home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!MARLIN.JCU.EDU.AU!BHMJS
- Message-ID: <199211090704.AA02840@marlin.jcu.edu.au>
- Newsgroups: bit.listserv.stat-l
- Date: Mon, 9 Nov 1992 17:04:14 +1000
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: Michael Smithson <bhmjs@MARLIN.JCU.EDU.AU>
- Subject: Measure of agreement for continuous variables
- Lines: 20
-
- Regarding R. Zambarano's request for a measure of agreement between
- two continuous variables, I can recommend two sources that might be helpful.
- One is an article by F.E. Zegers and J.M.F. Ten Berghe in Psychometrika (1985,
- v. 50, pp.17-24) on a family of association coefficients for metric scales.
- They consider a special case where the scales are uniquely (or absolutely)
- determined, which gives a similarity measure. There was some discussion
- in the literature which resulted in a later modification of their coefficients
- to take chance effects into account.
-
- Also, D. Dubois and H. Prade's 1980 book on Fuzzy Sets and Systems: Theory
- and Applications (Academic Press) gives some measures of similarity
- proposed for fuzzy sets that can be generalized to continuous scales in the
- nonnegative domain of the real number line. If the two variables
- are X and Y, then one such measure just requires summing min(X,Y) for all
- observations and dividing by the sum of max(X,Y).
-
- Dr. Michael Smithson
- Behavioural Sciences
- James Cooko University
- Queensland 4811 Australia bhmjs@marlin.jcu.edu.au
-