In cases where members are elected to represent single-member wards, or divisions, Queensland councils use the optional preferential voting (OPV) electoral system to elect members. This covers all mayoral elections, as well as council seats in most urban south-east Queensland councils.
Under OPV, voters are free to number as many or as few preferences as they wish. A single ‘1’ is sufficient for a vote to be formal. It contrasts to compulsory preferential voting (CPV), which is used for single-member electorates for federal and Queensland state elections, amongst others. CPV requires voters to indicate a preference for every candidate, and their vote is informal without a complete order of preferences.
While these two systems can play out in different ways over time, it’s easy enough in the short-term to compare how preferences flow under the two systems, and from that make some assessments of how Brisbane City elections would differ if CPV was used.
In this post I find that an increased flow of preferences between Labor and Greens would flip a number of seats away from the LNP, and make these elections closer.
At the 2020 Queensland state election, there were 23 seats that were mostly contained within the City of Brisbane, and in all 23 of these seats the final three-candidate-preferred count was between Labor, Liberal National Party and the Greens. I can then compare the 3CP number to the final 2CP to identify how preferences flow in like-for-like situations.
Of the 26 Brisbane City wards in 2020, eighteen of them only had a Labor, LNP and Greens candidate, so the primary vote also functions as a 3CP. Doboy isn’t part of the analysis due to the absence of a Greens candidate, and I’ve also excluded Pullenvale and Tennyson where an independent made the 3CP. This leaves another five wards where Labor, LNP and Greens made the 3CP after other candidates were excluded.
So we have 23 Brisbane City wards to compare to 23 Queensland state electorates to see how preferences flow between these parties in the same area, in the same year, but with different voting systems.
Of the 23 Queensland state electorates, the Greens were knocked out after the 3CP in 21 of those seats (the classic races). The only exceptions were the two seats won by the Greens. In Maiwar, the ALP came third, with their preferences helping re-elect Greens MP Michael Berkman. In South Brisbane, the LNP came third, with their preferences helping elect new Greens MP Amy MacMahon.
At the Brisbane City election, the Greens were knocked out in 18 wards, with Labor knocked out in the other five. Plus the Greens were knocked out in the Lord Mayoral contest, so I’ll include that in the analysis. There was no example where the LNP dropped into third place, as they did in South Brisbane, so we can’t compare LNP preference flows between Labor and the Greens, which was a major factor in the Victorian state election.
This map shows which non-LNP candidate made the 2CP in each electorate, and how much they overcame the third-placed candidate on the 3CP. I explained it in yesterday’s post.
So I have then compared the 3CP to 2CP to calculate the proportion of preferences from the third-placed candidate to each of the other two. Under CPV, the two numbers add up to 100%. Under OPV, there is a third portion, which is the votes that exhaust.
Voter group
Prefs to ALP/GRN
Prefs to LNP
Exhausted
Greens prefs (CPV)
80.1
19.9
Greens prefs (OPV)
47.0
10.1
42.9
Labor prefs (CPV)
74.6
25.4
Labor prefs (OPV)
41.3
12.2
46.5
Generally preferences from the Greens to Labor are stronger than the reverse – something I found when I analysed NSW preference flows.
The proportion of flowing preferences to the LNP tend to be a bit lower under OPV, but in raw numbers the OPV preferences are much less favourable to the remaining centre-left candidate, since the volume of preferences is smaller. Almost half of all votes exhausted in 2020.
Those numbers above are averages, but this next chart shows (in the case of Greens preferences flowing between Labor and LNP) each individual datapoint.
The X and Y axis show the proportion of preferences flowing to LNP and Labor respectively. In the case of CPV elections, those two numbers add up to 100%, so all of those datapoints make a neat line which follows the formula X = 1 – Y. In the case of OPV elections, the distance from the CPV line indicates what share of the preferences exhausted.
While there is some variations, there is a clear clustering. The gap between the CPV datapoints and OPV datapoints demonstrate why OPV tends to favour the candidate leading on primary votes – there’s just less votes available to chase down a lead.
So I then calculated roughly how the exhausted votes would need to split to turn the OPV preference flows into CPV preference flows. In the case of Greens preferences, I calculated that 77% of exhausted votes would flow to Labor. In the reverse case, I calculated that 71.5% of exhausted votes would flow to the Greens. I then applied those adjusted figures to the 2020 Brisbane City Council election results.
In every single case, the LNP’s position is weakened by the use of CPV. It would be enough for the Greens to win a second ward in Paddington, and for Labor to win Enoggera and Northgate (the latter with a margin of just 0.02%, or just nine votes).
That would translate into a council of 16 LNP, 7 Labor, 2 Greens and 1 independent – still a clear LNP majority, but much less dominant than the actual outcome. It would have also reduced the lord mayoral margin from 6.3% to 3.7%.
This wouldn’t have changed the overall outcome in 2020, but you could imagine it making a difference in a closer election.
Queensland used to use optional preferential voting, switching in 2016 in a sudden move in parliament without much notice.
In this case, I think the CPV outcome would have done a better job of representing how people voted. The LNP polled just 45.9% on the council ballot yet won 73% of the seats. While they did benefit from the usual majority bonus you see in single-member electorates, they also benefited from their centre-left opposition being divided.
While I think CPV would produce a fairer outcome, I remain a supporter of optional preferential voting. I don’t think it’s ethical to invalidate votes because they haven’t expressed the maximum number of preferences. But I think there is potential to have reforms that maintain OPV while doing more to encourage preferences – for example by instructing voters to number every box, even if you don’t enforce that rule through informality.
Ultimately, however, the answer is not CPV or OPV. It’s a more proportional system which ensures those exhausted votes can directly elect a representative. Labor and Greens voters can elect their own autonomous representatives who can then come to arrangements in the elected body, rather than relying on preferences to form majorities.