[EM] Bullet voting in STV-PR

[James Gilmour]

Subject changed - was under "IRV in action"

Markus wrote
> here are some quotations on bullet voting under PR-STV.
>
and gave five very interesting quotations.  (Details below)

My comment about bullet voting was exclusively in the context of a Condorcet
single-seat election.  There I can see a real advantage in bullet voting when
certain outcomes look likely based on public opinion polls and party canvass
returns.  Then bullet voting can be used to defeat the otherwise certain Condorcet
winner.

But Markus' quotations all related to STV-PR multi-seat elections.  Here the
situation is quite different.  Unlike the Condorcet count, I cannot see how bullet
voting can possibly bring any DIRECT advantage in an STV-PR election.  Your vote
stays with your first preference candidate so long as that candidate needs it to
secure election.  I know candidates and parties may sometimes recommend bullet
voting, but where is the real direct advantage?  I suspect there is a lot of
misunderstanding and misinformation put about.  I should like to see the evidence
that bullet voting contributed to the election of any candidate through the
operation of the STV-PR counting system.

There could be an INdirect advantage, if by recommending bullet voting a party
gets its supporters to do more of what it wants.  Perhaps the messages "Vote only
for X" or "Vote only 1, 2, 3 for X, Y and Z" are easier to put across and do
produce a more cohesive response from the supporters.  That sounds plausible, but
I wonder if there is any evidence for the different campaign approaches having
different effects on voter behaviour.


"Vote averaging" is a separate issue that comes into play where a small party will
win one seat and might win two on transfers.  If one of its candidates is very
popular and gets most of the party's first preferences, the second candidate may
get so few first preferences that he or she is excluded at an early stage in the
count.  So the second candidate is just not there to pick up the transfers when
they could occur later.  To avoid this, the parties try to "average" the first
preference support over both candidates so that both survive the early cuts and
both pick up later transfers, both achieve the quota and both get elected.  Bullet
voting doesn't come into this at all.


JG Comments of the Quotations on bullet voting under STV-PR:
>
> Harold F. Gosnell ("Motives for voting as shown by the
> Cincinnati PR election of 1929," National Municipal Review,
> pp. 471-476, 1930) writes:
>
> > A. Lee Beaty, a prominent Negro in Cincinnati, requested a place
> > on the organization Republican ticket, but this was refused. Beaty
> > then joined forces with the two independent Negro candidates and
> > became their campaign manager. He advised the Negroes to vote only
> > for the two Negro candidates and to express no other choices. The
> > purpose of this move was to deprive the organization Republican
> > candidates of the transfers from the independent Republican Negro
> > candidates.

Sounds like an (understandable) exercise in spite.  But all it did was deprive
those voters of the opportunity to have any further influence in the election.

>
> Frederick Shaw ("The History of the New York City Legislature,"
> Columbia University Press, New York, 1954) writes:
>
> > No one seriously believed that 9% of the city's electorate were
> > extreme radicals. But how did it happen that two Communists were
> > elected in 1943 and again in 1945? Part of the answer lay in the
> > technique they developed. Realizing they could scarcely seat more
> > than one candidate in any borough, they instructed their adherents
> > to cast a single vote, without secondary preferences. "Bullet"
> > voting, as it was called, assured such candidates of a solid block
> > of votes.

This sounds like a comment from a commentator who did not understand how STV-PR
works.  A "solid block" of first preferences with other preferences marked too
will have just as much effect as a "solid block" of first preferences alone.

>
> Warren Moscow ("What have you done for me lately?," Prentice-Hall,
> Englewood Cliffs, 1967) writes:
>
> > The Communists, with possibly 50000 dedicated supporters, were
> > unimportant in a state or city-wide general election, but they
> > were very important when consolidated behind candidates for the
> > City Council under the system of proportional representation,
> > which fostered bullet voting, that is, casting only one first
> > choice vote for a single candidate and ignoring all others.

I suspect this was an effect of district magnitude and geographical concentration
rather than bullet voting.  I cannot see how bullet voting could have achieved the
effect attributed to it.

>
> Basil Chubb ("The Government and Politics of Ireland," 1971) writes:
>
> > When the Sinn Fein Party appealed to its supporters in 1957 not to
> > give lower preferences to candidates of other parties, the number
> > of votes that became nontransferable as Sinn Fein candidates were
> > eliminated was very much higher than normal.

That should be no surprise.  That's what happens in all STV-PR counts.  If a paper
of an elected or excluded candidate has no further preferences on it, that paper
becomes non-transferable.  So if a locally significant party persuades its
supporters to vote only for its candidates, of course the number of
non-transferable papers will be higher than normal.  But that party will not have
secured any more seats by that advice.  All they will have done is to deny their
supporters the opportunity to determine who else will join them on the local
council or in the Assembly.

>
> Mart Bax ("Harpstrings and Confessions," University of Amsterdam, 1973)
> writes:
>
> > At this point in my argumentation one may object that fellow party
> > candidates must cooperate because under PR systems they need each
> > other's second and lower preferences to be elected. Indeed, Boissevain
> > observes that one of the main tactics of Maltese politicians is to
> > exchange second preferences. To some extent this holds also for Ireland.
> > However, a candidate is _never sure_ that his party colleague will keep
> > his promise; he might promise his second preferences to more than one
> > colleague. Indeed, he might even try to induce the electorate to give
> > no more votes than just for him. (This happened on a large scale in one
> > constituency during the 1969 Dail elections. The candidate in case won
> > with a landslide, he headed the poll, but with the result that the
> > second candidate of the party was eliminated. The successful candidate
> > expected to obtain a post as a junior minister, and bought a new suit
> > in anticipation. Unfortunately, however, the prime minister punished
> > him by not giving him the "prize".)

This sounds like a situation where the party concerned should have applied "vote
averaging", but the personality of one of its candidates got in the way of its
success.  The techniques for successfully promoting a team of candidates are quite
different from those used for single candidates in single-member seats.  One
complication in Ireland might have been that a significant number of TDs (members
of the Dail) were simultaneously local councillors.

But the reverse effect from that reported above has also occurred in a Dail
election.  One TD was so sure of his re-election that he urged his supporters to
mark their first preferences for the other candidate of his party in the hope that
they might then win two seats where they previously had only one.  His supporters
followed his advice in large numbers, but the effect was not what he had hoped.
The newcomer was elected and the TD was dumped!   Tactical voting under STV-PR in
real elections is very difficult, to the point of near impossibility.

James