[MUD-Dev] Codename Blue & Facets - Nick Yee's new studies

Nick Yee contact at nickyee.com
Fri May 10 10:20:18 New Zealand Standard Time 2002

From: John Buehler <johnbue at msn.com>
> Raph Koster writes:
>>> - The "Explorer" Bartle's type does not appear to exist.
>> I would say that the explorer type doesn't have material to
>> entertain it in the three games mentioned.
> You can say that, but then you'd have to reconcile it with this
> page:
>   http://www.andreasen.org/bartle/stats.cgi

The short answer is that a dichotomized test using a non-validated
categorization does not produce valid results.

The questions on the Bartle Test (which Bartle did not write) force
a choice between 2 options although there are 4 underlying
categories (none of which are validated). To see why this is
problematic, imagine I have a poorly constructed categorization of
food types: ice cream, Italian, stir-fry, and appetizers. Now I go
ask people questions like - When you eat dinner, do you prefer "ice
cream" or "Italian"? When you go out for lunch, do you enjoy
"stir-fry" or "appetizers"? And imagine after 30 questions from one
respondent, I notice that they choose "Italian" more often and I
conclude that this person is an "Italian" food type. Are you
comfortable with this conclusion? And if I sample 100 respondents,
and 40% of them are the "Italian" food type, what does this really

The dichotomization is problematic because it excludes 2 of the 4
options from each question, and it forces a dichotomy where none may
exist (Isn't it possible that someone may like both "appetizers" and

But the real problem is that the results of a poorly constructed
test of an unvalidated categorization are well ... not very


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

More information about the MUD-Dev mailing list