Sunday, 22 August 2010

What if we decided election winners using the Big Brother voting method?

This weekend, Australia went to polls in the 2010 Federal Election.

Elections make for some fascinating number analysis. As readers from a while back might remember, I love the statistics of elections. Australia has a preferential voting system, whereby voters list the candidates by order of preference. As opposed to the first past the post system used in Britain, the winner is not decided by who receives the most primary votes, but rather who is the most preferred candidate. Sometime soon I will write a post on the various voting systems used worldwide - see Plus for a great introduction to various voting methods - but for today we ask the question, what would happen if we used the Big Brother voting system?

Big Brother, for those who have been living under a rock, is a TV show in which around 15 house-mates are watched around-the-clock by TV cameras, which broadcast the show live to viewers who, at the end of each week, vote someone out of the house until there is only one contestant left. What if, instead of voting in political parties, we voted out the parties we didn't like?

Let's run an example on my local electorate, Grayndler. At the time of writing, the primary vote distribution looked like this (~70% of the total vote has been counted):

Name Party Votes
James Michael Cogan Socialist Equality Party (SEP) 849
Pip Hinman Socialist Alliance (SAL) 879
Alexander Dore Liberal Party (LIB) 16691
Anthony Albanese Australian Labor Party (ALP) 32406
Sam Byrne Greens (GRN) 17633
Perry Garofani Australian Democrats (DEM) 851

Total 69309

Under the current voting system, it looks like the Labor Party may win the seat, although there is still some uncertainty about this as Liberal party preferences will mostly flow to the Greens, meaning that on preferences there is some small chance that the Greens will win the seat. But what about under our new Big Brother system?

To test out this system, we need to make a few assumptions regarding preferences. We have no idea how voters listed their preferences - whilst, for instance, most Greens voters will preference Labor over Liberal, and many Socialist Alliance voters will preference the Greens over Liberal, I simply don't have the data. Anecdotally, many voters follow party "how to vote cards", meaning that they order their preferences how their favoured parties tell them. So let's assume, for the sake of this analysis, that every voter does this. Taking the party preferences from their senate preference flows list, we see that the parties list their preferences in the following way:

1 2 3 4 5 6
Socialist Equality Party (SEP) SEP GRN SAL LIB ALP DEM
Socialist Alliance (SAL) SAL GRN ALP SEP DEM LIB
Australian Labor Party (ALP) ALP GRN DEM SEP SAL LIB
Australian Democrats (DEM) DEM SAL SEP GRN ALP LIB

It was difficult to come up with the SEP list of preferences as they have three preference lists for the Senate and didn't actually make any effort to order the other parties in terms of preference but rather simply numbered their preferences down the page according to where the parties were written on the ballot. Weird. I suspect that because of this I have their preferences incorrect, but this is simply a worked example so don't hold it against me!

Let's now cross to Gretel Killeen at the Big Brother house.....

Week 1:
After battling it out with a number of pointless challenges and staying up late because they had nothing else to do, the first eviction saw an overwhelming majority of voters evict the Liberals. Using the vote table above and counting up the number of times a party was put as last preference on the ballot, the number of votes for eviction were as follows:

Week 1
Socialist Equality Party (SEP) 0
Socialist Alliance (SAL) 16691
Liberal Party (LIB) 51769
Australian Labor Party (ALP) 0
Greens (GRN) 0
Australian Democrats (DEM) 849

Week 2: 
A dancing-doona between the two socialist parties was the highlight of week 2. With no Liberals to evict, most voters had to turn to their second least-liked party. The Socialist Alliance pulled in an extra 32406 votes - for those playing along at home, these are all the voters who put Labor at number 1 on the ballot box and the Liberals at number 6, whilst the Socialist Equality Party, the ALP and the Democrats also picked up extra eviction votes. It's time to go....  Social Alliance.

Week 1Week 2
Socialist Equality Party (SEP) 017633
Socialist Alliance (SAL) 1669149097
Liberal Party (LIB) 51769-
Australian Labor Party (ALP) 0851
Greens (GRN) 00
Australian Democrats (DEM) 8491728

The following weeks saw an attempted turkey-slap by Labor on the Greens and an impressive Bum Dance by the Democrats. However, in the final week of the show, The Greens took out the seat of Grayndler using the Big Brother eviction rules.

Week 3Week 4Week 5
Socialist Equality Party (SEP) 66730--
Socialist Alliance (SAL) ---
Liberal Party (LIB) ---
Australian Labor Party (ALP) 85135175-
Greens (GRN) 0017542
Australian Democrats (DEM) 17283413451767

75% of voters preferred the Democrats to be evicted in the final round.

There are other ways in which one could run a Big Brother-style election and generally these methods would be likely to find the least offensive, rather than most preferred, party. I'd love to see this method run across the whole parliament!


  1. Perhaps they can call it "Australia's Next Top Politicians"

  2. This is true - on each of their senate pref splits, the SEP started with themselves as "1" and "2", then went to LIB on the first split, LAB on the second and GRN on the third and started numbering down. I didn't grab their how-to-vote card - didn't see them there at all actually - so I'm not sure what they did in the House of Reps. But the parties don't do any pref distribution in the HoR - you can only suggest people distribute prefs themselves in a certain way. In any case, I mucked this up once or twice, but each time the Greens and the Democrats were the final two parties, suggesting they are the least objectionable in Granydler.

  3. Actually, the SEP split their preferences LIB/ LAB/ GRN as they're all "bourgeois parties". I strongly suspect they would take the same approach to every other party, socialist or otherwise (they describe the Socialist Alliance as 'petty-bourgeois ex-lefts").
    So, to do the above exercise accurately, you'd need to divide SEP preferences evenly amongst all the other parties.