[EM] cluster analysis

[James Gilmour]

Forest wrote:
> I can think of two general approaches to applying cluster analysis to
> election methods:   <details cut>

I should be very interested to see how you would apply this to the STV-PR ballots from the 2002 Dáil
Éireann election.  Full ballot "papers" are available for the three constituencies that used
electronic voting:
http://www.meath.ie/election.html
http://www.dublincountyreturningofficer.com/Results_Download.html

I started to look at this, using GenStat for Windows, but the "missing values" (truncated ballots)
cause rejection of large numbers of ballot papers or total rejection, depending on the analysis
attempted.  (NB I have used GenStat for statistical analysis for many years, but this is for me,
both a new area of statistical analysis and a new area of GenStat use.)
James