Here's my .02: Some measurement distinctions are more important than
others. For statisticians, the distinction between, say, nominal and
interval is very important: computing the mean of a nominal variable is
nonsensical.
However, the distinction between interval and ratio scales is _not_ at all
important to statisticians-- how many _statistical_ tests can you think of
that are applicable to ratio data but not to interval data?
Therefore, in my text I acknowledge that there are four levels of
measurement and briefly define each. However, I then say that we will
collapse the last two together. Students are then required to know the
difference between "nominal," "ordinal," and "interval/ratio" data. That
pedagogical method has the virtue of teaching distinctions that will
actually be exercised throughout the textbook.
--Russ
_________________________________________________________________
Russell T. Hurlburt, Ph.D. Email: [EMAIL PROTECTED]
Professor of Psychology Telephone: (702) 895-0194
University of Nevada, Las Vegas Fax: (702) 895-0195
4505 S. Maryland Parkway
Las Vegas, NV 89154-5030 USA
http://www.nevada.edu/~russ/hurlburt.html
See Comprehending Behavioral Statistics at
http://psychology.wadsworth.com/authors/hurlburtr/cbs.html
_________________________________________________________________
