[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: Cluster analysis -similarity and dissimilarity measures
I am not aware of specific provision for ordinal
variables. In many circles part of the definition of
an ordinal variable is that the measurement scale
falls short of metric properties, so that the
existence of a similarity or dissimilarity metric
would appear arguable at best.
In some circumstances, it might be useful to
convert data to ranks or -- in this case --
something similar like "ridits".
Or, you could just say, code them 1 to 4 and
feed them in and see what comes out, being
aware of the arbitrariness of that.
There is cute stuff in Persi Diaconis' monograph
on group theory and statistics that might apply
here, but I'd be surprised at a Stata implementation.
Like the North Face of the Eiger, this material
is better approached from behind, e.g. through
Marden's book on rank data.
Diaconis, P. 1988. Group Representations in Probability and
Statistics. Hayward, CA: Institute of Mathematical Statistics.
Marden, J.I. 1995. Analyzing And Modeling Rank Data.
London: Chapman and Hall.
> I have a question about the measure options for conducting
> cluster analysis. I see that there are measures for continuous and
> binary variables, but what about ordinal variables? Is there a measure
> available in stata that I can use and just can't find? Or should I do
> some sort of standardization of the variables? The variables have 4
* For searches and help try: