qualitative vs. quantitative observations

[Tim Bergsma]

Matt wrote:

some people think unit and attributeDomain should be optional, or that
we need a
good default value if they are required.  For example, a default value
for
"unit" would be "undefined", and a default value for the attributeDomain
would
be a textDomain that matches a pattern of ".*". Need to figure out what
the heck
"dimensionless" is in STMML and how it relates to comment fields and
stuff like
that that aren't really data.  I prefer the latter (required but with
sensible
defaults). Need a decision on this. Hopefully during tomorrow's
conference call.
----------------------

Matt,

As we explore these issues, language conventions emerge that capture
ideas formerly implicit.  I suggest that comment fields, etc., are in
fact data, but they are qualitative, not quantitative.  The vast
majority of scientific data is quantitative, which tends to obscure the
fact that it is possible to conduct a completely scientific experiment
without numerical data.

The dimensionless problem remains, however!  Perhaps we should
distinguish between attributes that are inherently dimensionless (such
as a comment, or the color of your mood ring) and those that are
analytically dimensionless (such as moles per mole, or  liters per
liter, or percent).  

When I say that relative humidity is 15%, I'm really reporting the
calculation that the partial pressure of water vapor, say 0.285 kPa,
divided by the saturation vapor pressure, say 1.9 kPa, is 0.15.  By
dimensional analysis, kPa cancels, leaving a dimensionless result.  Of
course, if we bound the qualifiers to the dimensions, the cancelation
would not occur:  0.15 kPa-actual-vapor-pressure /
kPa-saturation-vapor-pressure.  These are the true units, and the number
0.15 is worthless apart from the implicit interpretive context. 
Unfortunately, the list of 'true units' is too large to be
standardized.  The options are 'dimensionless' or 'kPa /kPa, which is
helpful but incomplete.

I have to remind myself also that numbers are sometimes used to
represent merely qualitative observations, and for that reason are
inherently dimensionless.  E.g. Treatments 1, 2, and 3:  "Three what?"
you ask.  It could just as well have been A, B, and C, in this case. 
Also, observations of type "count" may appear to be dimensionless, but
the dimensions are implicit from the context:  e.g. 3 adult coccinelid
ladybird beetles per trap.  Inherently dimensional, but definitely
non-standard! (In that regard, standards like 'numberPerSquareMeter' are
a little misleading.  "Number of what?" must be inferred from the
context.)

regards,

Tim.

-- 
Tim Bergsma
LTER Information Manager
W.K. Kellogg Biological Station
Michigan State University
Hickory Corners, MI   49060
616/671-2337
tbergsma at kbs.msu.edu
http://lter.kbs.msu.edu