28May

After the Irish election, we're inspired to write an Explainer piece. It's answering a poser posed last month by Doctor Vee: Parties are only fielding as many candidates as they can possibly hope to win. In most cases, this is one. And while STV gives voters more choice on paper, this extra choice has actually made me feel a bit suffocated.

If you don't know what STV is, or how it works, then go back a month, read the links there, learn and understand. Remember, the Irish understand this system.

For the purposes of this example, we're going to use an example constituency of 60,000 people, electing five members from amongst six parties. The parties are as follows:

With five people to be elected, the quota is one-sixth of the vote. FF has two and a half quotas. FG has one full quota, and a spare 11%. Neighbour is 2% short of a full quota. SF, G, DP all have less than half a quota. Eyeballing these figures, we would expect the distribution to be: FF 2, FG 2, Nei 1.

On this basis, we can expect FF and FG to put up at least two candidates. Is it in FF's interest to put up a third candidate? Very probably; things can change between the close of nominations and polling day, otherwise the campaigning would be of no benefit whatsoever. If FF were to poll 45%, and have two candidates, the top of the poll might look like this:

That leaves 7000 FF votes - the best part of a quota - to transfer to the third candidate - but there is no third candidate! These votes might go to FF's preferred partners, but transfers between parties will tend to be even less reliable than transfers within parties. A third candidate might mean that no-one would be elected on the first ballot, but the transfers when the last candidate was excluded would tip both others over the top.

In practice, Irish parties tend to put up one more candidate than they can reasonably expect to win. Let's return to our original model, and put in some numbers.

So, using the approximate pattern of transfers we observed this week (DP to FF; G to Nei to FG, but with a large amount of random scatter), here's how things unfold.

Though he came seventh on first preferences, transfers from other parties have promoted Blair to be the first person elected. We'll play out the count:

At this point the count concludes, with Morrison, Budgen, Beaufort, and Storm elected without receiving a quota; the distribution of Marks (and surpluses) will force the other candidates over quota.

So, should parties field more candidates still? Perhaps not; let's see what happens if FF gets cocky and campaigns badly - the party fields a full slate of candidates, receive only 35% of first preferences, and find their preferred partners the DPs fall right away:

Between them, the FF candidates are down to about two-and-a-quarter quotas. Rather than put up even more long tables, we'll skip to what happens after the minor candidates and first FF candidate have been eliminated.

Spencer falls out, lifting the three remaining FF candidates above the Neighbours, then Brown's elimination elects Blair.

Look at the remaining FF votes. They've dropped from two-and-a-quarter quotas to barely two quotas. Blair's surplus (in reality, the third preference of Brown's transfers) still goes almost entirely to FG, third FF candidate Marks is eliminated, and here's the final score:

FF only has one person elected, the fifth seat has gone to FG. Why has this happened? It's simple - transfers are never 100% effective. Voters will not always follow party lines exactly - in the example above, some voters may have taken a shine to candidates Marks and Spencer, but think that Morrison is a cheapskate unworthy of their vote. They'd rather see someone - anyone - else elected ahead of him.

The strategy is to field at least as many candidates as seats you're certain of winning, but no more than you think you could possibly win. Generally, these numbers will either be identical, or separated by just one candidate. Fielding too few candidates is a sign of little confidence; fielding too many allows transfers to erode your support.

In the example cited by Dr. V earlier, Labour fielded two candidates, and four other parties put up just one. Three candidates were to be elected. We suspect that at least one of the one-candidate parties should have put up a second candidate, but it's difficult to predict which.

| Permanent link