Australians rejected the proposed constitutional amendment known as The Voice in a referendum on 14th October 2023. NO won the national vote by 20.1 percentage points and the state count 6-0 but the more important question is was my forecast right?
What was my forecast?
My article “Voice Referendum 1 – My Forecast Explained” gives you the following background –
- What is the Voice referendum
- How the referendum is conducted
- What the opinion polls said between 2017 & 2023
- How I made my forecast
- My criteria for declaring success
My final forecast & success criteria were –
- NO to beat YES by 17.0% i.e. 58.5% of Australian voters would vote NO and 41.5% would vote YES.
- All 6 states would vote NO.
- If the actual winning margin for NO was within 3.0% of 17.0% i.e. between 14.0% & 20.0%, I would regard that as success.
What was the outcome?
The Australian Electoral Commission Tally Room states 60.06% of valid votes cast were for NO and 39.94% for YES, a winning margin of 20.11% for NO. All 6 states and the Northern Territory voted NO with only ACT (Australia Capital Territory) voting YES. The proposed amendment to the Australian Constitution was therefore defeated.
More details on the results can be found here –
- Wikipedia – Wiki take their feed from the AEC but the page may have other links of interest
- Kevin Bonham – Australian election analyst who has written a number of blogs in relation to the Voice.
- ABC – the main Australian broadcaster who also give results by parliamentary seats. I will say the data visualisation used here is stunning and makes it so easy to read and follow on a smartphone.
Was my forecast a success?
Why couldn’t someone have stuffed the ballot boxes with a few more YES votes?! I got the 6-0 state count for NO right but by a whisker, the winning margin for NO of 20.1% is just outside my success range of 14% – 20%! If 39.95% had voted YES, then rounding to one decimal place would have made it 40.0% to 1 decimal place…
The reason for my +/-3% margin was that in UK general elections, such an error in the Conservative vote share lead over Labour can change the narrative and outcomes of elections. So whilst I could fudge this as a technical success, it is not quite good enough for my purposes so it is worth reviewing how I arrived at my forecast.
What lessons should I learn?
What I find annoying is that in a tweet posted 25th September 2023 on Twitter/X, I showed a method of trend extrapolation which predicted NO to win by 21% based on the polls at that time. This was based on within-house trends which proved a Trend Change had occurred after the Australian Senate passed the Voice referendum bill on 19th June 2023. However, when I updated my forecast on 12th October 2023, the last set of polls in October muddied the waters and led me to conclude a Step Change had occurred instead after that date. This resulted in a different trend extrapolation of NO winning by 16.5%.
The two different trend extrapolations are represented in this chart with solid red lines for Step Change and dashed green lines for Trend Change. The extrapolated leads for NO shown are based on all polls up to 14th October. As can be seen, extrapolation using trend change would have been spot on.
The within-house trends can be seen in this chart and it is the last set of black dots plotted in October which confused me. If I had ignored these, a trend change would be have been the obvious choice but October polls initially suggested the trend to NO had stopped which eventually led me to conclude a step change was the better choice for my extrapolation.
So why did the polls confuse me in October? The most likely answer is that Australians voted quite differently depending on the method of voting they used and pollsters probably struggled to account for this in their October polls.
In the UK, 80% of voters vote on election day with the remainder sending in postal votes in the 2 weeks before election day. In Australia, postal votes could be cast up to month before polling day and from the start of October, voters could also vote early at a polling station. Based on data provided by this excellent blog by Antony Green and shown in the table below, it appears only 45% of voters voted in person on 14th October whilst 35% voted early in person, 11% by post and the remaining 9% voting by some other early voting mechanism. Among the 46% voting early or by post, NO beat YES by 30% whereas among the 45% voting on polling day, NO beat YES by only 12%, a considerable difference that Antony states is very unusual for Australia.
As you can imagine, a pollster running a survey in early October would have had to ask people if they had already voted and then work out how to account for their answers in their results. If the survey had too few or too many early voters, then given the large differences seen in NO‘s lead over YES, it’s not difficult to imagine this causing polls to become more volatile in October. Had I known this was likely to be an issue, it’s possible I would have decided to place less weight on polls in October which would have been counterintuitive.
The lesson I take out of this for UK general elections is to be thankful that we have only 2 ways of voting! It is unlikely such an issue would be as prominent in the UK but it might pay me to look at UK opinion poll history in more depth. My approach till now has been to use only polls taken in the week before election day but perhaps since 2001 (when postal voting rules were relaxed) I might find some predictive power from using polls taken two weeks before election day.
Why did NO win?
The first articles are starting to appear to explain why YES lost so badly after starting out with a 55% lead. Many are making comparisons with Brexit in the UK & Trump in the USA from 2016. I have linked to a few of these and will add more as needed –
- Dr Katy Barnett’s reflections – I linked to an earlier article of hers at the start of this article and I found her postscript insightful.
- Helen Dale interview – Helen Dale is an Australian lawyer living in the UK and I found her exposition of the historical background to referendums in Australia very interesting. Having read other materials from both sides on the referendum, I think she fairly summarises the predominant NO voter motivations.
- Endorsements for Yes & No – a Remain 2016 vibe I immediately picked up on when I started following the Voice 5 months ago was the overwhelming endorsements by the so-called “great and good” of Australia. This Wiki links lists all public endorsements for YES and NO and the disparity is glaring.
- How similar are Leave & Trump voters? – I really hope a similar study is done in Australia to these “exit polls” carried out by Lord Ashcroft in 2016. By analysing these, I showed Remain & Clinton voters were almost identical in their underlying sentiments whilst Leave & Trump could be partly differentiated. I would not be surprised if we see something similar in Australia 2023 with YES voters mapping closely to Remain/Clinton voters whilst NO voters are only proximate to Leave/Trump voters.
— Would you like to comment on this article? —-
Please do leave your comments on this Twitter/X thread.
— Subscribe to my newsletter to receive more articles like this one! —-
If you would like to receive notifications from me of news, articles and offers relating to Elections & Polling, please click here to go to my Newsletter Subscription page and tick the Elections and/or Surveys category and other categories that may be of interest to you. You will be able to unsubscribe at anytime.
— Read some of my other blog posts on elections —
- Keir Starmer’s train to Downing Street in 2024 – I look at which tracks Starmer can follow to win a majority at the next election.
- My first attempt to forecast a local mayoral election which uses the Supplementary Vote system
- I am the most accurate forecaster of the 2019 UK General Election.
- Jeremy Corbyn’s road to a majority in 2019 and the roadblocks they face – my thoughts in January 2019 on the roadblocks in Labour’s way to winning the next election.
- How accurate are voting intention polls for UK General Elections?
More posts can be found by clicking on the Elections tab at the top of your screen.