🚀 go-pugleaf

Knowledge structures, such as Pathfinder Networks, HCS, and MDS have
been used to characterize information in complex cognitive domains
since the 1950's in Psychology.  Many of you reading this are familiar
with how they can be used to illustrate the structure of information.

There are a lot of experiments that show differing knowledge
structures, such as one from an Expert group and another from a Novice
group.  I'm analyzing some proximity data from an experiment, and I
want to know: HOW CAN YOU SAY TWO KNOWLEDGE STRUCTURES ARE DIFFERENT?
Of course you can eyeball them, comparing links, levels, and so on,
but how do you know they didn't just end up that way due to random
variation in the subject population?

Can anybody point me to some statistical techniques that allow you to
state, preferably with a p-value, that two knowledge structures are
really provably different?  It seems like I read something awhile back
that  analyzes the proximity matrices underlying the structures, and
is able to produce a p-value based on how much they differ, but I
can't remember exactly who did this.  Any references or comments would
be EXTREMELY welcome.

	Boyce
-------------------------------------------------------------------------
|W. Boyce Byerly			| Internet: wbb@acpub.duke.edu
|Department of Psychology, SHS		| Phone: (919)-477-0717
|Duke University			| 
|Durham, NC 27704			| 
-------------------------------------------------------------------------