or very important in a game, they tend to rate exploration (88.5%) and themes or plot (88.5%) most highly, then good characters (86.2%), graphics (79.3%), the opportunity to cooperate with other players (76.1%) and innovation in game design (74.6%) well ahead of other gameplay elements. Following these were unpredictable gameplay (68.9%) gameplay that make them think a lot (66.9%) and feelings of control while they play (66.8%). Complex strategies (60.1), imaginative gameplay (59.5%), constant excitement (56.0%), challenge (54.8%), and competition against other players (53.9%) ranked ahead of weapons and technology (48.1%), realism (41.3%), fast-reaction play (36.9%), and military or combat themes (31.9%). Finally, gamers were evenly split between feeling calming gameplay is important or unimportant.

In terms of online genre preferences, most respondents like or strongly like RPG’s and fantasy games online (85.7%), followed by fighting and shooting games (75.9%), real-time and turn based strategy or conquest games (67.4%), simulations (38.0%), platformer, maze and adventure games (36.0%), racing games (25.8%), puzzle, educational and board games (25.2%), sports games (17.4%), and finally gambling games (9.2%).

When asked specifically about online gaming, respondents report that communicating with other players is very important or important (87.5%), while exploration of new online game environments (83.6%) and teamwork (82.6%) are also important in typical online games. Secondary in importance are trying out new characters (79.9%), building a character's power, money or items to use again (77.4%), generosity and giving help to other players (77.1% - with only 5.2% saying it is unimportant), seeing old friends by playing (74.1% - with only 5.0% saying it is unimportant), practicing their skills (73.8%), and relying on and being relied upon by other players (73.2%). Tertiary elements include building a reputation (66.2%), making new friends by playing (65.5% - with only 8.0% feeling that this is unimportant), competition with other players (58.9%), trading items between players, characters or accounts (57.2%), and learning game secrets without help (57.2%). Other somewhat important elements include role-playing a character’s personality (55.9%), winning the game (51.0%), getting a good score (49.4%), impressing or charming other players (47.3%), joining a clan or guild (43.1%), puzzle solving (41.1%), defeating computer opponents violently (39.5%), learning game secrets from others (38.2%), and defeating other players violently (36.5%). Ranking poorest were learning secrets from magazines, guides or online (26.5%) and frightening, intimidating or dominating other players (22.0% - with 48.7% feeling this was unimportant).

A small number of online gamers have exploited a game flaw against other players in EQ, CS or other online games (10.1%, 15.3% and 18.5% respectively), but are more likely to do so if no other player directly suffers harm, particularly in EQ (24.5%, 18.9% and 18.5% respectively). Fewer player admit to having used a disallowed game hack or program against another, with more having done so in CS (3.4%, 7.5% and 4.5% respectively) and many more having done so against nobody in particular (7.3%, 8.2% and 15.0% respectively). Few players in EQ players, more than twice as many in CS players and many in other online games have intentionally betrayed their team mates or companions during gameplay (8.8%, 18.9% and 15.6% respectively).

Three in five (58.5%) of respondents have played EQ, with a similar number having played CS (55.3%). Of the regions in which significant numbers of respondents participated, Americans are most likely to have played EQ (68%), followed by Europeans (47%) and finally Canadians (34%) while Europeans and Canadians were more likely (63%) than Americans (53%) to have played CS. Most European EverQuest players are experts or pros at the game, Americans taking the middle road and Canadians are twice as likely to be EQ rookies and novices with very few pros. Counterstrike players in all three regions tend to be approximately equal in skill.

Transgressions	Scale: 1=”Strongly like”, 5 = “Strongly dislike”
When another player exploits a game flaw to get an edge against other players in EQ	4.38
When another player exploits a game flaw to get an edge against other players in CS	4.55
When another player exploits a game flaw to get an edge against other players in other online games	4.50
When another player exploits a game flaw to get an edge against other players in offline games	3.80
When another player exploits a game flaw to get an edge against nobody in particular in EQ	3.76
When another player exploits a game flaw to get an edge against nobody in particular in CS	4.10
When another player exploits a game flaw to get an edge against nobody in particular in other online games	3.85
When another player exploits a game flaw to get an edge against nobody in particular in offline games	3.36
When another player uses a disallowed game hack/cheat or other program to do harm to others in EQ	4.84
When another player uses a disallowed game hack/cheat or other program to do harm to others in CS	4.88
When another player uses a disallowed game hack/cheat or other program to do harm to others in other online games	4.85
When another player uses a disallowed game hack/cheat or other program to do harm to others in offline games	4.41
When another player uses a disallowed game hack/cheat or other program to do harm to nobody in particular  in EQ	4.26
When another player uses a disallowed game hack/cheat or other program to do harm to nobody in particular  in CS	4.45
When another player uses a disallowed game hack/cheat or other program to do harm to nobody in particular  in other online games	4.33
When another player uses a disallowed game hack/cheat or other program to do harm to nobody in particular  in offline games	3.84
When another player harms their team or group by intentional betrayal in EQ	4.75
When another player harms their team or group by intentional betrayal in CS	4.72
When another player harms their team or group by intentional betrayal in other online games	4.62
When another player harms their team or group by intentional betrayal in offline games	4.21
When another player harms their team or group by mistake in EQ	3.44
When another player harms their team or group by mistake in CS	3.52
When another player harms their team or group by mistake in other online games	3.42
When another player harms their team or group by mistake in offline games	3.41
When another player harms their team or group by being a poorly skilled player in EQ	3.77
When another player harms their team or group by being a poorly skilled player in CS	3.54
When another player harms their team or group by being a poorly skilled player in other online games	3.56
When another player harms their team or group by being a poorly skilled player in offline games	3.46

Here we see that CS players are more likely to take offense when another player exploits a game flaw against nobody in particular, while people are generally more forgiving of transgressions that occur in offline play than in EQ, CS or other online games. It is generally seen as less unacceptable to cheat if it doesn’t directly harm other players, using a hack is a little worse than simply exploiting game mechanics, betrayal is the grossest transgression and poorly skilled players and mistakes are less likely to upset other players than cheaters or exploiters.

 

Addiction and Online Gaming

Parents, educators, friends and government officials are often concerned about video games addicting players, particularly violent games. Due to its social nature some are concerned that online games pose even more threat, and the recent suicide of 21-year-old EQ player Shawn Woolly has only heated up the debate. While not everyone agrees on what addiction is, we set out to learn the opinions of the online gamers themselves.

Three common elements of addiction are displacement, social problems and issues of control or a lack thereof. The term displacement refers here to those times when other activities are put aside for gaming. Three out of ten (30.5%) admit to frequently playing online when they should be doing other things. Many players frequently lose sleep or stay up too late playing (27.8%). Social problems for gamers may often arise from addiction, such as has happened to 45.2% of respondents who have been in conflict with their friends or family about their gameplay in the past. Nevertheless, only one in 20 (4.4%) are frequently in such conflict now. Lack of control in one’s life can stem from addiction, and only 14.5% of respondents feel that their gaming is never out of their own control. Only 8.4% of players feel that they play too often.

Overall, nine out of ten (87.2%) of online gamers feel that some other people get addicted to online games, while only 5.4% disagree with that viewpoint.  Overall, 29.4% of respondents feel that they frequently play too often, yet only 18.4% feel that they themselves are addicted and 76.0% actually disagree that playing online games is an addiction to them.

In all of these circumstances, EQ players report more frequent addiction or associated feelings with 3.9% admitting that they are even now in conflict with friends and loved ones all the time over their gameplay, and 21.6% feel overall that they are addicted to EQ. On the other side of the coin, CS players report less frequent feelings of addiction regardless of that title’s higher level of violence, with only 10.4% admitting addiction. There is one exception, however, in that 4.3% of CS players admit conflict with family and friends over their gameplay even though all other indicators suggest a lesser feelings of addiction.

 

Gender and Gameplay Practices

It may be surprising that female and male gamers are very similar in many ways. Gender has negligible effect on gamers’ preferences neither between console and computer games, nor between online and offline games. Further, the amount of time these gamers spend playing games online each week is similar, with a few more females playing 25+ hours online than males (33% vs. 25%).

While both female and male online gamers tend to be 13 to 25 years of age, female gamers are twice as likely to be 40+ than males (9.8% vs. 4.9%), while males are more likely to be under 18 (28.5% of males are youths vs. 17.7% of females). Females are more likely to have children that they do not live with (9.8% vs. 2.9%), and females are more likely to be married or dating than single (40.0%, 40.0% and 20.0%, respectively), while only half of male online gamers are married or dating (26.5%, 24.0% and 49.5%).

Gender vs. Predominant occupation	Non - employed, non-student	Student	Home maker	Administrator / Owner of large business	Administrator / Owner of small business	Professional	Technician / Semiprofessional	Office worker / White collar	Tradesperson / Blue collar	Un -skilled worker	Sales / Service	Farmer / Fisher	Arts	Other
% of Males	2.6	42.5	0.2	11.1	0.4	3.2	14.6	6.3	2.8	0.8	3.6	0.2	2.8	8.7
% of Females	0.0	19.1	4.3	12.8	2.1	4.3	10.6	17.0	0.0	2.1	10.6	0.0	0.0	17.0

From the table above we can see that male online gamers are twice as likely to be students (2 in 5 males), and more likely to be unemployed, blue collar, technical or semiskilled workers, agricultural workers, or artists. Females are far more likely to be homemakers, though less than 1 in 20 is. Females are also more likely to be white collar workers, small business operators, sales or service people, or unskilled workers.

Females, in general, tend to be newer video game players (30% having played < 8 years, vs. 20% of males), while also being slightly more recent adopters of online play (14.0% having played < 1 year, vs. 5.2% of males). Both males and females rate equally gameplay elements such as feelings of control and excitement, the use of fast reactions, thinking hard and strategizing. Also, both genders rate the importance of innovation in game design and gameplay similarly. It is ‘very important’ to more females than males that games have great graphics (45.1% vs. 27.7%), characterization (80.4% vs. 47.1%), themes and plot (76.5% vs. 53.0%), exploration (64.7% vs. 49.7%), and imaginative play (37.3% vs. 22.8%). Fewer females rate the ability of a game to calm them down as unimportant (16.0% of females vs. 29.4% of males).

Many males find weapons and technology (49.4% vs. 35.2%) and competitiveness (56.3% vs. 33.3%) to be important or very important, and  they are less likely to rate unpredictability (6.1% vs. 21.6%) and combat or military themes (30.2% vs. 64.0%) as unimportant or very unimportant. While males and females on average feel the same about cooperation in gameplay, females tend towards indifference more than male players.

Both sexes are just as likely to have tried EverQuest (58.8% of females and 58.7% of males), while females are more likely to be expert or pro players (63.3% vs. 45.6%) while males are more likely to be novices or rookies (16.6% vs. 6.6%). This is consistent with the findings that both males and females rate role-playing games very similarly both on and off the Internet It is worthy of noting that roughly 10% more of both males and females like or strongly like RPGs online than offline.

Counterstrike, not surprisingly, is another story. Only 17.6% of female respondents have tried the game, compared to 59.1% of males, and only two female respondents in total claim to be of average skill. This is supported by the findings that 24.8% of males strongly liking “Fighting / Shooting” games offline and 46.3% of online, while only 4.2% of women rate fighting / shooting games that well regardless of whether they are played online or offline.

Other genres that the sexes disagree upon are: sports, which females are more likely to dislike (86.5% vs. 52.3%) and racing which more females strongly dislike (45.7% vs. 18.3%), while they like puzzle (55.0% vs. 21.9%) and gambling games (23.1% vs. 7.8%) far more often.

Females and males both feel very similarly about platformer / maze, simulation and strategy / conquest games both online and offline. It should be noted that both genders rated strategy games as being the same online or offline.

 

Socialization & Extra Activities

As already noted, the ability to make new friends and spend time with old friends is very important to online gamers, while social transgressions such as cheating or betraying one’s comrades are highly disliked. Cooperation generally ranks higher than competition, and very few online gamers felt that helping others was unimportant. Impressing other players and building a reputation was more widely seen as a priority than intimidating or dominating others. Role-playing did not rank as a top priority, but it was higher that joining clans, guilds or tribes.

Fully a quarter of respondents have created artwork, writing or other game related work for other online games (26.0%) and offline games (26.7%), with nearly as many having done work on EQ (21.4%)- much fewer have done so with CS (11.3%). When compared, EQ players are very likely to maintain a game website (19.8%) with CS players less so (10.8%). Online games in general tend to fall somewhere in between (15.8%) with the fewest respondents having sites for offline games (9.8%). More than two in five EQ players are more likely to have had helped run an actual in-game clan, tribe or guild (42.6%), while CS players are less so (27.1%), the two of which bracket the figure for online gaming in general (32.5%), while a surprisingly high number of respondents are do so for offline games (16.7%). Comparatively, few EQ players (4.3%) worked for the company, either volunteer or paid, while many more CS players did (19.1% - most of which presumably run servers), slightly fewer have done so for other online games (13.7% - again, likely many running servers), while one in twenty says that they have done so for offline games (4.9%). Likely, a few percent in each case were beta testers, at least in the online games.

Respondents report that playing EQ requires more money than playing other online or offline games, while playing CS is the least expensive option. Online games are generally more expensive than offline games as well. When it comes to using money in the game we find a similar response, with gamers responding in each case that having money in games is more important to gameplay than having real money to finance physically playing them.

1 = Very Important, 5 = Very Unimportant
Having in-game money to buy items in CS:
2.60
Having in-game money to buy items in offline games:
2.50
Having in-game money to buy items in other online games:
2.35
Having in-game money to buy items in EQ:
1.97

1 = Very Important, 5 = Very Unimportant
Having real-life money to play CS:
3.94
Having real-life money to play offline games:
3.30
Having real-life money to play other online games:
3.08
Having real-life money to play EQ:
2.66

All kinds of gamers are likely to talk about the game in real life (EQ ­ 85.9%, CS ­ 80.0%, other online games 88.7%, offline games 86.7%), but fewer have ever “role played” a character, especially so in Counterstrike (EQ ­ 69.9%, CS ­ 16.2%, other online games 54.3%, offline games 41.0%). Still, it is interesting to note that nearly 3 in 10 EQ players have never role played in a game sold as a “massively multiplayer role-playing game”. Also surprising is the fact that more respondents that play EQ reported to having communicated strategy and tactics during play (92.2%) than CS players (88.1%), even though CS is widely thought of as the most tactical popular first-person shooter. In fact, CS is only just ahead of online games in general (85.7%), although non-online games truly lag (56.8%) as they are obviously less multiplayer. No respondents have paid another player to help maintain or build their reputation or power, but a few EQ players (2.3%) have had someone do so as a favour. Some EQ (6.1%) and even a few CS (4.2%) players had someone take their place for fun, however.

We see the same recurrent pattern of social interaction when finding that EQ players are the most likely to have helped other players during play (85.9%) with CS players (62.5%) falling below the norm for online games (75.6%), and of course offline games are inherently less social and have seen fewer players help others (44.2%). There is another factor, however, in that most players have been playing offline games far longer yet have still been less likely to have helped another person.

Intense social relations are the most stirring form of network to resolve in online gaming. Nearly half of EQ players (47.5%) have made real-life friends through their play, while one quarter have done so in CS (24.8%). One third of players have done so in other online games in general (36.8%), and nearly one in five while playing offline games (18.9%). A few EQ players (6.0%) have pursued real-life romantic relationships through the game, behind other online games (8.3%). Even offline games (0.7%) are ahead of Counterstrike, in which nobody has pursued real romance. Pretend romance is different, however, fully a quarter of EQ players have role played pursuing romance for their characters (25.4%) with even a few CS players (2.2%) having done so. The online gaming norm falls in between (16.1%) with gamers finding their characters romance half as often (8.6%).

 

Gamer Archetypes

Early factor analysis and data reduction have yielded data that suggests four distinct archetypes of online gamers, but first, a brief explanation of how this works... Basically, factor analysis is a mathematical process that looks for commonalties between certain respondents, then tries to organize them into categories based on which other people answered similarly. For instance, a car manufacturer may take 800 participants and ask them a host of questions regarding their preferences in wheel base, sound systems, safety, color and so on. If everyone ranks safety equally high, or equally low, then the question of safety is not useful for differentiating between different categories of car buyer. Perhaps, however, they will find that those people that tend to rate large wheel base very highly compared to others rate a good sound system as far less important that others, comparatively, and so on. In this way, they can categorize car buyers. These categories, "factors", or "archetypes" as we call them.

In the case of our data, the most distinct are those we may call ‘warriors’ who prioritize weapons and technology, combat and military themes, realism, graphics, and to a lesser degree, fast-reaction and unpredictable play. Comparatively, they do not find interesting characters or being made to think a lot during gameplay to be very important Second, there are the gamers we may call ‘narrators’ who place priority on themes and plot, characters, exploration, using their imagination and thinking a lot, but they do not like games that are challenging and hard to master, competition with other players or combat and military themes. The third group could be called ‘strategists’. These gamers focus on complex strategies, challenging gameplay and mastery, being made to think a lot, use of their imaginations, and that gameplay be unpredictable, with everything else being comparatively unimportant. Finally, there are the ‘interactors’ whom rate competition and cooperation with other players above all else, while they do not care about unpredictability or being made to use their imaginations.

You can see the data summarized below on the "Rotated Component Matrix". Values close to zero, either positive or negative, indicate relative indifference to that element of gameplay, while highly positive values indicate a strong comparative regard for that element, while a highly negative value indicates a relative dislike for that element of gameplay.

Rotated Component Matrix

	warriors	Narrators	strategists	Interactors
graphics	.636	.301	-.098	.117
realism	.668	.011	.147	.200
weapons & technology	.829	.029	.084	.048
combat or military themes	.763	-.087	-.101	.104
characters	-.183	.770	-.073	-.048
themes and plot	.089	.780	-.012	-.036
complex strategies	-.016	.066	.749	-.044
fast reactions	.456	.307	.290	.289
imagination	.226	.584	.475	-.176
exploration	.221	.593	.058	.078
make me think a lot	-.121	.448	.608	.042
unpredictable	.385	.024	.475	-.266
competition with other players	.212	-.097	-.014	.882
cooperation with other players	.166	.038	.008	.843
challenging and hard to master	.039	-.239	.675	.165

Extraction Method: Principal Component Analysis.  Rotation Method: Varimax with Kaiser Normalization.
(Rotation converged in 5 iterations.)

Without getting into too much depth we can see a number of revealing details represented in the numbers here. In one brief example, we see that the Interactors rate both cooperative and competitive play foremost and equally; this does not mean that anyone that likes competition equally likes cooperation, not at all. What this does indicate, however, is that those who rate competition and cooperation very highly rate other amply elements so similarly that, provided our results are sound, they can be statistically grouped together! They put very similar emphasis on exploration, combat and military themes and unpredictability, and all the other elements we asked them about. For game developers this suggests that cooperative and competitive games can be quite similar indeed.

While archetypes seem intuitive at first glance, the significance of these findings is that they offer quantitative support for characterizing different kinds of online gamers, into four categories that may not have seemed the first four most intuitive breaks. It is not the case that the taxonomy suggested here is rigid and complete in stereotyping every player, instead, each gamer can be said to be comprised of a unique combination of warrior, narrator, strategist, and interactor. Once established, such a system for identifying a gamer’s personality would offer numerous avenues for more in-depth research on the social interactions and networks that are formed in online game spaces.

So, why four archetypes, and not three, or ten? On the "Scree Plot" below and to the right we see 13 different factors (or archetypes) distinguishing themselves to various degrees. We can clearly see that while factors 5 - 13 vary in Eigenvalue, the approximate slope of that section of the graph is nearly uniform, while the factors 1 - 4 (Warrior, Narrator, Strategist, Interactor) distinguish themselves more clearly from each other, in respective order. Note, also, that the graph bears some similarity to the positive portion of the ( y = 5 / x ) function.

On the left we see some tests used to determine if the factor analysis is statistically sound. A KMO Sampling test indicates a valid analysis at greater than 0.5, as does the null-hypothesis of Bartlett’s test of sphericity under 0.05. Also, the scree plot we looked at earlier indicates that all factor components (the 4 archetypes) have an Eigenvalue greater than one, as well as representing the steepest region of the slope, which we noted before. The summary of the Varimax-rotated component matrix indicates scores between -1 and 1 which indicate association with the values on the left to each given component, as we also noted earlier. You may note that the variable “fast reactions” appears with general equity across all components. Removal of this variable from the matrix increases the apparent differences between components, but doing so very slightly decreases the rotated sum of squared loadings which is indicative of how much behaviour is accounted for by the included components, i.e.: the amount of data that is kept throughout the reduction process. We decided, in order to maintain the integrity of the data, to leave this factor in for the time being.