Very interesting post -- just as a small note, I'm rather surprised to see that the mean grouping ratio for priests is only ~0.35, given that healers are the archetypical grouping character (rather difficult to solo a priest).

Given that priests are (by design) the most 'social' of the player classes, it'd be pretty cool to see a plot of mean grouping ratio vs. player level, just for the priest category -- I'm guessing that the mean grouping ratio for higher-level priests is *much* higher than 0.35.

Shouldn't you be sampling from a pvp server aswell? esp, for this kind of test

Good point, Gorman - we are in the process of adding a fourth data collection machine that will focus on a PvP realm.

Are you able to make any sense of group-size?

I expect it follows the same pattern (the levels and classes that group less often trend to group smaller when they do group; levels and classes that group more often, trend toward fuller groups), but it'd be interesting if that weren't so, or if there were an oddball class or level range that did the opposite.

Hey, just came over here from Terra Nova. Great site. I look forward to many more interpretations of the data.

I'm wondering if any attempt is being made to look at the psychographic preferences of online worlds. Our firm has done some research along this line, especially when it comes to gamers. You can read more of our research here at our blog: http://www.nslg.net/blog

Jeff - Good point. Unforunately, we can't get information on group size since we don't have access to it client-side. We do have guild size data though that we will release soon that does conform to your expectation.

just wanted to add that is actually very easy to solo as a priest, in contrast to what the first comment says. priests are one of the strongest classes you can play. they're not weak healer types that you'd get in any other game. they can output high damage, and couple that with their survivability(healing and shielding mainly) they can usually survive most any reasonable encounter, and even some you'd think they shouldn't be able to.

"players who choose to be warlocks are probably different from players who choose to be priests"

I find that sort of comment quite interesting, as my two level 60 characters are a warlock and a priest (with the former being the one I view as my primary character). When selecting a second class to play, I intentionally chose something quite different from my warlock, in order to maximize the novelty in my play experience.

My priest is specifically tuned to be a support mechanism for groups, as opposed to a strong soloer or PvP combatant. Despite this, I find myself playing my warlock exclusively in end-game raids. Part of this is because, while I enjoy playing my priest, a warlock in a raid has a very low-stress job. Success or failure of the group as a whole is not dependent on your split-second reactions, unlike a priest. A contributing factor is the relative class distribution. As uncommon and valuable as priests are, warlocks are even more so, especially in the Alliance. I have been the sole warlock in a 40-person Molten Core raid, which is quite beneficial to me from an item-acquisition perspective, even should my play style preferences not be an issue.

I have a 60 hunter and find it to be not entirely true that hunters/warlocks find it less likely to be grouped.

There are some classes who can potentially be more harmful to a group if they are played by idiots who have no idea what they are doing. A hunter with a complete lack of control over their pet can wipe a group fast. Warlocks, also a pet class, have the same problem.

Also most people, due to the fact that hunters were such a late class in beta, never learned how best to utilize a hunter. For instance... most groups want a warrior for a puller, but a hunter is a far better puller and has far more pulling skills to utilize than a warrior will ever have. The usual mindset is.. stick the guy with all the hps out front...when they should look at the hunter and see Feign Dead, traps, extra long ranged skills and realize that the hunter can pull more efficiently and with more control.

Still, even with that lack of understanding, I find that the reason I don't get into a group isnt because of the class, but because of the sheer number of that class around. A group of 10 doesn't need 3-4 of them to be hunters. It's not particularly well-balanced, but then again they don't need 3-4 priests either. It's about forming a well balanced group that covers all the bases you need for a quick, successful run.

My main is a level 60 hunter. He was very good at leveling solo so I did a lot of solo leveling with him.

I enjoy playing him in PvP however and NEVER wanted to PvP solo. He started to PvP a lot at level 30 so he was usually solo for leveling and usually raid grouped for PvP. I would usually level in the afternoon and PvP at night.

Now that he has been level 60 for some months I rarely play him EXCEPT for high level instance guild raids or PvP raids in the BGs.

I am a social type of person in real life AND in WoW, however the game encourages me to play solo for leveling and to group for instances and PvP. When I am soloing my characters I am STILL being social in guild chat. The data available to you probably doesnt show this type of social interaction.

I think the first graph is quite interesting; the proportion of time spent in a grou pis highly reflective of how much better off each class is in a group- the three main healing classes are grouped the most often, which makes sense as their abilities to heal in a group are their most effective and efficient abilities. On the other end, locks and hunters, with their pets, are sort of mini-groups on their own, and I don't think their efficiency improves much if at all by grouping. As for the gouping vs. level, this obviously shows the relative amount of content meant for groups at each level- the small spike around 20 from people trying out the first instances, and in the 40ish range (imo) from scarlet monestary, especially fo rthe horde- that particular instance seemed to be the most important one before Blackrock Mountain and the end-game stuff.