The preferential ballot favours the party with the most first preference votes

I have written several posts looking at the growing popularity of the preferential ballot/the alternative vote (AV) here in Canada – see this recent one, for example. I even attempted a redo of the 2011 Canadian federal election using the preferential ballot rather than our current FPTP. As I explained in that post, and in others, the big problem in attempting to forecast how the election would have played out using AV was the absence of data concerning voters’ preferences. Some polling firms would (and still do) regularly ask people which party was their second choice, but no one ever looked at voters’ potential 3rd, 4th, etc. choices.

However, a new poll by Abacus Data has done just that. According to Eric Grenier, in this article in the Globe and Mail, the poll asked respondents to rank seven parties from 1 to 7 (in other words, it used full preferential rather than optional preferential). I cannot find this data on the Abacus website. Mr. Grenier examined the numbers and posits that using a preferential ballot “would limit the ability of the Conservatives to win elections”:

With a preferential ballot, however, the Conservatives would come out further ahead. They would lead in 147 ridings on the first ballot (after distributing the marginal support for the smaller parties), compared to only 108 for the New Democrats, 76 for the Liberals, four for the Bloc Québécois, and three for the Greens (primarily due to an anomalous result in the poll in Atlantic Canada).

The Conservatives would have majority support in 60 ridings and win those automatically, while the NDP would win 23 seats on the first ballot and the Liberals 11. But that Tory advantage would disappear once the instant run-off was conducted.

The Conservatives would lose their first ballot lead in 30 ridings, and be reduced to only 117. The New Democrats would move ahead in 18 more seats and take 126, while the Liberals would win 17 more ridings and increase their total to 93. The Greens would hold on to two of the three seats in which they led, while the Bloc Québécois would lose all four.

I am not entirely certain how he comes to that conclusion. The last paragraph quoted above is particularly confusing to me.

Despite Mr. Grenier’s assertions at the outset of the article that the preferential ballot “is used in many jurisdictions around the world”, the only really comparable example available to us is Australia. Full preferential (where voters have to rank every single candidate on their ballot for the vote to count) is used at the federal level to elect the House of Representatives, and in some states, while a couple of states use optional preferential, where voters can choose to rank as many or as few candidates as they want. Indeed, many opt to rank only one candidate and optional preferential becomes a de facto FPTP ballot. This is what happened when AV was used in some provinces here in Canada in the past.

What Mr. Grenier seems to overlook is that the preferential ballot, in particular optional preferential, always favours the party which receives the most first preference votes – at least going by Australia’s long history with this form of voting. Grenier rightly notes that the Conservatives “would have majority support in 60 ridings and win those automatically”. However, things are a bit more complicated after that. Going by Australia’s experience, Conservative candidates would not, as Grenier posits, “lose their first ballot lead in 30 ridings” based on second preferences. It all depends on how close those Conservative candidates are to the 50%+1 needed to win the seat under AV. The closer they are to that mark, the fewer votes transfers they require. Consequently, a Conservative candidate with 45% of the vote on the first ballot count, would most likely still win the seat because they need far fewer votes to boost them over the 50% mark. Even if they were further from the 50% target, say at 40%, but the 2nd place candidate was well behind, say at 30%, the Conservative would still most likely win. Only in instances where two candidates were quite literally neck and neck on the first ballot count would the outcome be up in the air.

Readers interested in preferential voting should regularly read Antony Green’s Election Blog. Green is an Australian elections expert who blogs about both federal and state elections in that country, which, I reiterate, is really the only jurisdction at all comparable to Canada which uses the preferential ballot. As Green explains in this post:

At the 2010 Federal election, 64 of the 150 seats were won by a candidate with a majority on first preferences, and a further 75 won by the highest polling candidate at the start of the count after the further distribution of preferences. Optional preferential voting would have had little impact on these 139 contests.

However, in the 11 contests where the candidate leading on first preferences did not win, optional preferential voting could have changed the result.


The lesson here is that optional preferential voting always advantages the party with the highest first preference vote.

In other words, in the 2010 federal election in Australia, a majority of seats (139 out of 150) were won by the candidate who was ahead after the first count. Sixty-four were won by a majority on the first ballot, and 75 were won on subsequent ballots – by the candidate who’d been in first place on the first count. That is using full preferential. Only in 11 instances did the candidate who’d been leading on the first ballot fail to actually win the seat. Had optional preferential been used instead, in only 3 cases would the candidate in the lead after the first ballot have failed to win.

It is good that a polling firm here in Canada has finally started to explore voters’ preferences beyond their 1st and 2nd choices, but I don’t think Mr. Grenier fully understands how AV tends to play out – at least based on what happens in Australia.

Related Posts:

Radical Centrist