Subject- Nominal character

[Warren Lamboy]

 Subject:  Nominal characters

 In my opinion a number of separate issues are being confused in this
discussion
 of nominal characters, although I am ready to admit that perhaps the only
 confusion is in my head.  The broad categories of discussion seem to be:

 evolution of characters/character states
 coding of characters/character states for stat/math/phylogenetic analysis
 the data type (nominal/ordinal/etc.) of characters as they appear to us
 the data type we decide to use in treating the characters
 the "real" data type of the characters
 the physiological, biochemical, and environmental mechanisms determining the
     characters/character states

 As an example of what I mean, take the character of seededness in table
grapes.
 To the consumer of table grapes, it has two states, seeded and seedless,
based
 on the consumer's perception of how large the seed is with respect to the
size
 of the grape berry as well as the relative hardness of the seed coat and a
few
 other factors.  One might to tempted to call this a nominal character--how it
is
 coded for analysis is immaterial 0/1, p/q, 1/0, etc.  As I have suggested,
 however, the situation is not really either/or; it really boils down to
whether
 the hardness and size of the seed have crossed the threshold from seedless to
 seeded--and this threshold depends on the particular observer (eater of
grapes)!
  So the reality is that a number of different ordinal states or perhaps even
an
 approximate continuum of seededness exists, so that ordinal or continuous is
the
 "real" data type from one point of view, while as far as a grape wholesaler
is
 concerned, either the grapes have seeds or they don't (nominal). So we may
want
 to treat seededness as an ordinal character, a nominal character, or a
 continuum, whatever is best for our purposes; the character states would need
to
 be coded accordingly.  Actually, analysis of seededness using RAPD markers
finds
 several quantitative trait loci (QTL's) associated with perceived seededness,
so
 that the genetic or physiological mechanisms behinded seedlessness are
complex,
 and as a consequence, so probably was/is the evolution of the character
state.
 So the answer to the question of what kind of character seededness "really"
is,
 is possibly different from how it appears to us, is possibly different from
how
 we wish to treat it, is possibly different from it physiological/genetic
basis,
 is possibly different from how it evolved, and is possibly different from how
we
 code it for analysis.  As far as I can tell, the best that one can do is to
 state explicitly how one is treating a character and its character states,
and
 let someone else try to do better if they don't like what one did.  Arguments
 over whether a character is really nominal, ordinal, continuous, discrete
etc.,
 would seem not to be productive, in my opinion.  -  Warren