March 20, 2007

The Demographics of Game Choices

Many MMOs allow players to choose sides or affinities. Sometimes, like in City of Heroes/Villains, players can choose between a clearly good or evil side. In other games, like World of Warcraft, the good/evil divide is a little less clear cut. Of course, these choices don’t always revolve around good and evil. For example, they may involve choosing a race, or choosing between technology and nature. Here I’ll present some data on different hypothetical choices that a game may ask a player to make when creating a character or during character development. What we’ll see over and over again is that whenever a game asks players to make a choice, players seldom evenly distribute themselves. Instead, other factors, such as age, tend to be correlated with those choices. This in turn creates demographic differences between sides/affinities in games.

For example, let’s start with the good and evil split. The graph below shows both an age and gender difference. Younger players prefer to be on the evil side, and this tendency decreases as age increases. There is also a relatively more mild gender difference in the mid-ranges where male players are more likely to be on the evil side than female players. What’s interesting here is that the distribution roughly matches the 2:1 Alliance-to-Horde ratio in WoW.

====

Another choice that almost all MMOs ask players to make is to choose a race. To simplify matters, I choose 3 races - Elves, Humans, and Undead. Overall, there were almost no age differences, and the only striking difference that emerged was that female players were more likely to prefer playing Elves whereas male players tended to be more evenly distributed.

====

Another kind of choice that has less often been made explicit in MMOs is choosing a source of power. For example, a pet could be magic-based (e.g., elemental), nature-based (e.g., wolf), or technology-based (e.g., mechanical droid). Demographic differences with these three hypothetical choices were interesting.

Male players were about 3 to 4 times more likely to prefer a technology source than female players, and female players were about twice as likely to prefer a nature source as male players. Preference for a magical was roughly comparable between male and female players. There was also an interesting age trend among the male players. Preference for technological sources decreased as a function of age.

====

And finally, another functional divide in most MMOs is between dealing damage and healing damage. In this particular question, I focused on spell-casting and whether players would prefer a damage spell or a healing spell. Here, the gender difference is quite clear. Female players prefer healing spells over damage spells. There is also a mild age difference. Older players tend to prefer healing.

Added Note (03/25/2007): I wanted to add briefly how complicated these gender findings can be, and how they might be a reflection of other variables. For example, we also know that female players are more likely to be playing with their romantic partners. And it makes sense that a player's preferences might be shaped by their game-play history. Is it possible that men "encourage" their romantic partners to play healing/support classes to help their own playing styles? Or for example, women are more likely to be playing in the same room with someone else playing. Might this contextual difference lend itself to preferring more social modes of play? So I would caution against interpreting the graphs in this article as being purely driven by gender (i.e., women are biologically hard-wired to prefer passive, supporting roles), but rather, as the complicated outcome of gender, cultural, and contextual differences.

Posted by nyee at 4:45 PM | Comments (27) | TrackBack

July 28, 2005

WoW Gender-Bending

23% of male players (N = 939) listed a character of the opposite gender as their most enjoyable character compared with 3% of female players (N = 184). In WoW, men are about 7-8 times as likely to gender-bend than women.

Given a hypothetical pool of 1000 players:
840 would be male players
160 would be female players

Of the 840 male players:
193 would be playing a female character
647 would be playing a male character

Of the 160 female players:
5 would be playing a male character
155 would be playing a female character

Thus, altogether there are 348 female characters of which 193 (55%) would be played by a male player.
And there are 652 male characters of which 5 (< 1%) would be played by a female player.

In other words:
about 1 out of every 2 female characters is played by a man
about 1 out of every 100 male characters is played by a woman

The RL gender distribution is 84% male vs. 16% female.
The in-game gender distribution is 65% male vs. 35% female.

Men over the age of 18 are more likely to gender-bend (25%) than men under the age of 18 (10%). Players who gender-bend score significantly higher on the Customization and Mechanics motivations than players who do not gender-bend. In other words, the two primary reasons why players gender-bend are to be able to be more stylish and to optimize their character (via gifts?).

Posted by nyee at 4:28 PM | Comments (161) | TrackBack

WoW Character Race Demographics

Younger players tend to prefer being Trolls and Undead, while older players prefer being Orcs and Dwarfs. Interestingly, 3 Horde races have the lowest average age - Orc being the exception.

Night-Elves and Gnomes significantly more likely to be played by a female player. Orcs and the Undead are more likely to be played by a male player.

An overview of the motivation differences among players who prefer different character classes is shown below, followed by the detailed charts for each motivation. Only those motivations that were significant in an ANOVA are listed. What’s interesting is the repeated differences between the Gnome/Dwarf vs. the Undead/Tauren.

Click here for an overview of the motivations framework.


====


 

Posted by nyee at 4:25 PM | Comments (17) | TrackBack

WoW Character Class Demographics

Exploring the RL demographic and motivational differences in terms of character class selection has typically been difficult due to within-class variations among contemporary MMOs. For example, there is no easy way to classify character classes in Star Wars Galaxies or City of Heroes due to the extensive class combination possibilities. WoW, on the other hand, provides a clean class categorization. While there are still within-class variations, they are minimal compared with SWG/CoH variations.

From the survey data, there are significant age differences among players who prefer different character classes. Players who prefer Rogues and Shamans tend to be younger than players who prefer Warlocks and Hunters.

Female players are more likely to prefer Priests, Hunters, and Druids, while male players are more likely to prefer Rogues, Warriors, and Shamans.

An overview of the motivational difference is presented here. Only the motivations where an ANOVA showed significant differences among the character classes are listed. The detailed charts for each motivation follow. The motivational differences among the different character classes highlights an interesting potential effect of balancing classes. Given that classes do not appeal to players equally, nerfing or strengthening a class would have the effect of driving away or attracting players with different motivations.

Click here for an overview of the motivations framework.


====


 

Posted by nyee at 4:20 PM | Comments (37) | TrackBack

WoW Alliance vs. Horde

Overall, the demographic differences between Horde and Alliance characters are minimal. There are no gender differences. Male and female players are equally represented on both sides. There is a statistically significant, but substantively trivial, age difference. Players who prefer Horde characters (M = 27.5, SD = 8.0) tend to be slightly younger than players who prefer Alliance characters (M = 28.7, SD = 8.6).

Players who prefer the Horde score significantly higher on the Advancement, Competition, and Mechanics motivations than players who prefer the Alliance. In other words, players who prefer the Horde tend to be more achievement-driven, more likely to enjoy provoking and challenging each other, and more likely to be min-maxers. On the other hand, players who prefer the Alliance tend to score higher on the Role-Playing and Customization motivations than those players who prefer the Horde.

Click here for an overview of the motivations framework.

Posted by nyee at 4:18 PM | Comments (107) | TrackBack

February 21, 2004

Who Wants To Be a Jack-of-all-Trades?

In many class-based MMORPGs, class options include pure classes and hybrid classes. Pure classes excel at what they do, but have a more limited ranged of abilities, while hybrid classes have a larger range of abilities, but have limited expertise in them. The data presented here would also apply to skill-based systems because all these systems have a skill cap and players must decide whether to excel in one skill line or spread out, unless the system mechanics make one or the other a better option (like in DAOC).

Both gender and age were found to impact the preference for pure or hybrid classes. Female players tend to prefer hybrid classes. Also, older players prefer hybrid classes.

Among the motivation components and personality traits, the motivations “compete” (t = -10.00, p < .001, r = .18) and “achieve” (t = -7.36, p < .001, r =.13) were the best discriminators for players who preferred to focus rather than spread.

Posted by nyee at 12:15 AM | Comments (10) | TrackBack

Resource/Manufacturing/Combat Classes

Many recent MMORPGs have well-differentiated non-combat professions that are viable in and of themselves. In other words, unlike the EQ system, you don’t have to fight to get the resources or money (or rely on your guild) to buy the materials to skill up in crafting. The appeal of the resource and manufacturing classes were explored in comparison with the combat classes. Respondents rated the appeal of the following three broad classes:

If the following 3 types of roles were equally well-implemented and balanced, how much would each appeal to you?
Resource Gathering: Explorer/Geologist/Miner/Hunter/Trapper
Manufacturing: Blacksmith/Alchemist/Armorer/Tailor
Combat: Warrior/Archer/Healer/Sorcerer

Since most MMORPGs tend to be combat-focused, it was expected that the combat rating would be skewed higher. The goal was to explore whether particular slices of age, gender, or motivations would provide a reasonably good understanding of whether certain players were more likely to prefer certain non-combat professions.


The results seem to indicate that preference for these three board types of classes do not vary with gender or age very much, but may be better explored in terms of player motivations.

Posted by nyee at 12:05 AM | Comments (2) | TrackBack

Through The Looking-Glass

This dataset explores several aspects of identity projection through the use of avatars. When players create and play their characters, how many of them choose to act and behave pretty much the same as they do in real life, and how many deliberately try out new personalities? And do players tend to behave and act the same way across their different characters? Are there gender or age differences? And finally, do motivations for playing or personality traits have an influence on how identity is projected through avatars?

Female players are more likely to behave and interact in an MMORPG very similarly to how they behave and interact with others in real life when compared with male players.

Age, however, seems to be the more important factor.


The multiple regression results show that Introverts are more likely to behave similarly while Extraverts tend to behave more differently. Also note that Age and Extraversion do not correlate (r = -.02).

====


Players were also asked how consistently they act and behave when they are playing different characters. Gender and age had no impact on the outcome of the responses.

The multiple regression results show that it is the desire for immersion and role-playing that most differentiates players who behave consistently across their characters from the players who don’t.

====

Finally, respondents were asked whether they felt that they were more of “who you really are” in the MMORPG than in real life. There were no significant gender differences.

The multiple regression results echo the results from the questions on behavioral similarity, and show that Introverts are creating characters not only similar to themselves in real life, but that they act and behave in a way that they feel is more true of who they really are.

Posted by nyee at 12:01 AM | Comments (10) | TrackBack

September 3, 2003

The Demographics of Gender-Bending

Previous findings had shown that male players were significantly more likely to play characters of the opposite gender, but the age differences among male and female players was never explored in detail. It was surprising to find that older male players are the most likely players to gender-bend.

However, there was a possible confound in that older male players were also found to have more characters in general. This was not the case with female players.

====

And so instead of relying solely on the absolute number of characters of the opposite gender, another analysis was performed that used the proportion of characters of the opposite gender in terms of total characters. And this showed that older male players were indeed much more likely to have characters of the opposite gender.

Theories for male gender-bending I have explored in the past have included:

-Social gender boundaries more stringent in real-life for men, and in an anonymous space, men are more likely to explore gender roles.
-Female avatars receive more “freebies” and are treated “better” than male avatars.
-Male gender-bending is another form of dominating the female body.
-In PvP games, female avatars are perceived to be weaker and this might give the player a psychological edge against unsuspecting chauvinistic players.

None of these theories, however, explain why older men would be most likely to gender-bend. Any ideas?

Posted by nyee at 9:47 PM | Comments (77) | TrackBack

May 2, 2003

The Demographics of Character Creation

MMORPGs are designed to elicit personal and emotional investment. On a superficial level, we can point to the time investment that many MMORPG players put into these immersive worlds. On a deeper level, we can point to how players derive tremendous satisfaction and infuriation from their character’s successes and failures. And because players are personally invested in the character and the world, every decision they make within the construct becomes personally revealing.

Let’s consider a fairly crude example of getting at the complexity of projection. Most games make the character creation process as elaborate and fun as possible – an attempt to create a sense of uniqueness and individuality that is entirely yours. In a recent survey, respondents were asked to select the attributes they favored most in character creation among 4 choices. The gender differences were significant and for the most part expected.

Male players favor physical attributes that have an effect in physical combat, while female players tend to favor mental attributes that contribute to support spells and non-physical combat. Looking at the age differences, we find that younger players tend to prefer the STR/END and DEX/AGI while older players favor WIS/INT.

This has fairly interesting implications in terms of the individuals that make up a group in the game. The data suggests that the primary and secondary tank classes tend to be composed of younger players, while the offensive and support spell-casters tend to be composed of older players.

But the data on age differences is itself interesting. Notice that there is not much differentiation among attribute preference among players between the ages of 12-17, and that the differentiation occurs slowly after this point – the preference for STR/END and AGI/DEX decreases while the preference for WIS/INT increases.

As MMORPGs become more sophisticated and allow more character customization and in-game social decisions, the ability for these environments to elicit personally revealing information increases. As opposed to traditional techniques for projective personality testing (like the Rorschach inkblot test), the MMORPG also gives us an existing computerized framework for easily collecting this data. One could easily imagine tapping MMORPGs for personality assessment and screening techniques in the future.

Posted by nyee at 5:29 PM | Comments (18) | TrackBack

March 27, 2003

Identity Projection

Players think of and relate to their avatars in very different ways. Some choose to identify and personify their avatar with their own personality, while others objectify their avatar and see it as a pawn in an abstracted playing field. As the following graph shows, female players are more likely to see their characters as idealized version of themselves, and age has a greater effect on male than on female players.

Players who see their characters as idealized versions of themselves do not spend more time playing the game, but they do indicate a greater willingness to stay with the game indefinitely than players who disagreed with that statement, probably due to personal and emotional investment in their characters.

====

Players were also asked to indicate how much they agreed with the statement, “I think of my avatars as a part of an extension of me.” While there were no significant gender differences or age differences, there was a very clean positive correlation with hours played per week.

There was also a very clear positive correlation with indicated willingness to stay with the game indefinitely.

====

Finally, respondents were asked to indicate their agreement with the statement, “I think of my avatars as just pawns in a game.”

In full complement with the two data sets presented above, there was a negative correlation with both hours played per week and projected willingness to stay with the game indefinitely.

Together, these three data sets show how identification with the avatar affects game play per week and projected willingness to stay with the game, probably as a function of personal and emotional investment. Perhaps agreement with identity projection statements could be thought of a measure for emotional and personal investment into the game.

Posted by nyee at 5:34 PM | Comments (6) | TrackBack

January 1, 2003

Dr. Jekyll and Mr. Hyde

While most people might think that people are more likely to lie and misbehave online, the opposite is probably more true. As the following graph shows, about one-third of players feel that they act more benevolent online than they do in real-life.

It is also the case that about one-third of players feel they are more aggressive and initiative online.

The internet does indeed allow people to feel less inhibited, but as this pair of graphs suggest, the internet doesn’t turn people into pathological liars and thieves, but rather, courageous knights and brave warriors whose motives are benevolent.

Posted by nyee at 7:53 AM | Comments (19) | TrackBack