Kieran Healy

Posted
15 April 2003 @ 9pm

Tagged
Politics

Voting Systems

You don’t have to be a psephologist to share Matt Yglesias’s dislike of first past the post voting systems. The sole virtue of such systems is that they are easy to understand: the person who gets the most votes in a straight count wins the election. Most everything else about them is problematic. In particular, a small edge in the percentage of votes cast usually translates into a huge advantage in terms of seats in parliament or congress. Thus, despite representing nearly a fifth of the electorate in Britain, the Liberal Democrats can expect to win less than 8 percent of seats in a general election.

Ireland, the ole green ‘n’ lovely, has a multi-seat constituency, single transferable vote system. Here’s an explanation of how it works for the simplest case, the Presidential election, where there is only one seat to be won.

From the voters’ perspective, it’s almost as straightforward as First Past The Post. Multiple candidates run for a seat and you numerically rank them, with 1 being your first preference and so on down the list. (You don’t have to express a ranked preference: if you like, you can treat it exactly like FPTP; just check a box for one of them and leave it at that.) In the simplest case, a candidate is elected when they exceed the quota, which is defined as

No. of votes / (No. of seats + 1) + 1.

In the simplest case, a candidate receives more than the quota on the first count and is then elected. But in regular elections, there are multiple candidates and multiple seats (up to 5 per constituency in the Irish case). Once the first candidate is elected, his or her “surplus” of votes (over and above the quota) is taken and distributed to the other candidates on the basis of the next preference on the surplus ballots. If there are no surpluses to be distributed at the end of a count, the lowest-placed candidate is eliminated and their votes are distributed to the next preference candidate. Here’s an example of that process at work for an imaginary 2-seat constituency.

You can argue the merits of competing systems. (For instance, you can wonder whether very large countries could feasibly be administered by STV systems—Germany has a mixed system that addresses some potential problems.) But one thing that’s unarguably true is that systems like the Irish one make election nights vastly more interesting for political junkies. No question of it all being over before the polls close in California. Though simple from the voters’ point of view, STV systems offer endless possibilities for vote management by savvy political machines. You’ll usually have several candidates from the same party running in the same constituency, with the possibility that all of them (more likely at least two) could be elected. They also provide much more entertainment as the election night drags into the election weekend, with tightly-fought battles for the fifth seat in some constituencies coming down to the last view votes. If you go to the count center, you can (in a sense) watch your vote move through the counts as candidates are eliminated and surpluses transferred. Altogether much more enjoyable. If you like that sort of thing.


10 Comments

Posted by
Kevin Drum
15 April 2003 @ 10pm

Jeez, Kieran, we can’t even handle butterfly ballots and punch card ballots here in the U.S. and you want us to figure out a multi-seat constituency single transferable vote system?

You are dreaming, my friend…..


Posted by
claxton6
16 April 2003 @ 7am

gets on soap box

The Center for Voting and Democracy is an excellent resource for STV and similar systems, though it tends to focus more on proportional representation.

I’ve been keen on this for a while, but it’s only really starting to click what one of the most powerful arguments for it is. As I try to go back and relearn a little statistics, one of the points that’s coming through is that one tool for evaluating the usefulness of a statistical calculation is: How much of your population is used in the test. For example, the range of values in a population only gives you information based on two values, the high and the low. The mean, on the other hand, uses all values in your population or sample in its calculation. In this sense, the mean is more representative of your population/sample. (Of course, it matters what your use for the number is, blah blah blah.)

A similar argument, I think, can be made about electoral systems, in that you can look at how much of the population actually wants the winner in office. First past the post only really counts whatever it takes to get to plurality. In a large field, even three people, that can leave us with someone that significantly under half of the population actually wants in office, in effect covering (in the case of Clinton in 1992, if I remember the numbers correctly), say, 43% of the population. Other electoral systems can be designed so that you get to actual majority representation, which more accurately reflects what the population’s preferences are.

Gah. Sorry.

gets off soap box


Posted by
Matthew Yglesias
16 April 2003 @ 7am

The PR lie

Many readers have responded to my first past the post bashing with the tired strategy of pointing to the instability of government that proportional representation systems have allegedly brought to Israel and Italy. This misses the fact that there are…


Posted by
John Isbell
16 April 2003 @ 10am

“You’ll usually have several candidates from the same party running in the same constituency, with the possibility that all of them (more likely at least two) could be elected.”
Doesn’t this unduly favor big parties, with extra people and money?
“The rich get rich, while the weak ones fade
Empty pockets never seem to make the grade…”


Posted by
Henry Farrell
16 April 2003 @ 11am

In response to John, smaller parties usually do OK. Constituencies in Ireland have 3-5 seats available; in the larger constituencies, you’re likely to get a seat if you have 7-8% of the first preference vote, which is quite doable for most small parties. As Kieran says, PR-STV systems are a lot more fun for political junkies. They have manual counting of votes (although they’re probably about to get rid of it) so that election nights see pasty faced politicians at the count centre, waiting as “tallymen” and local experts of their party monitor the votes, and relay their likely chances of winning back to them, as boxes of votes from this or that end of the constituency is counted. The PR-STV system also has disadvantages; it encourages localism and favour based politics. The worst enemy of a sitting member of the Dail (Irish Parliament) is likely to be a member of his or her own party running in his or her constituency; the way to beat your enemy is not to appeal to party ideology (which you presumably share) but to tell your voters about all the things that you have done for them (new sports fields, getting the Council to fix the plumbing in your home etc).


Posted by
space
16 April 2003 @ 8pm

One side benefit of increasing the power of 3rd parties and 3rd party candidates is that it would actually draw more attention to issues. The Republicans have gotten a lot of mileage out of portraying Democrats as “far left” or near-communist. Yet, Democrats cannot simply avoid being tarred by the anti-American brush by changing party affiliation. The two-party system is too entrenched. However, if 3rd parties had more clout, a progressive candidate could simply switch to a party that does not have negative baggage attached and force opposition candidates to attack genuine, rather than fabricated and mythical, positions.


Posted by
nick sweeney
18 April 2003 @ 11am

Jeez, Kieran, we can’t even handle butterfly ballots and punch card ballots here in the U.S. and you want us to figure out a multi-seat constituency single transferable vote system?

Punch cards and butterfly ballots and odd mechanical machines and computer voting systems aren’t the issue here: in fact, like fish knives and cake forks, they’re sophisticated tools which make a simple task (voting) excessively difficult. (Look at the ease with which Canadian elections are conducted, with paper ballots and hand counting.)

I know, Kevin, that you’re being sarcastic about the terminology: but Americans can handle ‘rank the following features in order of preference’ or ‘are you highly satisfied, quite satisfied…’ on a consumer survey—Americans are, after all, the most surveyed populace on the planet—then they can certainly rank candidates in a similar fashion. And if you’re going to use computerised voting, it might as well be for a system that works better with computed counting, rather than one that can be resolved by a bunch of trained squirrels looking at paper ballots.

And, heh, the Electoral College looks like a convoluted and tricksy system to anyone outside the US.


Posted by
derrida derider
21 April 2003 @ 4am

If I can make a plug for my own small country (Australia), we’ve had an STV system for over a hundred years (tho combining it with multi-member electorates has only happened in some states in the last forty). What tends to happen is that you get clear majorities, and hence stable government (unlike proportional representation), but there is no way the two major parties can ignore the interests of smaller parties. It is not at all unknown for parties to get a smaller primary (ie first) vote than their opponents, but win government on the preferences of the minor parties.


Posted by
antoin@eire.com
9 August 2003 @ 4pm

http://www.eire.com//blogarchives/000045.html

Electronic voting should be a hot issue in Ireland. It’s the core of our democracy, and the government wants to


Posted by
HogBlog
25 June 2004 @ 3pm

Dick Cheney’s Potty Mouth

I’m sure you’ve heard this one before… Vice President Dick Cheney has been dropping F-bombs on Democratic Senators. Since this…