NAVIGATION:

 

 

Welcome to PlayOn 2.0

July 23, 2010 |

PlayOn 2.0 is a new longitudinal study of WoW players led by Nick Yee and Nic Ducheneaut at the Palo Alto Research Center (PARC). PlayOn 2.0 links self-report data (i.e., demographics and personality variables) with server-side behavioral data (i.e., from in-game and the Armory). Much of the current study expands on the infrastructure and lessons learned from the original PlayOn study where we used automated in-game census bots to track characters across 5 WoW realms.

Expanding on the Original PlayOn Project

While the original PlayOn study provided many unique insights into the social and game-play dynamics of WoW, one major drawback was our inability to link in-game metrics to real world (RW) variables such as player gender or age. More importantly, by only observing in-game data, we could only analyze data at the character level rather than at the player level. Thus, a player with many alts may be counted several times compared with players who play only one character.

One key feature of PlayOn 2.0 is in generating a data set that directly links RW variables with in-game variables. Another key feature is the longitudinal data collection allowing us to observe long-term changes in game-play. The third key feature is the cross-cultural comparison—in the current phase of the study, we are running data collection in parallel in the US, Hong Kong (HK), and Taiwan (TW). These three features allow us to explore many questions of interest to MMO players, and we hope to publish many of these findings here on this blog.

Study Design

Before we start, I want to briefly describe the study methods so readers have a better sense of the data we’re gathering:

  • We began the study by using a web survey as a recruitment tool, and also to gather basic demographic and personality variables. Volunteer participants were also asked to list up to 6 characters they are actively playing.
  • Once these characters were in our database, two automated data collection scripts activated.
  • The first script is an updated version of the PlayOn census bot. This bot cycles through the characters, and if they are currently online, will do a /who on them (gathering zone, level, and guild info) as well as a /who on any of their active guild members if the participant belongs to a guild. The collection interval for the current bot is roughly 45 minutes (i.e., to cycle back to the same character again).
  • The second script is an Armory scraper that gathers character profiles from the Armory. The Armory updates once a day if the character logged on the previous day. Thus, our script follows this schedule with a daily interval.
  • Once a month, participants are asked to fill out a short follow-up survey to input any new active characters they have recently created or brought out from hibernation.

During the recruitment phase, we recruited roughly 1950 US participants, 357 TW participants, and 134 HK participants. Due to infrastructure constraints, we randomly sampled from the participants to create a core data set based on roughly 500 US participants and 500 TW+HK participants. In short, while we have survey data from all participants, we are only actively collecting in-game and Armory data from about 1000 participants. To reach this 1000 goal, we over-sampled at 1100 to avoid potential losses due to missing variables, etc. Thus, for most of our posts, you’ll see us refer to about 1040 participants in the core subset.

One other constraint we had was that IRB (institutional review board) regulations did not allow us to collect data from minors. While this means we weren’t able to gather data from the below 18 crowd, we still were able to gather data from a very wide age range (18-65), and age correlations in the data set will likely provide useful age-related information.

Our Plans for the Blog

We plan on releasing 1-2 new blog posts every week. In the queue, we have posts related to PvE vs PvP preference, equipment level, the number of hugs given, tank/DPS/heal preference, and much much more. So you should go ahead and bookmark the site now.

More so than data sets we’ve explored in the past (i.e., original PlayOn Project or the Daedalus Project), there are many ways to slice and dice the current data set. Whenever we present one finding, it will be the case that we could have sliced it instead by region, by gender, or by age. Or there’s another way we could have looked at it. And another variable that is related to it, etc.. As we slowly but surely churn through the data, we ask ahead of time for your patience in that it may take us a while to answer specific questions, but the accumulation of blog posts here will hopefully provide a multi-faceted lens over time. And as much as possible, we will tag each blog post with the precise slice of data we used for the analysis.

Bookmark and Share
read more
comments (17)
posted in Announcements, Study Methods

 

Comments (17)

July 26th, 2010 at 8:43 pm Posted by Bonnie

Thank you for the fine explanation of the purpose and methods you used in this study. I participated in it in game as a WoW gamer and am very interested in reading your research findings and the discussions that ensue from that.

 

July 26th, 2010 at 8:49 pm Posted by Tyler

First, I am very glad to see that the PlayOn project is back in action. Congratulations on the new grant.

Second, might there be some way to get a RSS feed for the PlayOn 2.0 blog? All I can find is a RSS feed for the general PARC blog and it would be nice to have an additional feed specific to the PlayOn 2.0 project.

 

July 26th, 2010 at 9:37 pm Posted by Nick Yee

@Tyler Thanks for asking. Use this RSS: http://blogs.parc.com/playon/feed/rss/

 

July 27th, 2010 at 10:24 am Posted by Brian

Awesome, glad to see this in action again. I’ve been visiting Daedelus since I was playing EverQuest, so I’m always happy to see more things from you, Nick.

I’ll see if I can get the RSS feed on our website once I contact one of our web admins.

Hugs oughta be interesting, heh. It’s my personal second most used emote :x

 

July 27th, 2010 at 10:24 am Posted by Brian

And that’s totally not the smiley I was going for. Alas!

 

August 22nd, 2010 at 7:13 am Posted by Isaac

Your link to the old PlayOn project is dead. Could you put a link to the archive on your homepage somewhere?

 

August 23rd, 2010 at 10:51 am Posted by Nick Yee

@Isaac Sorry about that. Just fixed it.

 

August 26th, 2010 at 2:21 pm Posted by Lani Teshima

Thanks for the blog! I’ve added it to my Google Reader feed.

btw, Dmitri Williams says hi (I’m in his guild).

 

August 26th, 2010 at 2:58 pm Posted by Zek

Hmmm seems interesting, will definitely keep an eye out on this study. I use to spend a lot of time on mmo’s myself but i quit because it got rather repetitive and because it’s time consuming.

 

August 27th, 2010 at 7:11 pm Posted by Lindax

I’m still having some trouble with the sampling. It’s pretty clear what you did with the group you gathered up, but not how you got them in the first place. Given that a large part of your analyses rely on self-report data, I’m really curious about who chose to participate, who your recruitment was able to include and who it could not or did not (besides the minors). I’ve never seen any announcements anywhere recruiting players to the study (e.g., elistist jerks, wowhead, thottbot, wowwiki, Blizzard sites, etc.)… Can you say more about your recruitment process and how it might have influenced findings about, e.g., achievement.

 

August 28th, 2010 at 2:40 am Posted by Nick Yee

@Lindax Our biggest community announcement happened on WoW.com: http://www.wow.com/2010/03/17/gnome-rogue-with-pink-mohawk-wants-your-data/

Unfortunately, sites like thottbot, wowwiki, and Blizzard-owned sites do not publicize academic studies (and thottbot doesn’t even have a news list to begin with), but WoW.com is probably one of the top WoW-related news sites out there.

On the other hand, one good way of ascertaining any potential biases is by comparing our data to sites like warcraft realms that have WoW-wide census data. I’m not sure anything is directly comparable right now, but we will soon have data on class/race distributions and BG/Arena win ratios.

 

October 7th, 2010 at 6:32 am Posted by Mirko Suznjevic

Quote:
“While the original PlayOn study provided many unique insights into the social and game-play dynamics of WoW, one major drawback was our inability to link in-game metrics to real world (RW) variables such as player gender or age. More importantly, by only observing in-game data, we could only analyze data at the character level rather than at the player level. ”

I have solved this problem in my research by implementing an add-on which gathered data on player’s computers. I have gathered data about player behavior in terms of events generated by WoW API. In this way i can see how the player behaves not just the character.

I plan to monitor the player behavior further and to see how it changes during the lifespan of the game (how major expansions affect behavior and so on).

Downpoint of my research is small number of participants (i got 104 players for my last round).

If you would be interested in monitoring acctuall in-game behavior of the players throguh an add-on and connecting it to the real life traits of you participants let me know :) This is far more precise than just monitoring server statistics. I am planing to refresh the add-on for the new espansion and new round of my research anyhow.

And ofcourse i wish you good luck with the new round of your research will follow it for sure :)

 

November 4th, 2010 at 7:34 am Posted by Eridani

Hi!
Gosh…I have been “studing” video games social roles, behavior and communication (or interaction) for over a year and a half. Tought, it’s not until now when I will be able to publish something…
I was wondering. What do I need in order to be part of your group? I’m really interested in making a career as a research scientist.

 

December 4th, 2010 at 1:39 am Posted by Pep

Greetings,

I must say that so far I found the results quite interesting.
As a researcher (on a diferent area) and wow player, I came across in my mind of some of the questions you providing answers here.
I would like to see this data analysed in a multifactorial way. An idea could be to try and identify diferent profiles of wow players, how the variables themselves relate.
I haven’t see your questionaire, but one big discussion you see arround too is how much the game has a negative impact in real life. I honestly think is not a simple explanation but rather a given profile of gaming, related to idividual characteristics that ends up to have a negative impact. Do you think that in any way your study can provide answers to this question and eventualy help others to promote “healthy gamming pratices”? (I know it sounds scary, but is something we try to promote in our guild for example).

 

February 21st, 2011 at 6:41 am Posted by TheOtherWinter

Nic and Nick, when you enlisted folks like me to participate, you asked for access to our social networks like Facebook. I’m curious what you’ve done or plan to do with that information, since none of your results mentioned here have addressed that cross-over data.

 

February 22nd, 2011 at 2:34 pm Posted by Nick Yee

@TheOtherWinter: Two things hampered our social network data collection unfortunately. A lot of participants opted out of installing the FaceBook app. And due to IRB restrictions, we could not gather any data that was related to the participant’s friends (because those friends did not consent to study). Because of these two things, we haven’t prioritized analyzing that data. We might eventually get around to it, but we’re not sure at this point.

 

February 25th, 2011 at 2:31 pm Posted by TheOtherWinter

I’m not surprised at the opt-outs but it is too bad in some ways. I’d be curious if there were patterns to tracking of activity/mood FB updates against activity the previous 24 hours in game, or discussions of WoW-related topics or lack thereof.

 

Post Your Comment