New Daedalus: Demographics of WoW Players

A new issue of the Daedalus Project is out. Most of the new data focuses on the demographics of WoW players, emphasizing the relationship between in-game class/race/gender and RL gender/age/motivations.

Some highlights:

  1. Female players are older than male players. (link)
  2. Younger players prefer Rogues and Shaman. Older players prefer Hunters and Warlocks. Rogues and Shaman also score the highest on the Advancement (goals/achievement) and Mechanics (min-maxing) motivations.  (link)
  3. Older players prefer Dwarves and Gnomes, who also happen to score the lowest on all achievement motivations. Gnomes score the highest on the  Role-Playing and Customization motivations. (link)
  4. The RL gender distribution is 84/16. The in-game gender distribution is 65/35. 55% of female characters in the game are being played by men. (link)

I'm not as interested in WoW in particular as much as using WoW as an example of showing how RL demographics and personality factors map onto game variables - such as race. The WoW data provides one cohesive example of that.


Comments on New Daedalus: Demographics of WoW Players:

Tess says:

It would be interesting to compare the race and class motivation alignments with the race-class combination statistics. For example, I recall that in spite of the popularity of rogues, Gnome rogues are rather rare. This could be due to the fact that Gnomes and rogues show substantially different motivational patterns.

Posted Aug 1, 2005 5:36:45 AM | link

Edward Castronova says:

Interesting that older players play the solo characters - warlocks and hunters. Perhaps much of the hardcore/casual and solo/group debate among players is reflective of underlying differences in life situation more than anything else.

Posted Aug 1, 2005 6:00:20 AM | link

Lee Delarm says:

As a younger player turning into an older player I can say that I feel a rift between not only skill levels, but language barriers and interests.

The people who are older don't tend to get as much time online (I know I used to get LOTS) because of work, family and other commitments. Younger players get much longer stretches of time (especially during summer) to play to their hearts content with other worries or distractions.

The language barrier seems to be more of an up and coming type of thing, where the younger they are, the more "kthnx" or "u" or "wut" I see. This type of language totally turns me off to grouping, since nine out of ten people in a group (pick up group) will be using these phrases casually expecting me to decipher what they're spewing. Maybe this is partially just a skill issue and they haven't been taught proper typing but I'm seeing it to the level where I think they're just so lazy they refuse to type out the two extra letters per word (I still type 100 times faster than them somehow).

Lastly, I tend to get fed up reasonably fast when in a group that doesn't know what they're doing or is only interested in themselves. I've been in a LOT of groups (other games included) where the players will join and go afk almost immediately for 10 minutes or more. Other times the player will haphazardly agro monsters that we can't handle or will agro more monsters "on accident" (multiple times) because they're standing too close to the spawning areas. Myself not having the amount of time they do to devote countless hours to getting my corpse back 5 times in one night, I prefer to hunt "safer"...i.e. ALONE.

Posted Aug 1, 2005 6:36:04 AM | link

Scott M. says:

I sympathise, Lee. It really is very frustrating to get into a group with someone who seemingly lacks the basic rules for successful grouping. I just don't have the patience for it any more. I really wish an MMO would create a server or two for mature players; granted age is not necessarily a measure of maturity, but it's probably the best we've got. Maybe if they called is a "senior citizen" server or something it would discourage a certain bread of lazy, immature player.

Posted Aug 1, 2005 10:15:58 AM | link

Bart Stewart says:

Lee Delarm> As a younger player turning into an older player I can say that I feel a rift between not only skill levels, but language barriers and interests.
> I prefer to hunt "safer"...i.e. ALONE.

Likewise.

I have problems with grouping because my needs don't seem to be the same as what others in most groups are looking for. I group for security, and as payment for that security I try to help assure the security of others in the group. I'm willing to accept less loot over the short term (because loot is shared among the group's members) in return for more loot over the long term (because the group as a whole is more secure).

But too many times, people I've grouped with seem to have a kind of personal obliviousness to anyone else's interests or concerns, and have zero interest in anything longer-term than the current play session (if that). Grouping for them is just a mechanism for using other people to help them quickly score more high-level loot than they could get on their own.

Too many experiences like that are why I solo. Why should I (as a character) risk my life for my groupmates if it doesn't even occur to them to do the same for me? I wouldn't mind risking my life for the group if the other members of the group made the same commitment, but such players are so uncommon that it's no longer worth spending time looking for them. So I solo.

Which makes high-end content off-limits to me, meaning that I'm basically paying for features I'll never get to enjoy. And that's fairly annoying.

Why do designers treat grouping as a simple gating mechanism for high-end content? When that leads to people grouping who have no interest in promoting the group's welfare, isn't that a flawed design?

Are there ways to implement grouping that would motivate players to play as though the needs of the other group members were as valuable as their own?

Or am I the only one who'd play such a game?

--Bart

Posted Aug 1, 2005 1:25:38 PM | link

Jimpy says:

I'm in the same demographic as many of the above posters (older player/limited playing time) but have had a different experience.

I've found the combination of the guild function and the friends function can act as a fine filter. I've found myself in a guild that is basically full of married people. Between 9:00 p.m. and midnight, pretty much everyone online is a 30-something. I supplement that by putting anyone mature I meet on my "friends" list and, all told, I actually feel like a have a huge pool of possible playmates to draw from.

I agree that, for the "mature" player, pick-up groups are not a great gamble. But I'm curious how many older gamers make use of the guild and friend tools?

As for grouping being a gateway to higher level content, I think that might just be the nature of the "massively multiplayer" part of these games. If you *didn't* have to group to reach endgame content, I'm not sure you'd have an MMORPG anymore.

Posted Aug 1, 2005 1:52:39 PM | link

Thabor says:

Interesting to me that Rogues scored lowest on the Discovery motivation, when they have stealth based classes have a much greater potential for exploration.

I'm also a bit curious how Rogue ended up listed as both high and low motivation for Customization.


Interesting that older players play the solo characters

According to Nick's grouping ratio chart from the census data shaman is still on the lower end for grouping, and rogue is in the middle..

From my own experience I would consider most classes in WoW to be "solo" classes. Priest, Druid, and Mage being the more group motivated classes. Might be interesting to see how the the grouping ratio compares to the ratio of standard quests to elite/dungeon quests broken down by level range..

Posted Aug 1, 2005 3:01:57 PM | link

Nick Yee says:

Thabor - Thanks for pointing out typo. It should be Priests and Warlocks on high Customization.

Posted Aug 1, 2005 4:01:55 PM | link

Tess says:

Thabor said:
"Interesting to me that Rogues scored lowest on the Discovery motivation, when they have stealth based classes have a much greater potential for exploration."

I was totally puzzled by the exact same thing. Discovery is exactly the reason I play a rogue. I must be weird, or something.

Posted Aug 1, 2005 9:22:14 PM | link

Michael Hartman says:

I hate to ask such a boring question, but how is the sample created for this study. I read the article and all I saw was that his sample size was 1,019 people.

Posted Aug 2, 2005 2:40:05 AM | link

Lanky says:

Sir Bart said:

"Why do designers treat grouping as a simple gating mechanism for high-end content? When that leads to people grouping who have no interest in promoting the group's welfare, isn't that a flawed design?

Are there ways to implement grouping that would motivate players to play as though the needs of the other group members were as valuable as their own?

Or am I the only one who'd play such a game?"


Perhaps it is not a design flaw per se, but rather a common player flaw. I imagine there is a general ascending correlation between the age of a given player and his/her willingness to assist his/her group to succeed.

Im sure a similar correlation exists for other statistics, like level of education, et cetera.

The reason I postulate this is because several criteria, like age, or education, make it "easier" to understand that by helping your group, you help yourself. As long as WoW group challenges are designed to require or encourage that kind of play to succeed, then they are successfully designed in my humble opinion.

Also, the main reason I feel I can even put forward this position is that I play WoW as a level 60 mage in a guild that regularly defeats the hardest end-game content available. I believe that without a certain level of co-operation and organization and group comprehension, said content would be impossible.

Lanky

Posted Aug 2, 2005 10:26:59 AM | link

Nick Yee says:

Michael - Like all my surveys, the sample is a self-selected sample from MMO forums, news sites, message boards, and word-of-mouth. Naturally, self-selected samples are susceptible to bias, however, the survey sample does closely match server-side census data on several key variables.

Posted Aug 2, 2005 12:17:40 PM | link

Michael Hartman says:

Thanks Nick.

Given the un-scientific nature of the sample selection, what impact do you think this has on the research?

What effect does this have, in your opinion, on the conclusions one can draw from the data?

Have you made any efforts to figure out where the nature of the self-selected sample would inherently effect the data?

For example, forums are notoriously populated by the hardest of the hard core. Also, anyone willing to respond to a survey about WoW is going to tend towards being more hard core about the game. Would this fact tend to depress the numbers you find for older gamers, female gamers, or anything like that?

Posted Aug 2, 2005 12:55:08 PM | link

Nick Yee says:

Michael - It depends on what conclusions you are trying to draw. I focus on secondary relationships rather than primary frequencies. It's much less interesting to me that 16% of the population is female than that female players in general score lower on certain motivations and prefer certain classes. It is possible to spot gender differences (for example) even if the overall sampled gender ratio is skewed.

And the same thing is true for a skew in overall playing time. Say I have a higher ratio of high achievers in the sample, as long as I have some low achievers in the sample, the differences between those two groups probably reflect actual differences.

Here's a straight-forward example. The gender ratio in the real world is about 50/50. Say I get the average height of 1500 men and 200 women and I find from that sample that men on average are taller than women. Is it wrong to generalize because my sample is unrepresentative of the overall distribution or does it actually not matter?

And finally, the survey data dove-tails with the near-perfect server-side census data from the PlayOn Blog in several instances. The achievement scores of character classes (from the survey data) closely matches the leveling times of those classes (from the server-side data). Again, I'm not saying the survey data is perfect, but it does match up well with several key variables from the "scientific" server-side data.

Posted Aug 2, 2005 2:58:38 PM | link

Thabor says:

Here's a straight-forward example. The gender ratio in the real world is about 50/50. Say I get the average height of 1500 men and 200 women and I find from that sample that men on average are taller than women. Is it wrong to generalize because my sample is unrepresentative of the overall distribution or does it actually not matter?

I'll say that is wrong to generalize in that case. Or at least that it is wrong to be assertive about that generalization. In cases like secondary relationships they may stand a good chance of being correct, but they should always be supported by additional research when your inital sampling isn't representative.

Does it still not matter if your sample has indicated women on average were taller than men?


Posted Aug 2, 2005 6:42:25 PM | link

Michael Hartman says:

Thanks for the explanation Nick.

Posted Aug 2, 2005 7:27:16 PM | link

Nick Yee says:

Thabor - Again, like I've said twice already here in these comments, that's why it's cool that we have the PlayOn census data to reflect/complement/reconcile with.

I am trying to triangulate and not just standing on the survey data alone. It IS being supported by additional research using an entirely different methodology.

Please don't make me say it again.

Posted Aug 3, 2005 12:10:19 PM | link

Thabor says:

Nick,

I'm not trying to pick on you. The census data is valuable, and the survey data is also valuable. I have a very low tolerance though for generalizations being presented as fact, when they may or may not be supported. I'm not saying you are wrong, just asking you to be as precise in presting things in english as you would be presenting the mathamatic analysis.


I am trying to triangulate and not just standing on the survey data alone. It IS being supported by additional research using an entirely different methodology.

I have no doubt you put great effort into correlating your assertions. However not all of the survey data has any direct relationship to the cesus data you have so far.

A few specific cases:

Highlight #1: The census data doesn't tell us anything about age. This could use additional support, but this is a difficult one to get valid data on. The best you could probably do would be a slightly larger non-self selcting survey.

Highlight #2: Same as #1 for the age portions. The class to advancement correlation is most interesting since you have the the time to level data from the census to support it. The mechanics motivation could use additional support. Equipment and character build information might be able to better support this. Advancement and Mechanics almost certainly show some correlation. Which direction is it? IE: Do players select Rogues and Shamans because they are "better".

Highlight #3: Same as #1 for the age portion. The Advancement motivation could probably be supported by the census data. I haven't noticed a breakdown of time to level by race. Customization could probably be supported by additional information on equipment build as well.. IE( Low variance of equipment builds might inditance the Mechanics (min/max) while high variance might indicate Customization ).

Highlight #4: I don't know if your census information includes character sex or not. If it does you could support this by working backwards to see how close you come to the RL gender distribution.

Posted Aug 3, 2005 4:40:07 PM | link

Jim says:

Very interesting stats! I am 38years old and a hunter in bleeding hollow (horde).

I would love to play with all u guys and look me up if u come to my realm... (Gunnerx)

I agree with lots of the posts above...selfish people show their immaturity when u party with them. They will grab your loot over and over, open chests without rolling for them, and and as soon as they get the rare loot they came for they will drop from the party and hearth home.
grrr..
we need an old fart's realm.

p.s. I didnt know there were that many chicks playing wow... 45% of female characters are girls? I always thought they must be wowmo's...lol

Posted Apr 26, 2006 10:42:56 PM | link

Juan Incognito says:

I know a few of the people I play WOW with in real life (roughly 12, so not much of a sample size:(), and to be honest I can't say that your findings reflect what I can verify.

I would note the following: People with multiple 60 (or high level characters) usually have a variety of races represented. More often than not the first race chose was Night Elf. Most of the women I know who play tend to choose female gnomes as a primary character. I've found that a better determinent of RL sex than any other, go figure.

Posted Apr 27, 2006 12:45:18 AM | link