I suspect most of us are a combination of those codes. For example, I usually try to come up with a character concept first, then find the class/race that matches that concept best -- although the concept may be informed by the game mechanics, too. It's a very organic process for me.

When I rolled a Troll Shaman in wow, it was because it was the rarest class/race combination in the game (as it still is). When I started WoW the first time (before a 2 year hiatus) I rolled a Druid, because they were the rarest (and most commonly considered underpowered) class in the game.

I suppose the reasoning is something like; if they are rare and considered weak, if I do well it is because of skill not inherent unbalance.

Also its nice to be unique.

The Herald of Xotli is not a ranged class. *headshake*

Anyway. How did you treat the comments of those people who would be a combination of these options? Did you go by what seemed to be the strongest influence, or did you add them to multiple categories?

Good question, Ten. In cases where multiple categories were listed, we looked at whether there was one that the respondent emphasized (most of the cases). If it seemed all their listed categories were equal, we coded it as the first category in their response.

When I first started EQ, I picked characters that would start in Kelethin (a tree city with elevators) and easier characters (warriors, rangers, paladins), then I wanted a druid (Spirtit of Wolf, healing, buffs) for group play, then I hit some dark times and created a dark elf necromancer, a troll shadowknight and a wizard (to nuke). I roleplay a lot and have enjoyed all of these characters. EQ is great for allowing 8 characters per server, allowing me to try almost all races and classes.

Nick. I think you decision not to pull out multiple influences skews your results because fails to distinguish between those individuals that weight one factor highly and those that equate multiple factors equally. I certainly fall in the latter group. Why I chose to play a particular character varies from game to game, my mood, the alignment of the stars, the research I have done of classes and races, and what I had for dinner the night before. For me, the character creation process it by definition complex. While it may be helpful to try and tease out what is "really important" the respondents should have done that themselves using some type of Likert scale. For you to do it is capricious.

Daniel - Without a qualitative exploration, it's not at all obvious what items should be used in a multiple-choice survey for this question. And while I agree with you that this is a complex process, I would argue that starting with a close-ended survey format would constrain what you would learn.

There are seldom methods that can address all questions at once, and to suggest that this data set is somehow bad because it doesn't address one particular question is to miss the point that there are many different questions one can ask within a research area.

Had we gone ahead with a multiple-choice survey, others would have argued that I missed potential character creation reasons by not asking it in an open-ended format.

Good survey Nick. I usually pick the race first i want to play, than the class. I will be using this character alot and i dont want him to look anything less than the image i imagined.