As a final part in my series on trends in people voting early, today I’m analysing differences in how people vote, depending on whether they vote early or vote on election day, using the categories established in yesterday’s post.
Despite losing the last two elections, Labor won a majority of the two-party-preferred vote amongst those voting on election day at both of those elections. Indeed Labor has won the election day vote at four out of the last five federal elections.
The Coalition is much stronger on the early vote, having won the early vote at all of the last seven elections (although they barely won with 50.5% in 2007). As the early vote has increased, this outcome has become more critical to deciding elections. It would have been very hard to win an election despite losing the election day vote back in 2001, when almost 90% of all votes were cast on the day. Yet it looks like a majority of voters in 2022 will vote before election day. If the polls are right and Labor is on track to win a comfortable majority, you'd expect them to win the early vote for the first time this century.
There have been some theories about how much the election day and early vote will resemble each other as the latter grows. Will they become similar to each other? We haven't seen that in federal elections up until now. Indeed the two-party-preferred gap between election day and the early vote has grown at the last four federal elections, from 3.8% in 2007 to 5.3% in 2019.
Now I also want to consider how this could change after the COVID-19 pandemic. As more people cast their votes early, that changes the kind of person who casts such a ballot. Early voting may have traditionally leaned towards conservatives, but I don't think early voting makes someone a conservative. If progressive voters started voting early en masse, that trend could change. Indeed in the United States the 2020 election saw a massive gap between Republicans voting in person and Democrats voting through the mail.
This next chart shows the Liberal primary vote in the elections before and after the COVID-19 pandemic in four states and the ACT:
Every state shows a trend of the Liberal Party doing better in the postal vote than on election day. The pre-poll vote usually falls somewhere in between, but not always. This is the same trend as in federal elections, where pre-poll voting is more favourable for the Coalition than election day voting, but less so than postal voting.
Monday's blog post discussed how postal voting grew significantly in Queensland and Western Australia. The previous chart shows a noticeable shrinking of the gap in those two states, but not in other states.
The same shrinking is not clear on the pre-poll vote, which grew in every state. The gap did shrink slightly in Queensland, and more significantly in the ACT. The gap was largely steady in Tasmania and widened in Western Australia. South Australia does not separate postal and pre-poll votes, but the gap between these special votes and election day votes widened slightly.
I've also collected the same data for the two federal by-elections held in 2020. This chart shows the federal two-party-preferred vote for Labor.
Most of the growth in the early vote for these by-elections was in the postal ovte, and Labor significantly improved their vote in the postal vote, while making less progress on election day or in the pre-poll vote.
So there is limited evidence that a broadening of the early vote makes it similar to the election day vote, either before the COVID-19 pandemic or during it.
When I first learned about the statistic that Labor had won the two-party-preferred vote on election day, it made me wonder: how many seats would have changed hands if you counted just the election day vote, or just the early vote.
This map shows the two-party-preferred winner for each seat on the early vote, and on the election day vote. You can toggle between the layers:
On the early vote, the Coalition won 94 seats, Labor won 56, and the seat of Macnamara was a tie.
On the election day vote, Labor won 73 seats and the Coalition won 78. It's interesting that Labor won 50.7% of the election day two-party-preferred vote, but that translated into a slight Coalition majority.
On these boundaries, Labor holds 69 seats, plus they also won the two-party-preferred vote in the crossbench seats of Clark and Melbourne. So that is 71 seats, compared to 73 seats if you just count the election day vote.
There was only one seat where Labor won the early vote and lost the election day vote: Lingiari. In this seat, Labor won the sizeable remote vote while the Country Liberal Party did better in the urban towns. The CLP actually won the pre-poll vote, but Labor won the remote vote by a much bigger margin.
In addition to Macnamara, where the early vote was tied and Labor won the election day vote, there are 17 seats where the Coalition won the early vote and lost the election day vote: Bass, Blair, Boothby, Chisholm, Corangamite, Cowan, Dobell, Dunkley, Eden-Monaro, Gilmore, Griffith, Hunter, Lilley, Macquarie, Moreton, Perth and Richmond.
Most of those seats were actually won by Labor, but there were three seats the Coalition won despite losing the vote on election day: Bass, Boothby and Chisholm. Quite a small proportion of the vote in Bass was cast early (29.8%), with Boothby a bit high and Chisholm roughly at the national median. All three seats produced this lopsided result because Labor narrowly won on election day but the Coalition won by a large margin on the early vote.
Finally, I was interested in which seats had the largest gaps between the Labor two-party-preferred vote on election day and on the early vote, whether or not it changed the outcome. This map shows the gap in each seat.
Lingiari had the biggest gap of 12.7%. Just six other seats had a higher Labor vote before election day than on election day. The difference was small in Braddon, Fowler, Grey, Mallee and O'Connor, but Labor did 4.3% better on the early vote in McMahon.
On the Coalition side, their biggest gap was in Macnamara, where the Coalition polled 11.8% better on the early vote. I suspect this may be due to Jewish voters avoiding voting on a Saturday. Eight other seats had a gap of over 8 percentage points, with the biggest being Richmond (9.61%) and Boothby (9.65%).
