[MUD-Dev] Personality Types (was Anarchy Online (was Community Feeling))

John Hopson jwh9 at acpub.duke.edu
Thu Aug 16 14:54:12 New Zealand Standard Time 2001

At 02:58 AM 8/16/2001, Richard A. Bartle wrote:
> On 15th August 2001, Hans-Henrik Staerfeldt wrote:

>> What would be neat would be to have a survey with a thousand
>> questions, and do a components analysis to see which questions
>> were answered similarly.

> I really do not like questionnaires that do that, they mean you
> have to remember what you answered the first time. I once did a
> test at school with, ooh, it must have been well over 500
> questions. On page 30 there was some question that was the same as
> one on page 18 but with the sense reversed - instead of "put these
> vegetables in the order you most like them" it was "put these
> vegetables in the order you least like them". It took me a good 10
> minutes to find what I'd put the first time so I could make their
> software think I was more consistent than I was...

Um, I'm not sure this is a good objection.  Basically, you're saying
that redundant questions make it harder to mislead the test.  Which
is, as I understand it, part of the point of asking redundant
questions.  If we're building a survey to accurately judge how
people think about online gaming, we want it to be hard to cheat.
And as you said, it lets us get an idea of how "hard" their opinions
are.  We don't necessarily want players to be rock solid consistent,
we want a clear picture of what the player thinks.  There may be
legitimate reasons why a certain kind of player answers question A
one way and question B another, even if those two questions are
really similar.  The difference might even turn out to be important
to our understanding of players.

The other thing to remember is that having redundancy helps evaluate
the questions as well as the respondent.  If we designed a good
test, we should be able to split the test in half and the respondent
should get the same score on both halves.  If we group players by
their types, we can then look at the way different groups score on
each questions and thereby get an idea of which questions best
define the types.


MUD-Dev mailing list
MUD-Dev at kanga.nu

More information about the MUD-Dev mailing list