[EM] Re: Testing 1 2 3

James Gilmour jgilmour at globalnet.co.uk
Sat Jan 3 09:43:11 PST 2004

> In some political cultures there is a greater tendency to 
> vote the party ticket than others. I have just been looking 
> at the transfer patterns for the Bass division of the 
> Tasmanian House of Assembly for the 2002 state election for 
> the 3 largest parties. For the Labor party on each transfer 
> of surplus/ elimination of a candidate between 83 and 90% of 
> the votes transferred to other Labor candidates. For the 
> Green party between 87 and 91% of votes transferred to 
> continuing candidates of the same party. For the Liberal  
> party the figure was 90- 96%.

It should be no surprise that within party transfers are particularly high where candidates are
listed on the ballot paper by party.  In these Tasmanian elections each party is listed in a
separate vertical column on the ballot paper, with Robson rotation of candidates' names within each
column.  This design of ballot paper encourages voters to mark successive preferences within one
party before going on to consider another party.  (British electors do not like or want parties
listed in separate columns across the ballot paper.)

This is an example of yet another type of factor that has to be taken into account in the analysis
of the results of real elections.  It certainly makes realistic modelling much more difficult, to
the point that I would doubt its value.

> Also in Australian Senate elections with "above" and 
> "below" the line voting  you can sometimes find that 
> over 99% of a candidate's surplus transfers to the next
> candidate of the same party.

I have to confess that I regard this implementation of STV-PR as a travesty of what I believe STV-PR
is about.  "Above the line" and "below the line" voting effectively reduce STV to a party list
system of PR.  In such circumstances any analysis of transfer patterns is almost meaningless.


More information about the Election-Methods mailing list