Details of Factor Analysis
A principal components analysis was performed to arrive at a parsimonious representation of the associations among the 39 items. 10 factors were extracted with eigenvalues greater than 1. Together, these factors accounted for 60% of the overall variance. The chart below shows the factor loadings of the survey items used.
The scores for all subcomponents were generated for each of the 3200 respondents using a regression method. Another principal components was performed on the 10 subcomponent scores. 3 factors were extracted with eigenvalues greater than 1. Together, these 3 factors accounted for 54% of the overall variance. These 3 factors are largely uncorrelated (r's ~ .10). The chart below shows the factor loadings of the subcomponents on the 3 main components.
Nick - fascinating stuff. Are you submitting for publication in a peer-review journal? If so, do give the reference when it's published. Two points:
1. This needs a snazzier name...!
2. What aspect of the model are you personally least comfortable with?
Nice work! As with Alasdair, this needs to be put together for peer-review submission. Some thoughts:
1. I notice all 39 items remain in the final survey, yet you indicate some high correlation between what should be somewhat unique constructs. I don't see refernce for your factor loading decision matrix (i.e. factor loading >.5 on a single factor, no higher than .3 on any other factor or some similar rubric). Is it possible that a few items could be omitted from this?
2. Given the above, you've used Principle Components Factor Analysis. If you are expecting signfiicant correlations between constructs/dimensions have you considered running Factor Analysis using Principle Axis with a Varimax rotation? This might better capture the unique clusterings within a correlated dataset.
I'm in agreement with you that Bartle's index is simply too restrictive; there are too many cross loading patterns. I've always wondered about the griefer as achiever aspect as well (I notice you found a correlation here)... I suspect that griefing and "something else" are dimensions of achievement.
I notice a couple of your alphas are a bit low (below .70). I doubt this would be a problem in an exploratory study (given that they're over .60 you're fine per Nunnaly 1978), but there is some potential to investigate for tweaking in the future. Have you looked at alpha if item deleted to see if a specific items removal "helps?"
Anyhow, good work and a very good read. Please update if this article (in any format) is going to publication, I'd love to cite it in some work I'll be doing.
Alasdair asked: "What aspect of the model are you personally least comfortable with?"
All the subcomponents have fairly normal distribution except for the Competition component (skewed low). I'd like to see that component be more about Competition than Griefing.
Troy - Thanks for the stats feedback. Will definitely check those things before sending it for review anywhere. There were some cross-loadings - will play around more with it to see if I can get it to be cleaner.
Nice work but as one who has participated here since the very first survey , I am not so surprised .
However there where I am frankly surprised is the proposed treatment of "Escapism" .
The alpha is low and the loading in Immersion is even lower .
I suspect that your Immersion axis is not right as it is .
In any case "Escapism" shouldn't correlate very well with the other 3 factors .
If "Escapism" is surely a motivation factor to ... escape something somewhere , it is not very specific to MMORPGs .
That's why it might (but I have not your detailed data to assess it) be misleading to put it at the same level as other , specifically MMORPG related factors , like role-playing or looking after in game-mechanics .
I also suspect that the "Escapism" correlates to other factors outside of Immersion and more particularly to socialisation .
I am particularly sensible to that aspect because I know many people playing MMORPG (myself being one of them) who would score extremely high on the 3 Immersion components but extremely low on the "Escapism" component because they are perfectly happy with their RL and are not "escaping" anything .
Symetricaly I can imagine cases (even if I don't know such cases personaly) scoring extremely high on the "Escapism" component but rather low on the other 3 components of Immersion .
After all Immersion is like taking holidays on Greenland (if you have never been to Greenland , I recommend it :)) and it doesn't imply that you are escaping something specific .
So if "Escapism" may imply Immersion , the opposite should not be statistically significant .
However I understand the need to have the "Escapism" component in a MMORPG study .
I suspect that people driven primarily by that motivation and even should they be a minority are precisely people who could be more subject to "addiction" and some other pathological effects .
I suspect that there are 2 kinds of "Escapism"
1) The harmless one where the person "escapes" from the work and stress of the every day's life by having a recreative activity - reading a book , watching TV , taking a drink or having a nice cool session of MMORPG .
This kind could correlate to about anything - Immersion , Socializing ...
2) The potentially dangerous one where the person has a problem with RL altogether because he/she doesn't feel as having a right place in the RL or is missing something vital and transferring/compensating to a MMORPG instead .
I could imagine this kind correlating with Achievement and perhaps Socialization but probably not with Immersion .
Those are the people that get said in game "Get a life" .
To make a long story short : I feel that the "Escapism" is a meta category that has to be differentiated more in order to fit in your present 3 factors .
Unless you discover that there is a 4th , not yet clearly identified motivation axis .
The problem is more semantics than anything else. The word "escapism" and "escapist" has negative connontations all over the place. People feel uncomfortable saying they use games to escape. I know I am. Maybe questions that are more neutral-sounding and don't mention escape would make escapism a more popular answer choice.
I don't think there is really "good" and "bad" escapism. It's just that you can't overdo escapism. This is hard to measure (how much is too much?) and even harder to admit in a survey. An addict will usually say that the game helps him to relax and deal with his RL problems better.
I didn't actually mean my comment as a semantic one .
My main remark was technical - the subcomponent called "Escapism" (whatever it might mean) has a low loading in the "Immersion" axis and a low alpha (see the details at the end of Nick's study) .
That means that it doesn't fit well in the Immersion axis .
And if a component gets low or strange correlations that might (but must not) mean that it is badly defined or is itself a combination of other , not yet clearly defined , factors .
So it merits probably a new set of questions to get a better understanding how this component behaves and if it can be considered as a "clean" component .
I am currently developing an MMORPG, and though I lack money and employees, I am absolutely deadset on finishing my design scheme and getting it programmed.
My question is, in the way of developing a social and economic structure, what do you most recomend for, and what do you advise most against?
I'm a big fan of your work, thatnsk for listening.
I am absolutely praud of you guys,just keep it up.
Hey there. I'm working on a big document right now for my employer that refers to reason people play games, and what's fun, and what isn't. I've been looking into Bartle's 4 player types, and saw some more site actually use those as "parts" of what define the player (92% socialiser, 8% killer, 60% explorer, 40% achiever -- Sums up to 200%, strangely enough)
Anyway, I also did some more research on the subjects, and found Nicole Lazzaro's 4 keys the reason why people play games (Hard Fun, Easy Fun, Altered States and People Factor)
Now, here's the interesting part : Your grid only uses 3 "overarching" factors for your 10 factors, which is Achievement, Social, and Immersion.
I tied those to the 4 keys, which gave me:
- Achievement = Hard Fun (Ok, I admit, the competition part is actually in good parts People Factor)
- Social = People Factor (And here, we see the chatting, which is usually made more simpler if enough Easy Fun tools are availlable, i.e. inns, etc)
- Immersion = Easy Fun
But that leaves out Altered States... Which I tied to Escaping.
That leads me to believe... that perhaps Escaping is a category of its own...?
Well, I didn't do any studies in the matter, I'm actually just a programmer that loves game design and MMOs, and takes a good amount of my time reading about these kind of things, at and off work hours.
| ) ( ) |-|
It would be interesting to study how a player's motivations are affected by the game environment, rather than the player's personality.
For example, I like to play solo, and try to create characters that are self-sufficient, but I do love to group when I have a good group. A large part of the reason that I prefer to solo is that it is difficult to find good groups. My main motivation for soloing is avoiding little kids, griefers, and bad players. The motivation has more to do with the other players' behavior than with my own personal preferences. If I could find a game with lots of skilled, mature players, I believe I would group more often than solo. Perhaps it is my game experiences that have molded me into a solo player.
Do you plan to make this an online self-assessment tool, the way the Bartle test has been? I would be very interested to see my own results (as well as other players').
Good luck! Your site is very thought-provoking.
Daemone - Here is a link to the assessment tool.
The Bartle types are derived from NLP's "Meta Programs". It's good to see someone verifying empirical observations from NLP. Thanks.
"Kingdomality" is another classification built on NLP. The four personality types defined in Kingdomality: Challenger, Helper, Maintainer, Explorer map directly to the Bartle's types.
Cheers, Oleh K.
PS Results of your survey do not contradict the types - I think, you have misclassified Maintainers/Killers.
From my own experience I think you are missing a very vital category and that is developing skill in playing the game for its own sake.
My fondest memory in any MMORPG was playing two accounts simultaneously in Everquest while raiding Plane of Time. I had to switch tabs at least every 2 seconds between my Enchanter and Bard who were in different areas so I could keep crowd control on two different groups of monsters at once. If I could pull it off it was rare for anyone else in the 72 man raid to notice, if I screwed up even a tiny bit the raid would wipe and we would have spent 50+ collective man hours and achieved nothing.
I would suspect that a significant portion of your respondents similarly seek to master the game rather than to dominate or impress other players.
You may be interested to find your results changing based on how much time each individual player has to play the MMO in a given timeframe. Socializing aspects may not seem very important to the mechanics-minded players if they can only be around for a handful of hours at a time.
I would hypothesize that the amount of time each contributing response can remain in the game given a set time frame would directly correlate to the mixture of answers recieved from that individual.