[EM] Biasing STV for women: not a possibility. Ireland transfer values in European Parliament elections

[Craig Carey]

http://www.environ.ie/elections/howelect.html

Ireland's STV algorithm that elects members to the European
Parliament calculates the transfer value for the transferring of
papers away from winners, as the smaller of 1, and the surplus
divided by the wieght of the transferrable papers. This way of
calculating the transfer value of course results in less wasted
votes. In my opinion this is the less usual way, and this way is
also better. My opinion is that the wastage of votes is much too
high still even after the Irish fix, and it should be increased so
much that there would be a resulting increased wasting of votes in
the transferring away from losers, but that may require pulling the
transfer value out of STV altogether which is not something that
people have an imagination for. It is worth noting that the page
makes it clear that the transfer value is clipped back to 1 (and if
that were not done, then some quite alarming examples can be
constructed showing that the transfer value should not exceed 1).
[The Multiwinner Approval Vote of Mike Ossipoff has the defect that
the transfer value can rise much above 1, which would doubtless
prevent the method becoming much favoured.]

The Australian Senate uses the larger denominator. Ref (5) of
Appendix A of: http://www.aec.gov.au/pubs/electoral_systems.htm Few
would copy the Australians without careful checking.

-----------------

A numerical check on the howelect.html page that shows a 1994
election in Ireland for people to stand for the European Parliament
http://www.election.demon.co.uk/ep1994.html

The quota is 139,967, and Rev. I.R.K. Paisley exceeded the quota in
the first stage with a surplus of 23,279. There were 17 candidates.
The sum of the papers transferred away is only 22273. Maybe the
candidate's transfer value was 1, and there were 1006 (=23,279-
22,273) votes that had a 1st preference for that candidate and no
2nd preference. By simply using ordinary STV there seemingly would
have been a difficult to argue for wastage of 0.087% of the popular
vote.

75+524+39+219+55+5+25+306+36+44+10+56+16082+4717+80 = 22273

Here is an article by James Connelly that describes the smaller
denominator transfer value but which fails to mention that it is
truncated back to 1 if it exceeds that:

http://www.solent.ac.uk/socsci/jc/voting/stv.html


--------------------

The LWVWA has a significant PDF document on voting systems and
social aims and methods. The report has comments on the Limited Vote
and so on. The report takes a rather 'high altitude' view.


    An Evaluation of Major Election Methods And Selected State
    Election Laws,  Fall 2000,
    By The League of Women Voters of Washington Education Fund

     http://www.lwvwa.org/election/

     http://www.lwvwa.org/election/emstudy.pdf


The CVD might have tweaked the STV method renamed it, given it a
brandname.

As far as I can tell, the Choice Vote method is tweaked (? data?) and
the CVD is dominated by men. If all possible interests of election
method selectors who select STV variants are partitioned, then they
could be classified like this:

   * STV variants that male councillors favour
   * STV variants that are not favoured by male councillors

Then a next step here is to note that councillors that are not male
councillors tend to be women councillors. It just seems to be that the
brick layers that created the Choice Vote have some freedom to tweak
the method. It can be speculated that the criteria might be the
production of a Choice Vote method that they favour, i.e. one that is
just what is needed. Above in this text there is described two
different ways to calculate the transfer value. Another tweak option
is Hare vs. Droop, and another is that quotas that vary with stages,
and there is sure to be more. They are political in that the name of
the candidate and party is not actually written into the counting
method.
I have never actually seen the wording of a CVD Choice Vote method.
A real need is for some group to post in the Choice Method. At the
least, we want to see if it changes over the years.
Maybe the use of a fairer Irish method that wastes less of the intent
of voters is opposed unacceptable to one of, but not both of, the
CVD and the Women's League of Washington. Probably most subscribers
will have a lot of trouble sustaining an interest in this. Can the
women plainly rule out tweaking for a purpose of favouring women?.


---------------------------

The report takes a bit of small stand against the tweaking of
vote data. Methods seem to be safe. [This is maybe more advanced
than the LWV women felt competent to set restrictions onto]

    "Safeguards from Fraud
    There seems to be a very strong public trust in the veracity of
    our election system. Yet there are remarkably few investigations
    made which could provide assurance that security of the ballot
    box is as solid as most voters would like to believe. The trend
    from poll voting to mail voting removes more of the process from
    public view. The need for knowledgeable observers is great."

Jimmy Carter once said he must have seen all possible types of
electoral fraud in his own state of Georgia. ['seen them all' or
something]. They must be strongly stressed to get high tech and
shift the fraud into the method. Old attempts to write fraudulent
vote mis-counting hardware wasn't a huge success in USA. No comment
on the CVD (minimal interest).

A danger could be that after electoral reformers move in and the
possibilities for altering STV become restricted. Do women regard it
more in their interests to waste votes transferred away from winners?,
or from losers?. The latter option seems right. Just what data
indicates for unusual/new/female candidates is not known. It would
vary from place to place, and there might not be enough interest.

I'd prefer to carefully avoid use of real data, but we should know
if the bias directions indicated by self-interest of minor groups,
and by strict mathematical principle, agree or diverge.


---------------------------

Correction

I withdraw my allegation that the Washington Women's League is
intending to violate Article 25 (b) on the International Covenant
on Civil and Political Rights. I have no evidence for this
new stand, and nearly entirely none for the original allegation
that was in the subject field of a message I sent to this EM list.